Schlaepfer v Australian Securities and Investments Commission

[2021] NSWCA 129

Court of Appeal

Supreme Court

New South Wales

• Summary available
• Amendment notes

Medium Neutral Citation:

Schlaepfer v Australian Securities & Investments Commission [2021] NSWCA 129

Hearing dates:

28 July 2020

Decision date:

30 June 2021

Before:

Meagher JA at [1]
White JA at [2]
McCallum JA at [111]

Decision:

(1)   that the primary judge’s order as to costs be vacated;

(2)   that the appeal otherwise be dismissed;

(3)   that Mr Schlaepfer pay one third of the costs of the proceedings below and of the appeal.

Catchwords:

DEFAMATION – Publication – Slander – Where appellant relied on contemporaneous email to prove publication of the words allegedly said – Whether slander proved in the terms pleaded – Whether appellant identified by the matters complained of – Whether imputations complained of by appellant conveyed to the ordinary reasonable listener

DEFAMATION – Defences – Defence of qualified privilege at common law and under Defamation Act 2005 – Whether reasonableness required to be proved to establish defence at common law having regard to defence as pleaded – Whether reasonableness established – Defence of justification – Where ASIC sought to establish truth of imputation of market manipulation without attempting to establish the individual trading of any particular trader or pod of traders – Contention that the relevant company had engaged in the impugned conduct on “any one or more” of 23 “occasions” – Expert opinion evidence concerning proof of impugned conduct by the establishment of certain metrics observed in the company’s trading – Whether sufficient to establish truth defence

Legislation Cited:

ASIC Market Integrity Rules (Securities Markets) 2017 (Cth), r 5.6.12

Australian Securities and Investments Commission Act 2001 (Cth), ss 1, 11(4), 12A(2),(6)

Civil Procedure Act 2005 (NSW), s 56

Corporations Act 2001 (Cth), ss 761A, 798F, 798G, 1041A,1041B,1308A, 1311

Criminal Code (Cth), s 11.3

Defamation Act 2005 (NSW), ss 25, 30

Supreme Court Act 1970 (NSW), s 101

Cases Cited:

Amalgamated Television Services Pty Ltd v Marsden (1998) 43 NSWLR 158; [1998] NSWSC 4

Brooks v Fairfax Media Publications Pty Ltd (No 2) [2015] NSWSC 1331

Chase v News Group Newspapers Ltd [2002] EWCA Civ 1772

Corby v Allen & Unwin Pty Ltd [2014] NSWCA 227

Cush v Dillon;Boland v Dillon (2011) 243 CLR 298; [2011] HCA 30

Director for Public Prosecutions (Cth) v JM (2013) 250 CLR 135; [2013] HCA 30

Dyson v Associated Newspapers Ltd [2020] EWHC 188

Fairfax Media Publications Pty Ltd v Pedavoli [2015] NSWCA 237

J’Anson v Stuart (1787) 99 ER 1357

Lange v Australian Broadcasting Corporation (1997) 189 CLR 520; [1997] HCA 25

Lindholdt v Hyer [2008] NSWCA 264

Papaconstuntinos v Holmes a Court (2012) 249 CLR 534; [2012] HCA 53

Prince v Malouf [2014] NSWCA 12

Radio 2UE Sydney Pty Ltd v Chesterton (2009) 238 CLR 460; [2009] HCA 16

Readers Digest Services Pty Ltd v Lamb (1982) 150 CLR 400; [1982] HCA 4

Reynolds v Times Newspapers Ltd [2001] 2 AC 127; [1999] UKHL 45

Salomon v Salomon [1897] AC 22

Wootton v Sievier [1913] 3 KB 499

Category:

Principal judgment

Parties:

Daniel Schlaepfer (Appellant)
Australian Securities & Investments Commission (First Respondent)
Greg Yanco (Second Respondent)

Representation:

Counsel:
RG McHugh SC, M Richardson, A Harding (Appellant)
J Hmelnitsky, M Lewis, PJ Holmes (Respondents)

Solicitors:
Mark O’Brien Legal (Appellant)
Ashurst (Respondents)

File Number(s):

2019/383494

Publication restriction:

None

 Decision under appeal 

Court or tribunal:

Supreme Court of New South Wales

Jurisdiction:

Common Law Division

Citation:

[2019] NSWSC 1644

Date of Decision:

26 November 2019

Before:

Fagan J

File Number(s):

2016/342827

HEADNOTE

[This headnote is not to be read as part of the judgment]

The appellant, Mr Daniel Schlaepfer, was the principal, director and ultimate beneficial owner of a company, Select Vantage Inc, which traded on stock markets worldwide, and the sole shareholder of Merlito Securities Company Ltd, the company through which Select Vantage traded on Australian stock markets. Select Vantage’s trading activity was undertaken by a large number of individual traders, organised across a number of trading locations (pods), whom it permitted to day-trade with its capital. Select Vantage retained a significant portion of the net revenue of its traders and remitted the balance to the manager of the relevant pod for distribution to the traders in that pod.

In November 2014, the respondent, the Australian Securities and Investments Commission (ASIC), became concerned that Select Vantage and Merlito were engaging in a form of market manipulation known as “layering”. Without notice to Mr Schlaepfer, ASIC sought, successfully, to persuade the companies’ existing Australian stockbroker to terminate Merlito’s trading account. (A previous broker had closed Merlito’s account in August 2014, albeit for reasons which remain unclear on the evidence.) ASIC’s head of market supervision, Mr Greg Yanco, then had a series of telephone calls and in person meetings with senior executives of other major stockbrokers to warn them of ASIC’s concerns. After all but the last of these conversations had taken place, ASIC published a market surveillance update which noted that an overseas firm had recently had its market access terminated by two brokers. The update also referred to regulatory action for layering taken against Select Vantage in Japan and by the UK Financial Services Authority against another company, Swift Trade Inc, and described a “former trader at Swift” as the founder and owner of Merlito and Select Vantage.

Mr Schlaepfer commenced proceedings for defamation. Relying on an email sent by one of the executives shortly after he spoke with Mr Yanco as evidence of the substance of what Mr Yanco had said, Mr Schlaepfer claimed that Mr Yanco’s statements on each of the relevant occasions conveyed a number of defamatory imputations, including that Mr Schlaepfer was engaging in unlawful market manipulation, and true innuendoes, including that Select Vantage and Merlito had engaged in stock market manipulation and that they had engaged in market manipulation serious enough to cause two stockbrokers to terminate their relationships with the companies. ASIC denied that Mr Yanco had made statements in the terms pleaded and contended that Mr Yanco’s actual statements did not identify Mr Schlaepfer and did not convey any of the pleaded imputations or true innuendoes. It also pleaded the defences of qualified privilege (both at common law and under Defamation Act 2005 (NSW), s 30) and justification (Defamation Act, s 25).

As to the latter, ASIC contended that, at the start of the day’s trading, Select Vantage traders, acting in concert, would establish a substantial long position in a thinly traded stock. They would then attempt to create a false impression of demand by placing bids for large numbers of shares at prices sufficiently below the highest competing bids to make them unlikely to be filled. The intended effect (so it was contended) was to place upwards pressure on the share price. Over the course of the day the traders would maintain a large volume of these non-genuine bids, cancelling and resubmitting or amending them to ensure they were unlikely ever to execute, and unwind their long position with the benefit of the rising prices. The principal evidence in support of ASIC’s justification case was the expert evidence of Professor Putnins, whose conclusion was based on 10 statistical measures of stock market activity (metrics) which, in his opinion, were indicative of layering, and statistical tests for coordination in the trading activity of Select Vantage traders.

The primary judge preferred Mr Yanco’s evidence of what he had said during the phone calls and meetings to the email relied on by Mr Schlaepfer. His Honour concluded that Mr Schlaepfer was not identified by Mr Yanco’s statements; that the words said by Mr Yanco did not convey any of the imputations or innuendoes pleaded; that the statutory and common law defences of qualified privilege were made out and not defeated by malice on ASIC’s part; and that although Mr Schlaepfer was not personally involved in any wrongdoing, the innuendo that Select Vantage and Merlito had engaged in stock market manipulation was substantially true, having regard to Professor Putnins’ metrics and the evidence of coordinated activity.

Mr Schlaepfer appealed. The issues on appeal were:

1. Whether the primary judge erred in not finding the six slanders proved as pleaded (by reference to the contemporaneous email of a party to those communications).

2. Whether the primary judge erred in finding that Mr Schlaepfer had not been identified, in view of the market surveillance update, issued after all but one of the publications, and evidence that some of the recipients of the publications had identified him.

3. Whether imputations (b), (c) and (d) and true innuendos (a), (b) and (d) were conveyed (there was no challenge to the findings that the remaining imputation and true innuendo were conveyed).

4. Whether the defences of qualified privilege at common law and under Defamation Act, s 30 were made out.

5. Whether Select Vantage and Merlito had been engaging in stock market manipulation. Mr Schlaepfer contended that Professor Putnin’s 10 metrics were either simply not indicative of layering or explicable as the product of restrictions on Select Vantage traders’ activity (such as their inability to short) and the aggregation of the trading activity data of many individual traders.

Held, dismissing the appeal (McCallum JA, Meagher JA agreeing and White JA agreeing subject to two reservations)

As to issue (i)

by McCallum JA at [176] and [181], Meagher JA and White JA agreeing at [1] and [2]:

Mr Couper’s contemporaneous email should be accepted as an accurate record of what Mr Yanco said to the six brokers. The primary judge erred in not finding that the six slanders were proved as pleaded.

As to issue (ii)

by McCallum JA at [191]-[194], Meagher JA and White JA agreeing at [1] and [2]:

The primary judge erred in finding that Mr Schlaepfer had not been identified by the matters complained of.

As to issue (iii)

by McCallum JA at [214] and [221], Meagher JA agreeing at [1], White JA expressing a reservation as to that conclusion at [4]-[6]:

The primary judge erred in not finding imputations (b), (c) and (d) conveyed. Imputation (b) was conveyed to any person who knew that layering is a form of market manipulation (which, as the primary judge found, would include all of the recipients of the matters complained of). Imputations (c) and (d) were conveyed in the ordinary and natural meaning of the words said by Mr Yanco. It followed from those conclusions that true innuendos (a), (b) and (d) were also conveyed.

As to issue (iv)

by McCallum JA at [242], Meagher JA and White JA agreeing at [1] and [2]:

The primary judge was correct to hold that the defence of qualified privilege at common law was made out. Accordingly, the appeal should be dismissed.

by McCallum JA at [260], Meagher JA and White JA agreeing at [1] and [2]:

The fact that Mr Yanco’s communications were made without any opportunity for Mr Schlaepfer to respond was fatal to the element of reasonableness required to be established in order to make out the statutory defence. Accordingly, the primary judge erred in holding that the defence of qualified privilege under s 30 of the Defamation Act was made out.

As to issue (v)

by McCallum JA at [307], [368], Meagher JA agreeing at [1], White JA disagreeing:

The evidence adduced by ASIC was insufficient to sustain the defence of justification. It did not establish that Select Vantage or Merlito were engaging in market manipulation.

by White JA at [109]:

The evidence did not establish the truth of the imputations and true innuendoes pleaded. However, the primary judge was correct to find that the evidence of Professor Putnins established that Select Vantage or Merlito were engaging in market manipulation.

Judgment

1. MEAGHER JA: I have had the benefit of reading in draft the judgment of McCallum JA and agree, for the reasons her Honour gives, with the orders she proposes.

2. WHITE JA: I have had the very considerable advantage of reading in draft the reasons for judgment of McCallum JA. Subject to two reservations, neither of which is dispositive of the appeal, I agree with her Honour’s reasons and I agree that the appeal should be dismissed.

3. These reasons assume a familiarity with the reasons of McCallum JA.

4. My first reservation concerns grounds 5 and 6 of the notice of appeal which challenge the primary judge’s conclusion that the imputations alleged by Mr Schlaepfer would not have been conveyed to the stockbrokers to whom the matter was published. Mr Schlaepfer contends that the primary judge ought to have asked whether the imputations would be conveyed to the ordinary reasonable listener. I agree with McCallum JA (at [203]) that the stockbrokers to whom Mr Yanco spoke would be more likely, rather than less likely, to draw the imputations alleged than would an hypothetical ordinary person.

5. Where a plaintiff contends that a defamatory imputation that would not be conveyed to an ordinary reasonable reader, viewer or listener was conveyed because the recipients of the defamatory publication were in a narrow or specialised field, it is necessary for the plaintiff to plead by way of true innuendo the special facts known to those recipients which are alleged to convey the defamatory imputation to them (Readers Digest Services Pty Ltd v Lamb (1982) 150 CLR 400; [1982] HCA 4 at 505-506; Radio 2UE Sydney Pty Ltd v Chesterton (2009) 238 CLR 460; [2009] HCA 16 at [49]-[52], 481-2; Corby v Allen & Unwin Pty Ltd [2014] NSWCA 227 at [180]-[181]).

6. In the present case it was the defendant who contended that the imputations alleged by the plaintiff would not have been conveyed to the particular narrow class of listeners to whom Mr Yanco spoke by reason of special facts known to them. In principle it might be thought that it would be open to a defendant to rely on extrinsic facts that showed that words which, in their natural and ordinary meaning, convey a defamatory imputation did not convey that meaning if these facts were known to all the persons to whom the matter was published (Gatley on Libel and Slander 12^th ed at [3.23]). It is unnecessary to pursue this issue as no such extrinsic facts were pleaded by ASIC. In any event, I agree with McCallum JA (at [205]) that the imputations would have been conveyed to the stockbrokers to whom Mr Yanco spoke who had the specialised market knowledge.

7. My second reservation concerns the defence of justification. As McCallum JA observes (at [310]) the only finding this Court is asked by Mr Schlaepfer to review is the primary judge’s finding that some orders were placed by Select Vantage traders that were manipulative and in contravention of ss 1041A and 1041B of the Corporations Act though it was not possible to say how many orders were involved or to identify the particular orders or series of orders involved. McCallum JA’s decision in Brooks v Fairfax Media Publications Pty Ltd (No 2) [2015] NSWSC 1331 and the authorities to which her Honour there referred establish that that is insufficient to sustain a defence of justification. However, counsel for Mr Schlaepfer said that it was important for his client’s reputation for the finding to be addressed by this Court, not only on the sufficiency of the defence raised, but on its merits. I accept that that is so. Mr Schlaepfer submitted that the primary judge had failed to engage with the answers he and Dr Carr propounded to Professor Putnins’ evidence.

8. For the reasons which follow, I agree with the primary judge’s conclusion.

9. The primary judge summarised the model of trading of Select Vantage’s traders as follows:

“[20] The business model of Select Vantage involved trading activity of 1,800 individual traders located at various places around the world, each of whom was authorised to use the company’s capital to trade on stock markets to which Select Vantage had direct market access through brokers. The company set the parameters of trading. A significant limitation was that only day trading was permitted; that is, each trader could buy and sell shares but had to close out his or her position on each exchange at the end of each trading day. No stock was permitted to be held overnight.

[21] The traders were engaged in groups called “pods”, each of which had a particular physical location. There was a manager for each pod. A large number of the traders were located in China. Because of the minor difference between that country’s time zone and Australian Eastern Standard Time, Select Vantage’s Chinese traders were active on the ASX and Chi-X. They were authorised by Select Vantage to trade through Merlito’s direct access relationship with Macquarie. The plaintiffs contend that the individual traders were free to place bids and asks and to buy and sell stock according to their individual strategies and decisions.

…

[81] … About half of Select Vantage’s traders were domiciled in China and because of the closeness of the two countries’ time zones many of them traded on the Australian market.”

10. There was evidence that the traders received a degree of common training and there were remuneration incentives based on group performance. However, there were no incentives for remuneration otherwise than on a pod by pod basis. The performance of multiple pods was not aggregated to calculate remuneration.

11. Both Mr Schlaepfer and Mr Kruyne, an analyst at True North Vantage (a subsidiary in Mr Schlaepfer’s group of companies), gave evidence of controls designed to prevent manipulative trading. The primary judge summarised the evidence of Mr Kruyne on this issue (at [249]-[250]):

[249] Mr Kruyne said that Select Vantage’s trade surveillance team had remained fairly constant at five analysts during 2013 and 2014. Throughout those years, the filters in Select Vantage’s software known as Ginger were set to raise an alert if, following execution of a trade, there was a cancellation of orders on the opposite side of the book within 10 seconds. That criterion remained the trigger for identification of layering throughout 2014. In cross-examination it was clarified that an opposite-side cancellation within 10 seconds would not engage the alert if the cancelled order had been resting on the book for a significant period of time. For other types of manipulation another filter was in place. This identified any placements and cancellations of orders in quick succession. The surveillance team also carried out analysis to try to identify collaboration in the placement of orders between traders in different locations around the world.

[250] Mr Kruyne said that in 2014 Select Vantage’s filters on the entry of orders restricted its traders to 3 orders per side per exchange per stock. Combining the ASX and Chi-X, a trader could have six bids and six asks resting at any time on a particular stock. Attempts by traders to place orders in excess of this limit would be blocked before the orders could reach the broker’s electronic placement system. Some traders were granted exemption from this limit and could place a higher number of orders. However Macquarie itself set an upper limit of five orders per side per exchange per stock.

12. It is important to say something more on two of the constraints on the traders. Traders were prohibited from holding stock overnight and were constrained from short-selling. Neither the short-selling constraint nor the prohibition on holding stock overnight was considered by Professor Putnins in analysing the trade data until the hearing. The short-selling constraint was explained by Dr He in his oral evidence:

“Q. Were there limitations on short selling?

A. Yes.

…

Q. You tell me what the limitation was.

A. So, it’s against naked short selling.

HIS HONOUR

Q. Meaning?

A. Meaning that you have to have - you don’t have to own the shares, but have to have access to, to borrow those shares.”

13. The importance of this constraint was pressed by Mr Schlaepfer on appeal as it did not feature in the reasons of the primary judge. It was said to undercut a number of Professor Putnins’ metrics as it was said to explain why the traders entered a significant number of buy orders because they were unable to trade until they had access to stock. The prohibition on holding stock overnight was also said to be relevant for the same reason, as no trader could start the day with a long position. Further, the requirement to revert an inventory to zero before the end of the day was said to explain why there was significant selling activity by the traders towards the end of the trading day and particularly in the last half hour.

Challenge to the 10 Metrics

14. The primary judge accepted six of Professor Putnins’ metrics, rejected one and did not consider three in detail (at [205]). The metrics were explained in detail in Professor Putnins’ report. I shall consider them in turn.

Metric 1: Unbalanced Quoting

15. This metric looks to the displayed orders in the limit order book to show whether there is an imbalance between bid side and ask side dollar volume. The metric is expressed as a percentage where 0 per cent is equal dollar volume between bid and ask orders and 100 per cent is all dollar value on one side of the market. A market participant who is layering will generate a quoting imbalance as they use the imbalance to create a false impression of the buying or selling interest in a stock in order to influence other traders and drive the stock price either higher or lower.

16. The primary judge summarised Professor Putnins’ evidence in relation to unbalanced quoting (at [207]-[209]):

[207] … The dollar volume of each order is derived by multiplying the number of shares by the price at which the order is placed. Professor Putnins found that for each stock/day the total dollar volume of Select Vantage’s buy orders significantly exceeded the total dollar volume of its sell orders. He worked out the proportion of this excess, the difference between the dollar volume of buys and sells, relative to the combined total of all orders (buy and sell) for the day, expressed as a percentage. The average across all the stock/days was 85%. Professor Putnins carried out the same exercise for all other traders in each of these stock/days and found that their corresponding proportion was 18.1%.

[208] Professor Putnins expressed the following conclusion about this highly unbalanced pattern of Select Vantage’s ordering (or “quoting”):

The resting (unexecuted) limited orders of a layering strategy a given point in time are likely to be highly unbalanced (more buy volume than sell volume or vice versa). The imbalance is used to create a false impression of buying or selling interest in order to influence other participants and market prices. […]

On all of the stock-days, [Select Vantage’s] average displayed order imbalance is above 50%, reaching levels as high as 98% on some stock-days, and having an average of 85% across all of the stock-days. These levels indicate a strong tendency for [Select Vantage] to have highly unbalanced displayed orders resting in the limit order book.

[209] This was further explained in Professor Putnins’ oral evidence as follows:

Select Vantage has roughly 12 times more quoting activity on the buy side than the sell side, yet they buy and sell equal volumes on a typical day. In other words, putting the quoting activity in perspective relative to the amount of trading activity, there's 12 times more, in other words there's abnormal quoting activity on the buy side of the market.

17. The primary judge relied on this metric in concluding that there was layering of the market. Mr Schlaepfer argues that this metric is unable to indicate layering as any imbalance is explicable by the constraints placed on the Select Vantage traders not considered in Professor Putnins’ reports, and by legitimate trading techniques.

18. It can readily be accepted that the need to establish a long position before entering sell orders and the requirement that the traders commence each day with no inventory is a legitimate justification as to why the traders would enter a number of limit orders to purchase stock at the start of continuous trading. In cross-examination Professor Putnins’ was unshaken that the addition of this consideration would not impugn the inference of layering that could be drawn from a trigger of this metric:

Q. Assuming that that is right, that individual traders could not short sell, could not put on an ask unless they had the stock in hand through earlier filled bids on the day, with that one exception of that one trading day and one stock. Would that have a bearing upon the capacity of the trader to undertake layering in cycle? In other words, to go back from layering up bids and getting an ask accepted to cycle back into layering up asks in hoping to make a bid? Would the constraint upon short selling affect that?

A. Not necessarily, your Honour, because what the individual trader could do if they wanted to engage in a cyclical layering strategy is first establish a long position in the stock, then from that point maintaining inventory at a positive level, cycle that inventory up and down and at the end of the day close out that inventory position. What it would prevent is the cyclical trading strategies that are sometimes seen in markets where there is no need to first establish a long position and one can just cycling with long, short, long, short, long short.

Q. But from what Mr Harding is saying about the effect of the evidence to date, and we'll just assume that that's correct, in order for the trader collectively to switch back into layering up a lot of asks behind the touch price, with just one bid on the other side in the hope of moving the price down presumably, the traders who did that collectively again, is it correct, would have had to have acquired a substantial quantity of stock and accept the risk of exposure of holding quite a substantial quantity of stock to put it in as a series of asks at various steps?

A. That's correct, your Honour. I think it constrains the ability to run a cyclical layering strategy if your constraint is to keep your inventory in a positive domain, and not being able to place asks that exceed your inventory position.

Q. Whereas if they layer up on the bid side, there's no question of having to have stock in hand in order to make as many bids as you like, it's just that as you make them, if they are hit and result in trades and are successful, you're going to have to find the money unless you sell out, correct?

A. That's correct, your Honour.

19. Although the need to establish a long position at the start of trading may explain the existence of an order imbalance at the start of the day, and so much was accepted by Professor Putnins, it does not explain a very high score on this metric which is indicative of an imbalance that persisted over the course of a whole trading day, rather than merely the first hours of trading.

20. Professor Putnins accepted that a high result on this metric could be consistent with traders acting as speculators. Speculators, who are traders who have a view about the ultimate price direction of a stock, are likely to be unbalanced as they seek to acquire or sell out a long position consistently with their view as to the stock’s price trajectory.

21. Mr Schlaepfer submitted that the justification for that imbalance over the course of the day was that it was legitimate for the traders to leave any unfilled or partially unfilled buy orders from the opening period of trading in the hope that they would be filled later in the day at what would presumably be an attractive price. That is, a speculator who believes that over the course of the day a stock will increase in value may leave a limit order at a lower price on the book meaning that if it was to be filled it could hopefully be sold later in the day when the stock is anticipated to be trading at a higher price.

22. Mr Schlaepfer’s counsel put such a scenario to Professor Putnins which was accepted:

“Q. On those assumptions it might be a perfectly legitimate strategy for a trader with those constraints at the beginning of the day to place one or more non-aggressive buy orders in the hope of obtaining a fill at a favourable price, mightn't it?

A. Okay.

Q. Do you accept that?

A. Yes.

Q. For example, a trader who knows that a stock typically trades in the band 20 to 21 cents might place an order at, say, 19 cents in the hope of obtaining a fill at that price in the expectation that the stock would rebound and the trader would be able to exit at 19.5 cents or higher. Do you follow?

A. I do, but why place the order at 19 cents when the market is at 20, 21?

Q. In the hope that it might move down.

A. Why not place it when the market moves down and execute at 19?

Q. There's no reason why you couldn’t place it down there and then just let it sit there in the hope that the market moves, and in fact that would be a perfectly sensible idea, wouldn’t it, in order to obtain queue priority at that price?

A. If you obtained queue priority then you should have high execution rates.

Q. No, that's not my question?

A. Okay.

Q. My question was whether the strategy or the scenario that I explained to you might be perfectly commercially sensible and reasonable for a trader who wishes to obtain queue priority at the lower price of 19 cents?

A. Sure.”

23. This metric was also challenged on the basis of alleged methodological flaws. Dr Carr challenged the utility of this metric as not accounting for the time relationship between the imbalanced quoting and when trades actually execute. This challenge took two forms. First, the metric treats an order imbalance that arises from a trade entered “five hours earlier in the trading day in the same manner as an identical order that was placed immediately before a trade”. Secondly, the metric “gives a higher score for cancelled orders that stayed open for a long period of time”.

24. If one were looking to draw inferences about when particular orders were entered and cancelled, or the likely price effects of particular orders, then this metric would not be of assistance for the reasons given by Dr Carr. However, as the case mounted by ASIC is one of persistent unidirectional layering over the course of the trading day, the minutiae of the imbalance at particular points in time is not obviously relevant. It is the fact of the imbalance which may create a false impression of demand or market interest which is of relevance.

25. On its own this metric is unable reliably to indicate layering as it cannot exclude legitimate trading practices. A speculator who has a particular view about the price direction of a stock is likely to be highly imbalanced. That view is reinforced when one considers the constraints on Select Vantage traders. Although this metric cannot exclude the trading activity of legitimate speculators it does demonstrate that Select Vantage traders were on average significantly imbalanced in their trading activity and, albeit consistently with the short selling constraint, that the imbalance was due to substantial bid side limit orders.

Metric 2: High Quoting Activity

26. Rather than looking to an imbalance between quoting activity on each side of the market, this metric purports to measure the dollar volume of the resting bids and asks attributable to Select Vantage traders as a percentage of the total bids and asks. The metric was expressed as an average over the course of a day’s trading where the constitutive data was calculated each time there was a change in the status of the limit order book (i.e. a trade was made or an order was cancelled etc).

27. Professor Putnins observed that over the course of trading in the impugned stocks Select Vantage traders averaged approximately 20 per cent of the total dollar volume of bids and 1.7 per cent of asks. The total range of bid side activity across the various stocks when taken individually ranged from 10.5 per cent to 28.5 per cent.

28. The primary judge summarised the effect of Professor Putnins’ evidence as follows (at [211]):

Placing a large proportion of the resting orders on one side of the market (buying or selling) is not a typical characteristic of a legitimate short-term speculative trading strategy. Short-term speculators that have a prediction about future price movements usually want to trade in the direction of the predicted price movement with minimum signalling of their trading intention to other market participants. […]

The limit orders of a layering strategy are likely to represent a substantial proportion of the depth … on one side of the market (buying or selling) as the layering strategy is attempting to mislead the market. Using a large volume increases the probability that the layering orders have a substantial effect on the appearance of supply or demand and therefore the desired effect of influencing the price or trading decisions of other market participants. […]

On average [… Select Vantage’s] quoting on the bid side represents about one fifth (20.3%) of all quoting activity (dollar depth) on the bid side. This share ranges from 10.5% to 28.5% indicating a fairly high quoting activity on the bid side of the market …

In contrast [Select Vantage’s] share of the quoting activity (dollar depth) on the ask side is a mere 1.7% on average – less than one tenth of its quoting on the bidside. [Select Vantage’s] quoting on the ask side puts the bid side quoting level into perspective. [… On] a typical stock-day [Select Vantage] buys and sells an equal volume of shares. Therefore, the high volume of bidding does not reflect a tendency to buy more shares than the amount sold. [… There is] 12 times more quoting activity on the bid side than the ask side, even though the volumes bought and sold are approximately (or exactly) equal. This comparison suggests that [Select Vantage’s] quoting activity is excessive (by a factor of 12 times) relative to the volume that it buys.

29. Professor Putnins was challenged on this metric in cross-examination on the basis that the short selling constraint and the aggregation of the trading data of traders who never obtain a fill and those who trade actively on both sides of the market might give rise to a misleading impression:

Q. But take the case of the other - I forget how many I gave you. I think I said 100 orders placed at the beginning of the day in various parcels by SV traders, only one of which got filled, for the other 99 who might leave their bids on the order book, on your calculations those bids contribute to your calculation of this metric in a manner that indicates layering, is that right?

A. They contribute to the imbalance between the bids and the asks.

Q. And the level of quoting activity because those particular traders never get the opportunity to place an ask, do they?

A. If, yeah, if they, if they don't end up buying and if there's a constraint then they don't sell, right.

30. It may be accepted that there is a risk that this metric would present an inflated figure due to the aggregation of the traders without excluding those traders who never obtain a fill but leave their limit order standing in the hope of being able to trade later in the day. That being said neither expert demonstrated either way what proportion of traders fell into such a category or whether it was of significance. However, should ASIC’s case on collusion be made out then the inclusion of such resting orders would be material. Such orders, although never executed, would be consistent with a wider strategy of inflating the perceived bid side interest and liquidity.

31. It was also said that a trigger on this metric was not necessarily indicative of layering as high levels of quoting would be consistent with legitimate trading. Specifically, it was said that high levels of quoting were a feature of liquidity provision strategies where the trader typically seeks to trade on both sides of the market taking advantage of shifts in the spread by providing liquidity to other traders. High levels of quoting would be consistent with both providing liquidity and leveraging profit to be made through executing economically meaningful trades on both sides of the market.

32. Professor Putnins accepted in cross-examination that high levels of quoting could be consistent with liquidity provision strategies however he did not accept that such a strategy was a likely justification when unbalanced quoting was also present in the trading data (metric 1):

Q. Thank you, and just one last question. On this particular metric, I just wish to ask one or two further questions. You accept don't you, and I think this table in exhibit P graphically depicts it, that it is to be expected that certain market participants such as liquidity providers will present a substantial proportion of the depth of the market at any one time?

A. That is correct. It wouldn't be, it wouldn't be unbalanced though. A liquidity provider may be a substantial proportion of the market overall across both sides but you wouldn't expect for a liquidity provider an imbalance. So say just the 20% number, that could be a large liquidity provider that accounts for 20% of, of the, of the quoting activity. But the imbalance is not a typical characteristic of a liquidity provider.

33. Dr Carr challenged the basal proposition that there should be, generally speaking, equality between the bid and ask side activity as part of a liquidity provision strategy. The debate between the experts went as follows:

Professor Putnins:

“… Now why is that comparison a relevant one? It's a relevant one because we know from the analysis, and this is something that I think Dr Carr and myself both agree across, across the reports, that Select Vantage tends to buy and sell an equal quantity of shares in a given security on a given day. In other words, they try and not hold inventory positions overnight. Because that happens to be an empirical feature of the data, then if I compare the quoting activity on the buy side with the quoting activity on the sell side, that is in effect holding constant the amount of buying or selling that is executed.

What I find in that particular characteristic in my primary report is that Select Vantage has roughly 12 times more quoting activity on the buy side than the sell side, yet they buy and sell equal volumes on a typical day. In other words, putting the quoting activity in perspective relative to the amount of trading activity, there's 12 times more, in other words there's abnormal quoting activity on the buy side of the market.”

Dr Carr:

“Q. Do you wish to respond to that?

A. Right, right, and that's, that's, that's exactly what I was thinking earlier when I said this is very significant. The professor seems to make certain implicit assumptions. In this case there's an implicit assumption over here that the number of bids and asks ought to be equal if you are buying and selling equal amounts through the day. He states that we agree that Merlito bought and sold equal amounts through the day. The data shows that. But for that, he seems to expect that the number of bids and the number of asks also must be equal and then he says, well, Merlito put 12 times more bids than asks. But that's not necessarily so. In fact, the rest of the market put on more bids than asks as well. In fact, statistically significantly so. So the presumption that the bids and asks must be equal is not necessarily true. It's not theoretically true and it's not born by the data as well, even on these days.

Q. Does that mean that you disagree with Professor Putnins' statement at transcript 987 line 49 that there's abnormal quoting activity on the buy side of the market?

A. Yeah, I don't think we have any basis to conclude one way or another.”

34. It may be accepted that simply because more bids were entered than asks one cannot conclude that there was trading activity inconsistent with genuine liquidity provision strategies. However, if the imbalance that ASIC propounds here (12:1) is soundly determined then it would be difficult to conclude that such an imbalance was consistent with liquidity provision strategies. The primary judge did not say (at [239]) that where the disproportion between bids and asks is by a factor of 12:1, the trading is ipso facto consistent with layering. However, that disproportion “in combination with other features” may be consistent with layering.

34. Dr Carr challenged the manner in which Professor Putnins calculated the average imbalance of 12:1. Dr Carr undertook his own statistical analysis of the cancellation rates of bids and asks by reference to how long the respective order remained on the limit book prior to cancellation. Taking the trading in ADO on 15 September 2014 as an example:

36. What this demonstrates is that of the total cancelled orders, the asks were cancelled more than twice as quickly when compared to the bids. Some 67 per cent of eventually cancelled bids remained on the limit book for longer than 10 minutes, where only 51 per cent of eventually cancelled asks remained on limit book longer than five seconds. Dr Carr opines that as Professor Putnins calculated this metric by looking to the percentage of the dollar value of orders on each side of the market attributable to Select Vantage traders at each time there was a change in the order book, the result for the bid side depth is inflated as those longstanding orders would be counted in the metric potentially thousands of times. The inverse is true for the ask side where the majority of orders would only be counted in as many calculations as take place within 30 seconds.

37. This metric presents an average over the course of the day’s trading. Dr Carr’s challenge would be significant if it were necessary to determine the nature of the imbalance at particular points in time, such as within 30 seconds of a trade actually taking place. That was not the issue. The primary judge rightly considered that at times the utility of Dr Carr’s challenges to the evidence of Professor Putnins was undermined by his narrow conception of layering (at [237]). ASIC sought to prove the degree of quoting activity on each side of the market and the inferences that can be drawn from their relative differences over the course of a trading day.

38. However, unless the case for collusion can be established by other means, the failure to exclude the potentially high number of traders who never obtain a fill casts doubt on the utility of this metric to indicate layering on the bid side. Should the case for collusion be established then the colour such bids would take would be consistent with layering. On the trading days in question Select Vantage traders accounted for a significant proportion of the resting bid orders throughout the trading day.

Metric 3: Abnormal Cancellations

39. The third metric tests for high levels of cancellations as a proportion of total orders. The logic underpinning this metric is simple. The archetypal feature of layering is the entering of orders that are not intended to be executed.

40. The primary judge summarised Professor Putnins’ analysis of this metric (at [212]-[214]):

[212] Professor Putnins’ third measure is an abnormally high rate of cancellation of orders. He worked out the ratio of Select Vantage’s order cancellations to the number of its trades, in each stock on each of the relevant days. He found that this varied between 10:1 and 1.9:1 (expressed in his report simply as ratios of “10” and “1.9”). The average was 3.4. Professor Putnins made the same calculation for all other traders combined and derived an average of 0.8.

[213] Professor Putnins’ opinions about this metric are as follows:

Layering strategies involve abnormally high order cancellation rates (a high ratio of cancellations to trades). This is because the layering orders are not intended to execute so they will typically end in a cancellation, and may also need to be cancelled and resubmitted as market conditions change (such as when the market moves towards the layering orders) to maintain a low execution probability. […]

[T]he characteristic of abnormally high cancellation rates is a feature of [Select Vantage’s] trading. On all of the stock-days, [Select Vantage’s] cancel-to-trade ratio exceeds that of other market participants by a clear margin. [It] is between 251% higher and 1081% higher than that of other market participants depending on the stock-day. [… On] the average stock-day, [Select Vantage’s] cancel-to-trade ratio is about 450% higher than that of other market participants. This is a substantial difference in cancellation rates, consistent with the presence of layering orders in the Relevant Order Activity.

[214] I am satisfied that Professor Putnins’ methodology is effective to ensure that the high cancellation rates do not merely reflect legitimate influences on the market, such as news specific to the stock. He gave this oral evidence:

I deliberately construct the [metric so] that it controls various market wide effects that could be specific to a given security or specific to a given day. [… For] example, if different levels of volatility of different securities lead to different cancellation rates, different markets have different cancellation rates, whether there's news flow or not news flow on a particular day can affect the cancellation rate and so forth.

So the way to control all of those various factors that can affect the natural, the normal cancellation rates is to conduct the analysis within a given stock within a given day, and within that given stock and within that given day you hold constant things like the particular stock, the characteristics of the stock, the news flows and so forth, and so the relevant comparison holding those factors constant becomes between Select Vantage cancellation rates and the cancellation rates of the rest of the market […].

41. The general thrust of the reasoning is that the rate of cancellations to trades was economically significant and outstripped the balance of the market on average.

42. Mr Schlaepfer submitted that a high rate of cancellations is consistent with legitimate liquidity provision strategies; a proposition which was accepted by Professor Putnins. However, that by itself fails to explain the significantly higher rate of cancellations when compared to the rest of the market which is presumably made up of traders pursing a range of trading strategies including liquidity provision. Counsel for Mr Schlaepfer put to Professor Putnins that on the particular trading days it is possible that the balance of the market was made up of predominantly speculators or long-term traders. In such circumstances the difference between traders acting as liquidity providers and the balance of the market would be exacerbated:

Q. Would you assume please that the rest of the market in these particular stocks on these particular days was heavily comprised of speculators and long-term investors, eg, because they're attracted to the stock or wishing to sell it because of recent news?

A. Okay.

Q. Would you also assume that Select Vantage traders were predominantly engaged in liquidity provision strategies?

A. Okay.

Q. On those two assumptions, that may well result in SV, Select Vantage, having a higher cancellation rate than the rest of the market, mightn't it?

A. Yes.

43. Dr Carr also suggested that various factors such as information flow may contribute to an explanation as to why the rates of cancellation are comparatively higher.

44. Although the logic may be accepted, the assumption regarding the balance of the market is not established. As Professor Putnins explained in examination in chief the purpose of comparing the trading of Select Vantage traders to the rest of the market is to capture a representation of what “a typical trader” is on the day in question in relation to that particular stock. There was no evidence that that representation would be anything other than a diverse group of traders with various strategies, motives and information. It is unlikely that on the 23 days in question it just happened to be the case that the rest of the market was overwhelmingly employing strategies which would result in low cancellations. This is especially so when one accepts that there were a number of brokers who exhibited higher rates of cancellation than Select Vantage traders on 19 of the 23 relevant dates.

45. The assumption that the “traders were predominantly engaged in liquidity provision strategies” is inconsistent with Mr Schlaepfer’s argument that in relation to metric 1 that course of trading conduct could be explained by reference to significant numbers of speculators. In his first report Dr Carr states that the data in his opinion is representative of the traders pursing “diverse strategies”. This tension is present throughout the challenge to the metrics and it was not explained how the traders could be predominantly engaged in conflicting trading strategies without that appearing in the data.

46. Counsel for Mr Schlaepfer also put to Professor Putnins that the rate of cancellations might be increased by the short selling constraint, particularly if no fill were achieved by those traders early in the day’s trading:

Q. But I'm asking you to assume a trader who has no inventory position--

A. Okay.

Q. --who has zero stock and yet has some open orders on the market coming into say the last ten minutes of trading. Would it not make sense for that trader subject to the constraint to be flat at the end of the day, in order to avoid obtaining a fill on those open orders, to cancel them?

A. I would expect those orders for someone that doesn't hold inventory and knows that they need to be flat at the end of the day to be cancelled well in advance of the last half hour or the last ten minutes because if you go into the last part of the trading day with the possibility of picking a position that you can't flatten, that's a, that's a substantial risk.

Q. So you would expect to see cancellations then of open orders?

A. You would expect to see cancellations, yes.

…

Q. The fact that Select Vantage traders are subject to this constraint means that they may legitimately demonstrate a higher cancellation rate than traders who are not subject to that constraint, you would agree?

A. It's possible.

47. Counsel’s question assumed that traders who did not obtain a long position were not engaging in manipulative conduct. If the finding as to collusion stands then it is possible to infer that those traders who did not obtain a long position but nevertheless entered resting bid orders which were cancelled at day’s end played a role in creating an artificial impression of the stock.

48. The final criticism of this metric arose from cross-examination of Professor Putnins:

Q. Taking you very quickly back to metric 3, abnormal cancellations, pages 164 to 165, it's the case isn't it that you've also brought to account in this metric dark order cancellations or cancellations in the dark market?

A. Sorry, which page are we looking at?

Q. 164 and 165.

A. So, so in this metric I haven't done additional analysis. This metric is as it stands, this wasn't one that we discussed yesterday, so figure 5, just to try and remind myself exactly how this is constructed, is a cancel to trade ratio taking the number of order cancellations, dividing it by the number of trades in a given stock day. So, this has, it could be each stock activity in both ASX and Chi-X. So, there were no additional constraints imposed on this metric.

HIS HONOUR

Q. So, in that one do you recall whether you included trades that had been cancelled within the dark market?

A. I believe they're included, your Honour.

49. The inclusion in the calculation of orders cancelled in the dark market (the dark market is discussed below at [78]) is said not to be indicative of layering and may result in an inflation of the number of cancellation. By definition an order entered in the dark market cannot create a false impression of demand as it is not visible to other market participants. The cancelling of such orders would, without more, seem to be consistent with bona fide orders which have been subsequently cancelled for non-manipulative reasons. Neither expert gave any indication as to the breakdown between lit and dark market trading although it may be inferred that the majority was conducted in the lit market. Nevertheless, this criticism of the metric is well made and is likely to give an inflated perspective of cancellations.

50. It may be accepted that a liquidity provision trading strategy may result in high rates of cancellations and that dark side cancellations may have inflated the metric. Nonetheless, these are not complete answers to this metric. The primary judge did not err in saying that it can be concluded that on the relevant trading days Select Vantage traders’ exhibited rates of cancellation that were significantly higher than the balance of the market on average.

Metric 4: Low Execution Probability

51. Orders which are indicative of layering will generally be placed in such a way that they deliberately have low execution probability. Execution probability can be assessed by looking at the number of price steps between an order and the touch price or where in the queue an order is positioned to execute at a particular price point. The ASX and Chi-X give execution priority to an order based first on price and then on a first in time basis at each price step.

52. The primary judge summarised Professor Putnins’ conclusion as follows (at [215]-[218]):

[215] Professor Putnins’ fourth characteristic is a low probability of execution. To measure the probability of Select Vantage’s orders being accepted so as to result in an executed trade, the Professor calculated the distance of each of its bids away from the best price on the order book. The distance was allocated one unit where the order was one price step away from the best order; that is, away from the highest price on the bid side or from the lowest price on the ask side. The distance was allocated two units for any order that was two price steps away from the best, and so on. Professor Putnins averaged these distances for all of Select Vantage’s buy orders and, separately, for its sell orders. The average across all stock/days for the bids was a distance of 1.8 and for the asks a distance of 0.2.

[216] The following are Professor Putnins’ conclusions on this measure:

Placing orders in the market that deliberately have low execution probability is not a feature of a typical legitimate short-term speculative trading strategy. Short-term speculators that have a prediction about future price movements will generally want to execute trades to try and profit from their predictions and therefore are likely to place orders such that they have a reasonably high probability of resulting in trades. One cannot profit from orders that do not execute, unless those orders are used to obtain favourable execution prices for other orders and are therefore not intended to execute (eg layering orders). […]

Layering orders are placed in the market in such a way that they deliberately have low execution probability: either at price steps behind (away from) the best quotes, or at the back of a long queue of orders [with earlier time priority] at a given price step. If low execution probability is achieved by placing orders away from the best quotes, we should see lower price aggressiveness (orders further from the best quotes) on the side that contains layering orders (the side with more quoted volume). […]

Given the results for [the second characteristic] indicate excessive bidding activity by [Select Vantage], consistent with layering the bid side of the market but not the ask side, if some of the bid orders are used for layering, we would expect to see lower execution priority [scil probability] for [Select Vantage’s] buy orders than its sell orders.

The results [for the fourth characteristic] show that the tendency described above is indeed a feature of [Select Vantage’s] quoting activity. [Select Vantage’s] buy orders tend to be placed further behind the best quotes than their sell orders, giving [Select Vantage’s] buy orders lower execution priority [scil probability] than their sell orders.

The results indicate that Select Vantage places its buy orders in the market in such a way that they are likely to have lower execution priority/probability than their sell orders.

[217] Professor Putnins found that Select Vantage’s placement of its buy orders further away from the best quotes than its sell orders correlated with significantly lower execution rates actually achieved for its buy orders, relative both to the average execution rates for all limit orders, buy and sell, of all traders (36.61%) and to the average execution rates for Select Vantage’s sell orders (13.33%). In Professor Putnins’ opinion:

This evidence further supports the notion that most of the buy orders placed by [Select Vantage] behind the prevailing best bid are not intended to execute.

[218] Professor Putnins accepted that the further a bid is away from the best price and the lower its execution probability, the less influence it is likely to have on moving the price. However, he considers that bids first placed relatively close to the best price would be influential. Select Vantage traders’ high rates of cancellation and amendment of bids led to a large volume moving to positions of low execution probability.

53. Professor Putnins concluded that the rate of bid execution for Select Vantage traders was 3.99 per cent and for ask side execution the rate was 13.33 per cent. In the case of the latter that rate was marginally higher than the average ask side execution rate for the rest of the market (10.42 per cent). Professor Putnins opines that Select Vantage traders are 334 per cent more likely to execute a trade on the ask side than the bid side on average. He also considers that the rate of 3.99 per cent is considerably lower than the execution rate for the balance of the market on the bid side where the limit order is behind the best prevailing bid (11.89 per cent). Only limit orders (as opposed to market orders) were included in this metric.

54. It is said that from the low rate of execution it can be inferred that a large portion of bid orders was not made with an intention to trade and are consistent with layering. When this is combined with the available inferences from metrics one and two it is said that what emerges is an image of persistent layering of non-executing orders on the bid side of the market.

55. Mr Schlaepfer challenges the primary judge’s reliance on this metric on the basis that it is consistent with legitimate trading by speculators and is explained by the short selling and inventory reversion constraints.

56. Professor Putnins accepted that speculators who have a view as to where ultimately a stock price may land may enter a number of limit bid orders at the beginning of the trading day at lower price steps in the hope of obtaining favourable executions as a result of having high queue priority at that price point. However, the entering of new limit orders of this kind is less likely as the day progresses where there exists a restriction on carrying stock overnight because, as explained by Professor Putnins, there is a greater risk of adverse outcomes should a bid order be executed towards the end of the day’s trading. One would expect that such orders would be cancelled as the day progresses as regardless of the speculators’ ultimate view as to the direction of the stock price, if one of these limit orders trades near day’s end there may be insufficient time for the stock price to rise before the trader must aggressively reverse his position potentially at a loss. It can be accepted, as Dr Carr opined, that traders may have particular strategies that they implement in the closing auction at the end of the day’s trading in order to obtain favourable trades; but the risk remains.

57. Professor Putnins rejected the suggestion that this would provide a total answer to the triggering of this metric, as he was of the view that one would see higher rates of execution if the purpose of the limit orders were to secure queue priority at lower price steps. Professor Putnins’ position appears to be borne out when one considers that the rest of the market achieved executions on the bid side three times as often as Select Vantage traders when they placed limit orders behind the best prevailing bid (3.99 per cent compared to 11.89 per cent). Although this might be the result of the rest of the market being more aggressive with their limit orders, that reinforces the idea that limit orders placed at great distance from the touch price are less likely to be legitimate as the chance for execution is extremely low. Furthermore, this casts doubt on whether the entering of the limit order bids are designed to obtain stray favourable executions when they appear to have lower queue priority than the rest of the market which is contrary to their stated purpose. The primary judge put the issue like this during Professor Putnins’ cross-examination:

“Q. If you measure aggressiveness just by price level, then this comparison of percentages … doesn't necessarily indicate that the rest of market has been putting on its bids behind the best at a higher price than SVs, but it may indicate that just in some other way the rest of the market is acting in such a way to make it significantly more likely that its bids behind the best will in due course be filled?

A. That's correct, your Honour.

Q. Relative to SVI?

A. That's correct, your Honour, right.

Q. By a factor of three-fold?

A. That's right.”

58. It can be accepted that the entering of lower price bids in the hope of stray executions at favourable prices could form part of a legitimate speculative trading strategy. Coincidentally such conduct would generate the appearance of layering the buy side of the market. However, if that were the case then one would expect the traders to achieve rates of execution that are at least remotely similar to other market participants who are placing bids behind the touch price. That has not occurred here. Neither the short-selling constraint nor the restriction on holding inventory overnight explains this particular variance in those execution rates.

59. The rate of execution for ask side orders (13.33 per cent) slightly exceeds the average for the balance of the market (10.42 per cent). When compared to the rate for bid side orders it suggests that Select Vantage traders were more aggressive on the ask side of the market by placing orders closer to the touch price and/or dealing with them in a way differently from bid orders to ensure queue priority. The latter could include not amending orders which has the effect of refreshing the order’s position in the queue. But it may be consistent with Select Vantage traders reversing out of long positions aggressively in order to take advantage of inflated prices. It could also be consistent with Select Vantage traders reversing out of the same position in order to comply with the constraint regarding holding stock overnight should the traders be in a long position at the day’s end.

60. This metric is capable of supporting an inference of layering conduct. The presence in the data of very low rates of bid execution is consistent with the entering of non bona fide orders in order to build a false impression of demand. The suggested explanation that this metric is consistent with legitimate trading and the short-selling constraint is not compelling in light of the very low rates of execution comparative to the balance of the market when entering bids behind the touch price.

61. I consider this to be a strong indicator of layering which allays the reservations expressed above concerning metrics 1, 2 and 3.

Metric 5: Inventory Reversals

62. This metric was rejected by the primary judge and is not the subject of a notice of contention. The primary judge said that this metric could not be relied on (at [219]):

Professor Putnins’ fifth characteristic is that of buying volume during a trading day being approximately equal to selling volume. I do not put any weight on that characteristic in the present case because, although it is present, it is attributable to Select Vantage having adopted a business model of not permitting its traders to hold any stock overnight and requiring them to close out all positions at the end of each day’s trade. There could be sound commercial reasons for limiting activity to day trading in this manner and I do not regard this feature as supporting an inference of market manipulation.

Metric 6: Trades oppose Quotes

63. This metric looks to the imbalance between communicated trading intention and actual trades. The metric tests whether, at each time that the Select Vantage traders executed a trade, the dollar volume of resting orders was on the same side of the market as the executed trade. The primary judge summarised Professor Putnins evidence in this way (at [220]-[221]):

“[220] Professor Putnins’ sixth characteristic is that of trades commonly being executed on the opposite side of the order book to the side on which there is an imbalance of resting orders. He made a count of how many trades were executed by Select Vantage on the sell side at a time when the dollar volume of its resting bids outweighed the dollar volume of its resting asks. Similarly, he counted how many trades were executed on the buy side when the balance of resting orders was the other way. He also counted the number of trades that were executed on the same side as the imbalance of resting limit orders. Professor Putnins then calculated what percentage of the total number of trades was represented by those that were executed on the opposite side to the imbalance. The range was found to be between 25% and 56% with an average of 48%.

[221] Professor Putnins expressed these opinions:

Executing trades on the opposite side of the market to the traders’ resting order imbalance […] is unlikely in a typical legitimate short-term speculative trading strategy. At a given point in time, a speculator with a prediction about the future price movement will typically want to buy or sell, but not be bidding to buy and then actually selling, or vice versa, offering to sell but actually buying the stock. […]

Market participants engaging in layering will often have trades executed on the opposite side of the market to their resting order imbalance […]. Put differently, layering often involves trading in a direction that is opposite to what trader’s order imbalance would signal as their intended direction of trade because their order imbalance is not a true reflection of their intentions.

To put [the average of 48% as measured for Select Vantage] into perspective, note that many legitimate trading strategies will have a score of 0% on this characteristic […].”

64. The existence of a dollar imbalance in favour of resting orders on the bid side as against executed asks is said to be consistent with layering in two ways. First, it is said to aid in the drawing of an inference that the resting bid orders are not intended to be executed as they are inconsistent with actual trading intention manifested by the executed trades. Secondly, it is said to indicate that the trader is executing inflated trades on the ask side as a result of the false impression of bid side demand, thereby changing the behaviour of other market participants.

65. In 2014 Select Vantage had about 1800 traders around the world and they placed approximately 400,000 orders per day. Select Vantage used a surveillance tool called “Ginger” that screened transactions with filters designed to reflect rules and regulations around the world. (at [75], [76]). Mr Schlaepfer gave evidence that that software would raise an alert if collaborative trading were identified where a number of traders in the one pod placed a substantial number of resting limit orders for stock listed on a particular exchange in order to carry out a layering strategy (at [78]).

66. Mr Schlaepfer placed significant weight on the fact that although there were incentives for remuneration within a particular pod, there were no incentives that would follow from the aggregation of the performance of multiple pods. But as a single person or entity could own multiple pods, and then appoint a single person to oversee the pods, this consideration is of little weight.

67. Professor Putnins’ opinion was challenged on a number of grounds, one of which was his use of zero as a relevant comparator. In Professor Putnins’ view zero was the appropriate comparator by reference to the expected behaviour of speculators and liquidity providers. He said:

“Metric 6: ‘Quotes opposes trades’

Speculators

‘Executing trades on the opposite side of the market to the trader's resting order imbalance (e.g., having more buy orders resting in the market but then actually executing sells and vice versa) is unlikely in a typical legitimate short-term speculative trading strategy. At a given point in time, a speculator with a prediction about the future price movement will typically want to buy or sell, but not be bidding to buy and then actually selling, or vice versa, offering to sell but actually buying the stock.’

Liquidity Providers

‘Liquidity providers will at times execute trades that are in an opposite direction to their resting order imbalance (e.g., buying when they have more sell orders in the limit order book or vice versa) because they typically quote both sides of the market and trades could occur on either side. However, due to the inventory control mechanisms of liquidity providers, they are likely to more often trade in the same direction as the imbalance in their resting orders. Liquidity providers typically modify their resting orders to revert their inventory towards zero. For example, when a liquidity provider has accumulated a long position in a stock, they will typically want to sell the stock and therefore are likely to provide more liquidity on the ask (sell) side and vice versa.’”

68. Mr Schlaepfer argued that the use of zero was inappropriate as genuine liquidity providers could also generate a non-zero result on this metric. Dr Carr opined that many short-term traders could produce results on this metric of around 50 per cent. Professor Putnins was cross-examined on the provenance of the comparator of zero and its utility. In cross-examination he gave the following evidence:

“Q. And assume that the liquidity provider places a buy order of 50,000 lots at 89 cents and a sell order for 100,000 lots at 91 cents. Would you make those assumptions?

A. 50,000 at 89 and what did you say, 100,000 at 91?

Q. 100,000 at 91. So buy order at 50,000 at 89, sell order 100,000 at 91. So on those assumptions, there is a sell side order imbalance at that time, isn't there?

A. That's right.

…

Q. … roughly speaking, over time, over a stock over time, you'd expect to see, wouldn't you, the stock moving upwards about 50% of the time and downwards 50% of the time?

A. Over time, yes, but not in a particular circumstance where you're providing a particular order imbalance. That doesn't hold.

…

Q. But purely by statistical chance, there is a chance that the price will move down to fill the buy order. That's right? Of 50,000 shares.

A. There is a chance, that's right. It's not a 50% chance.

…

Q. So on that circumstance you have a trade on the opposite side of the resting order imbalance, do you not?

A. That's correct.

…

Q. None of the articles that you've referred to in your report provide support for the assertion that you would expect to see 0% for liquidity provider on this score?

A. I can find references if you want references. What I'm saying is that my discussion in section 6 that talks about what is likely or unlikely to be observed for various legitimate trading strategies isn't something that I've just created on my own imagination. This is underpinned by the market microstructure literature on how trading plays out in markets. As to while there isn't a specific reference given for this paragraph, these are sort of - this is economic reasoning that is supported by the field. Now, what that implies, right, there's probably no paper that writes down that if you were to compute it to metric with an implied score of zero, what that implies follows by logical reasoning.”

69. It may be accepted that the use of the comparator of zero could exclude legitimate trading that inadvertently triggered the metric. Nonetheless it is a question of degree. As Professor Putnins said in cross-examination:

“Q. The hypothesis is that anything more than 0% is something that will trigger this metric; is that right?

A. Anything significantly above 0% in both a statistical and an economic sense, so small variations from zero we discussed last week would be economically meaningless because by chance you would expect them to be triggered now and then under perfectly legitimate trading strategies. It’s when you see a substantial deviation from zero.”

70. Mr Schlaepfer also submitted that the metric was consistent with the short-selling constraint on Select Vantage’s traders. As with Metric 4, the suggested explanation that the metric was consistent with legitimate trading and the short selling constraints is not compelling in light of the very low rates of execution.

71. Mr Schlaepfer’s third challenge was that Professor Putnins’ methodology of aggregating all Select Vantage traders’ trading activity could generate false triggers on this metric. He submitted:

“So one particular point about aggregating all of the traders together is there was evidence which I'll give your Honours the references to shortly, but there was evidence that the traders were prohibited from collaborating with each other. They were told they weren’t allowed to communicate with each other; they were not allowed to do that. As soon as one aggregates all of the data, all of the tradings, one is implicitly starting from a proposition that they are collaborating, and yet that was very much in issue as we see later on. So there's a real problem with circularity in the way in which Professor Putnins approached the matter.”

72. It is said that if one aggregates all the trading activity the myriad of different trading strategies being employed may result in a false positive which is only consistent with layering if it is found the traders were colluding. It may be, for example, that large numbers of resting bid side orders are attributable to speculators who never obtain fills whilst the comparatively smaller (in dollar volume) ask side trades are attributable to liquidity providers. Professor Putnins conceded that it was possible that false positives would arise on this metric due to this issue.

73. This critique would be well made if there were no evidence or basis for inferring collusion between traders. But if collusion can be inferred as, for the reasons below, I consider it should, then this metric is supportive of the primary judge’s finding of layering.

Metric 7: Cancels oppose Trades

74. It was said that manipulators engaged in layering would often cancel orders on one side of the market after they executed an order on the opposite side of the market. The logic for this is that if layering orders are never intended to be executed, they will be cancelled once the market has been pushed in the appropriate direction. The primary judge summarised Professor Putnins’ evidence regarding this metric as follows:

“[222] The seventh characteristic is that of cancelling orders on one side of the market after executing an order on the opposite side. Professor Putnins counted both the number of bid orders cancelled after an ask had executed and the number of sell orders cancelled after a bid had executed. He also counted the numbers of orders that were cancelled on the same side as that on which a trade had been effected. Professor Putnins then worked out what percentage of all orders cancelled following trades was represented by those that cancelled on the side opposite the transaction. The percentage ranged from 40% to 70% and averaged 52%.

[223] Professor Putnins gave these opinions about this seventh characteristic:

Manipulators engaging in layering will often cancel orders on one side of the market after they execute an order on the opposite side of the market […]. This occurs because layering orders are not intended to execute and once they have served their purpose of pushing the market towards the price of any bona fide orders resulting in executions, the layering orders can be cancelled. […]

[This] is unlikely in a legitimate short-term speculative trading strategy unless the cancelled orders were not intended to execute but were instead intended to get the favourable execution price for the trade (layering). In particular if the speculative strategy is an intraday one […] in which the intention is to reverse inventory towards zero by the end of the day, then after having bought the desired quantity a speculator would typically want to sell that position for a profit, or after having sold the desired quantity a speculator would want to repurchase that position for a profit. Such trading would result in the opposite characteristic: placing sell orders after buying (not cancelling sell orders) and placing buy orders after selling (not cancelling buy orders). […]

To put [Select Vantage’s] numbers into perspective, note that many legitimate trading strategies will have a score of 0% on this characteristic […]. Therefore, cancelling orders following trades on the opposite side of the market, a characteristic of layering, is consistently present in [Select Vantage’s] activity on all stock-days and accounts for a considerable fraction of [Select Vantage’s] order cancellations on individual stock-days as well as overall.”

75. Dr Carr criticised Professor Putnins’ approach to this metric on the ground that there was no differentiation between cancellations that took place immediately after an executed trade, and those that were cancelled minutes or possibly hours later. That criticism was misplaced and was due to Dr Carr’s description of layering as involving rapid cancellations whereas Professor Putnins was of the view that that was not the only way that layering could be used in markets. He said that layering could be used in markets in a way that did not require cyclical patterns.

76. Dr Carr also criticised this metric for failing to take account of changes in information or other changes in the market which may have led to all market participants adjusting their position. It is unnecessary to consider in any detail Professor Putnins’ response to that criticism. Metric 7 is not compelling where ASIC’s case was not one of cyclical layering. There was no imperative to cancel orders after a trade as there was no attempt to layer the ask side of the market.

Metric 8: Dark opposes Lit

77. The primary judge did not rely on this metric in finding that Select Vantage’s traders had engaged in layering conduct (at [224]). His Honour said:

“I found Professor Putnins’ eighth characteristic, regarding Select Vantage’s use of the unlit market, of some additional weight. However, I do not find it necessary to rely upon that element and will refrain from discussing it. To do so would require a description at length of the operation of the unlit market.”

78. A dark market is one where the trades and orders entered by individual market participants are not made public to other market participants. This stands in contrast to the lit market where the other participants can observe the trades that take place and see the level of supply or demand in a stock at the various price levels displayed in the limit order book. In this metric Professor Putnins calculates the proportion of times Select Vantage traders’ dark order imbalance opposes their lit order imbalance. This is said to demonstrate inconsistent trading intentions, as on the lit market Select Vantage traders may be entering orders on the bid side to a greater degree than the ask side (see metrics 1 and 2) whilst doing the opposite in ‘secret’ on the dark market.

79. Professor Putnins determined that the percentage of time that Select Vantage traders’ positions in the respective markets were inconsistent varied between 8 per cent to 100 per cent over 22 trading days. On 1 January 2019 there were no orders by Select Vantage traders in GCN on the dark market within the best 10 price levels. The average over the 22 days was 43 per cent. This is said to be consistent with layering in general and particularly in this case where it is alleged the lit market imbalance was persistently on the bid side which was said inter alia to create a false impression of demand. The presence of the imbalance indicates that Select Vantage traders were attempting to reverse out the long positions secured in the various stocks in secret so as not to undermine the image of demand on the lit market.

80. Mr Schlaepfer challenged Professor Putnins’ use of zero as the appropriate comparator. For the reasons above concerning use of the zero comparator, it is possible that legitimate trading could trigger the metric. Nonetheless, it is a question of degree. Professor Putnins’ evidence revealed that on 22 of the trading days, Select Vantage’s traders were, 42 per cent of the time on average, entering orders which generated an imbalance inconsistent with the lit market order book. If it can be concluded, as I think it can, that there is collusion between traders, then this metric is a strong indicator of layering as trades executed in this way would not undermine the image of bid side demand.

314. In those circumstances I am not persuaded that mere correlation provided a basis for inferring coordination. The regression analysis was subject to a similar problem, because clustering errors by stock and by day would not account for (innocent) correlations referable to the factors discussed.

315. Finally, ASIC points to the unlikelihood of hundreds of independent traders converging on the same thinly traded stocks on the same days. What might motivate such strategic behaviour, and therefore explain its adoption by a large number of traders, is explained below. Again, for sole traders with limited training and resources, the set of viable strategies was probably small. That is, of course, also a reason to think coordinated manipulation might prove attractive; there is force in the primary judge’s conclusion that Select Vantage’s trading model was in that sense “objectively calculated” to encourage traders to engage in coordinated manipulation: at [256]-[257].

316. In my assessment, however, the independent evidence of coordination is equivocal. If the trading activity data demonstrates that market manipulation likely took place, for the reasons given above – the number of different traders whose activity the data captures, and the absence of an incentive for traders who were not already long to push up prices – that manipulation was likely coordinated. But that trading data must be analysed without the benefit of a pre-existing assumption that coordination was likely rather than merely possible.

A legitimate trading strategy?      

329. Before turning to Professor Putnins’ metrics of layering, it is necessary briefly to explain the competing innocent explanation offered for the presence of those metrics in Select Vantage’s trading data. The suggestion is not that all Select Vantage traders adopted the same strategy (Mr Schlaepfer’s case is that they were independent) but that reasons could be given for why they might decide to pursue similar legitimate strategies that would generate the patterns of activity seized on by ASIC.

330. After ASIC had communicated its concerns arising from Dr He’s analysis of Select Vantage’s trading in OIL stock over several days from 17 November 2014, Mr Kruyne (as already noted, an employee of a Canadian subsidiary of Mr Schlaepfer’s group of companies) undertook his own analysis of the impugned trading in the hope of allaying the concerns ASIC had raised. He set out the results of that analysis, together with an explanation of features of Select Trading’s method of trading, in an email which was sent to Macquarie on 26 November 2014. The following account is drawn primarily from that explanation.

331. Mr Kruyne contended that Select Vantage’s increased activity in OIL stock was “directly related to publicly available news or increases in volume & volatility”. He said that news was important to Select Vantages traders (who, as already noted, were day traders) because it would “draw other market participants to the stock and allow [Select Vantage] traders to provide liquidity and try to capture the spread” (in place of making long-term directional bets). In this context, the “spread” refers to the difference between the limit (that is, listed) price of the highest bid and the lowest ask resting on the order book. If there were market participants willing to “cross” the spread by meeting the terms of an outstanding limit order, traders would be able to “capture” the spread by buying and selling equivalent quantities of a stock over the course of the day. In placing resting limit orders for other traders to fill, they also served as liquidity providers. Mr Kruyne said “Given the intraday nature of the trading done by [Select Vantage] traders, the focus of these individuals is always on market microstructure, and intraday metrics (volume, volatility, bid/ask size) rather than fundamentals such as market cap or earnings”.

332. Traders following such a strategy (seeking to provide liquidity to more aggressive traders) would look for stocks with relatively high trading activity. Two useful indicators of activity would be recent changes (at or around market open or the previous day’s close) in trading volume or pricing volatility. Select Vantage traders focused their attention on non-indexed and low-priced stocks, to avoid competing with high-frequency trading firms and to take advantage of the fact that Australian equity markets’ fees are calculated in terms of basis points rather than per share.

333. A trader following this strategy would obviously seek to provide liquidity on both sides of the spread. But Select Vantage required its traders to close out their positions at the end of each day, so they were unable to enter orders on the ask side unless they had either already purchased stock on that day or were able to borrow stock to assume a short position. As already noted, there was generally not a pool of shares from which traders could borrow.

334. It was therefore necessary for Select Vantage traders to place bids, and have those bids wholly or partially filled, before they could place asks and attempt to “capture the spread”. Typically, they would place orders at or before opening, to secure execution priority, and leave their orders on the books in the hopes of an aggressive fill (bearing in mind that these stocks were identified as having higher than usual trading volume or pricing volatility). It followed that they would necessarily place a higher number of bids than asks, and that their orders would usually remain on the book, until executed or cancelled, for an extended duration.

335. ASIC responded to that explanation by submitting that there would be no reason for traders adopting a liquidity provision strategy to place bids significantly below the touch price. Their goal was (ex hypothesi) to trade, which they could not do until they acquired stock; there was therefore no reason to enter and leave open large bids with a very low chance of being filled, which would be counted towards their internally imposed limits of three orders on each side of the market, and the amount of capital with which each trader could trade.

336. As to the order limit, which applied per stock, it is not implausible that inexperienced day traders might prefer to place low-probability but possibly high-reward bets over attempting to buy in at the price of the best prevailing bid at any given time. The gist of this explanation was put to Professor Putnins. The suggestion was that traders might leave non-aggressive bids in place over the course of the day, so that they would enjoy queue priority if the market price fell. That assumed, possibly on the basis of the historical price of the stock, that if the market price fell it was likely subsequently to rebound, allowing a more profitable sale. Otherwise there would be no reason not to place an order at or close to the touch price and exploit the existing spread.

337. Professor Putnins rejected that assumption as the basis of a legitimate strategy, because it “would imply that the market is not even weak form efficient, which I think [is] inconsistent with a large, large body of empirical literature”. That response made two untested assumptions, first as to the efficiency of capital markets for thinly traded, non-indexed low-cap stocks and secondly as to the rationality of Select Vantage’s traders.

338. Separately, Dr Carr speculated in his reply report that the trading activity “appears consistent with a mix of arbitrage and liquidity provision strategies”, the former being a reference to arbitrage taking advantage of price differentials between the ASX and Chi-X. He did not expand upon that cryptic observation in his oral evidence. Dr Carr’s position was ultimately that it was “virtually impossible” to account for the various different strategies and information flows which might explain the trading data. That does not explain the degree of correlation, which suggests that the diversity of strategies was not as great as Dr Carr was evidently prepared to assume.

Professor Putnins’ metrics

339. Turning to Professor Putnins’ ten metrics of layering, they were:

1. Unbalanced quoting: the resting limit orders at any given point in time were highly imbalanced by dollar volume, so that the difference between the value of resting bids and resting asks was a very large proportion of the sum. For Select Vantage, this proportion was greater than 50% on each stock/day – on average it was 85% – and much higher than that of other market participants, who averaged 18%. Professor Putnins considered that this imbalance was used to create a false impression of buying or selling interest in order to influence market prices.

2. High quoting activity: Select Vantage was responsible for a very high proportion of the volume of resting limit orders on the bid side of the market – between 10.5% and 28.5% over the relevant stock/days, averaging 20.3% – but very little on the ask side, on average 1.7%.

3. Abnormal cancellations: Select Vantage’s cancel-to-trade ratio, the ratio of orders cancelled to orders executed, exceeded that of other market participants by a clear margin on each stock/day. On average its ratio was 3.4, whereas the ratio was 0.8 for the rest of the market.

4. Low execution probability: Select Vantage was much more aggressive on the ask side than the bid side. On average, its bids were placed at a much greater distance from the touch price than its asks; and the average execution rate of its bids below the highest resting bid was 3.99%, much lower than the 11.89% rate for other traders’ bids below the highest resting bid and the 13.33% rate for its own asks above the lowest resting ask.

5. Inventory reversals: Select Vantage did not hold positions overnight. Professor Putnins treated this as consistent with but not greatly probative of layering. In fact it was required by Select Vantage of their traders, and for that reason was given no weight by the primary judge.

6. Trades oppose quotes: Select Vantage executed an average of 48% of its trades on the side of the market that was not the side with the larger volume of resting orders, whereas “many legitimate trading strategies”, including traditional liquidity provision, would score close to 0%.

7. Cancels oppose trades: a high percentage – from 40% to 72%, depending on the stock/day – of Select Vantage’s cancellations were of orders on the opposite side of the order book to the most recent trade, whereas, again, traditional liquidity provision would score close to 0%.

8. Dark opposes lit: averaged over all relevant stock/days, [2] Select Vantage’s order imbalance in the dark market, where limit orders were not displayed (but might still be executed against), was in the opposite direction to its imbalance in the lit market 43% of the time. The obvious inference is that Select Vantage tended to have an ask-side skew in the dark market, where its orders could not readily create a false impression of demand, slightly under half the time.

9. Quoting opposes inventory reversion: the trading of a liquidity provider would generally exhibit inventory reversion in the sense that their order imbalance would mirror their inventory position – that is, they would tend to quote more on the ask side when their inventory was positive, and vice versa – although not necessarily at all times, as Professor Putnins conceded. Depending on the stock/day, Select Vantage’s quoting activity was inconsistent with inventory reversion between 51% and 100% (and on average 93%) of the time.

10. Quoting opposes trading intention: as Select Vantage traders were required to be flat at the end of the day, their trading intention towards the end of the day must have been to wind down their accumulated inventory. But in the last half hour of the continuous trading phase of each stock/day [3] a buy-side resting order imbalance was observed at all times, save for one stock/day on which it was present 67% of the time.

    2. The dark market data was unavailable on one stock/day.

    3. There were two stock/days for which the measure could not be computed, because inventory was already zeroed by 3:30 pm.

340. Many of the differences between Select Vantage and the market average for these characteristics were statistically significant. That assists only in rejecting a hypothesis that is in any event plainly untenable, namely that Select Vantage’s traders in aggregate behaved like the average market participant.

The effect of aggregating trading activity data   

341. Mr Schlaepfer submits that metrics 6 and 8 may be an artefact of aggregating the trading data of traders acting legitimately and independently. Supposing that traders were acting independently, there would be no reason to expect that trades would not regularly (and randomly) oppose quotes, or the dark market imbalance oppose the lit market imbalance. That was Professor Putnins’ own evidence in his first report.

342. However, it was also Professor Putnins’ evidence that for many of the metrics, aggregating the trading activity data would have the opposite effect. If other traders were not layering (and their individual data therefore did not display characteristics of layering), aggregation would dilute rather than exaggerate most of the characteristics of layering in the firm-level data.

343. In his first report, Professor Putnins’ evidence was that “the two effects act in opposite directions” and it was difficult to say what the overall effect of aggregation would be on the likelihood of correctly observing the characteristics of layering (that is, observing the characteristics of layering when layering behaviour is present to some extent, and not otherwise). And he conceded that one effect of aggregation was that “some characteristics will be more useful than others in identifying layering”.

344. In his oral evidence, Professor Putnins’ qualified that opinion, suggesting instead that aggregating the data of “a large number, or even some number of independent traders” would generally result in fewer indicia of layering. That evidence may be accepted in the general terms in which it was expressed, although its force is limited in a case such as the present where it is already clear that many traders were trading in a similar fashion (as discussed above) and the question is whether that behaviour indicates sinister collusion or admits of an innocent explanation.

The proper approach to the metrics   

345. ASIC emphasised the point, made repeatedly by Professor Putnins in his evidence, that the metrics cannot be analysed, rationalised and explained away individually. What is necessary, so it was submitted, is to conduct a “probabilistic exercise” having regard to the joint likelihood of observing the presence of all of the metrics in Select Vantage’s trading activity data on the hypothesis that there was no manipulative activity.

346. It may be accepted that the Court must consider all of the evidence taken together. However, it would not be permissible to treat the improbabilities of observing each characteristic in legitimate trading activity as straightforwardly compounding. As I will explain, the probabilities are not independent.

347. For example, taking metrics 1 and 2 (unbalanced quoting and high quoting activity), if other market participants quoted in a roughly balanced fashion, as the evidence suggests they did, then a company trading on Select Vantage’s model, with a pronounced bid-side imbalance at most points in time, would necessarily have higher quoting activity on the side the subject of the imbalance. Given metrics 1 and 5 (unbalanced quoting and inventory reversals), the side the subject of the imbalance would have lower execution probability (metric 4). The high cancellation rate and low execution probability (metrics 3 and 4) are obviously related because higher cancellation rates reduce the likelihood of execution and orders less likely to execute are more likely to be cancelled. Metric 6 (trades oppose quotes) follows from metrics 1 and 5 (unbalanced quoting and inventory reversals) where, as here, the enduring quoting imbalance was on only one side of the market.

348. Given imbalanced quoting on the bid side, a high cancellation to trade ratio, and inventory reversals (metrics 1, 3 and 5), cancellations opposing trades (that is, cancellations being of orders on the opposite side of the order book to the most recent trade: metric 7) is not itself surprising. This is not a case in which particular cancellations following shortly after trades are identified as suspect – indeed, the evidence was that the number cancellations surrounding a fill was relatively low: primary judgment at [243]. And if resting bids had a low execution probability relative to asks (metric 4), there would be no reason for cancellations not to oppose trades, or for quoting activity by volume to be consistent with inventory reversion (metric 9). The same point might be made, albeit with less force closer to the day’s close, about metric 10 (quoting opposes trading intention).

349. Professor Putnins’ metrics emphasise how different Select Vantage’s trading patterns were from the average of all market participants, although it is difficult to know how much significance to attribute to the mere fact of that difference. For some metrics, “many” legitimate strategies would result in scores close or equal to zero, but how many was not explained. Metrics 8 and 10 raise further questions about the genuineness of at least some of the low-priced, high-volume bids in the lit market. Ultimately, however, the extent to which the ten metrics do more than show that Select Vantage traders placed and maintained over the course of the day large numbers of low-priced bids that were cancelled or left on the books rather than executed is much slighter than was acknowledged by the primary judge.

Insufficient evidence for a finding of manipulation

350. Dr Carr addressed Professor Putnins’ opinion in detail in his “rebuttal” report. In summary, he was of the view that Professor Putnins’ conclusion (“that it was likely that Select Vantage engaged in a form of market manipulation known as layering”) was not reliable because his measures did not identify particular orders and trades that contained the basic elements of layering as commonly understood by regulators and market participants. Dr Carr did not consider that the opinion that traders had created and maintained artificial prices could be sustained in the absence of any identification of particular orders that created artificial prices and a benchmark “but for” price for the relevant stock or at least some statistical certainty of deviation from the “but for” price.

351. While Dr Carr analysed the 10 metrics in detail in that report, it is not necessary here to address every point. It is enough to highlight some key points.

352. Professor Putnins demonstrated that Select Vantage’s order to trade ratios were high (between 3.1 and 10.8 on the relevant stock/days) and indeed considerably higher than those of the rest of the market (on average around 250% higher than the average for other market participants). Dr Carr expressed the opinion that the levels of cancellations by Merlito traders observed in ASIC’s defence dataset were within the reasonable range of cancellations observed in equity trading in the US and Australian equity markets. Professor Putnins explained that Dr Carr’s opinion on that issue did not have adequate regard to differences in market structures between Australia and the US.

353. But Dr Carr also gave persuasive evidence that, on most stock/days, other market participants had higher rates of cancellation than Select Vantage, and cancelled orders more quickly after entry. Elsewhere, Dr Carr explained that the time sensitivity of order placement was a significant factor in that analysis; he said Professor Putnins took no account of that. The primary judge dismissed Dr Carr’s evidence because Professor Putnins’ other indicia of layering were present in Select Vantage’s trading activity data, which diminished the likelihood that there was an innocent explanation for Select Vantage’s cancellation rates (as opposed to those of the other market participants, whose high cancellation rates “may have some other explanation”): at [240].

350. With respect, that reasoning is flawed. The question was not, what is the likelihood that there is an innocent explanation for Select Vantage’s higher cancellation rates, conditional on the presence of the other metrics? Obviously, the likelihood of an innocent explanation for any one metric was lower given the presence of the others. The question was whether the likelihood Select Vantage was engaged in manipulation was informed by the presence of high cancellation rates, conditional on the presence of the other metrics. The significance of Dr Carr’s evidence was that, while Select Vantage’s cancellation rates were high, they were not outliers on the relevant stock/days. Professor Putnins’ analysis of metric 3 asserted that “layering orders are not intended to execute so they will typically end in a cancellation, and may also need to be cancelled and resubmitted as market conditions change”. Dr Carr noted that this analysis acknowledged the need to cancel and resubmit orders as market conditions change but that nowhere did Professor Putnins conduct any analysis of the changing market conditions or identify a proper benchmark as to what was an “abnormal” cancellation rate as opposed to a normal rate. Dr Carr’s analysis indicates that the degree to which the third metric (abnormal cancellations) is independently informative is low (if non-zero).

351. Metric 4 (low execution probability) is based on Professor Putnins’ claim that “layering orders are placed in the market in such a way that they deliberately have low execution probability: either at price steps behind (away from) the best quotes, or at the back of a long queue of orders at a given price step”. Dr Carr’s opinion was that the use of the term “deliberately” in that assertion had no economic basis. While the evidence was that the execution probability of Select Vantage’s traders’ bids was low relative to the rest of the market, it was not shown that the execution probability was low relative to the quoted prices. It remains possible that the low execution probability was largely or wholly attributable to the distance of Select Vantage bids below the best quoted bid. Dr Carr noted that traders place limit orders away from the touchline “intending and hoping that their orders get executed because market prices fluctuate”. In other words, the placement of limit orders at a point away from the touchline does not in itself indicate an intention that the order not execute. Select Vantage’s high cancellation rates are not informative on the point, because those cancellation rates might in turn be the result of Select Vantage’s bids being systematically lower than the rest of the market (and hence more likely to be cancelled rather than executed). Separately, Dr Carr noted that orders having a low execution probability were less likely to have a price effect on the market (and so not obviously part of a “deliberate” layering strategy).

352. Professor Putnins’ analysis does show that Select Vantage’s bids were much less aggressive (that is, much further from the trading price) than their asks, but that is not suspicious in itself. Leaving high-volume, high-priced asks on the order book, on the assumption that a reversion to the historical trading price band would be likely should the order be filled, was not an available option in view of the inability to short. It is far from obvious that Select Vantage traders could be expected to adopt a symmetrical volume and pricing strategy in view of the asymmetrical constraints under which they operated.

353. The divergence between dark and lit market behaviour is at least superficially suspicious. Dr Carr’s evidence in reply was that, by his calculations, Select Vantage’s cancellation rate was higher in the dark than lit market for both bids and asks, and accordingly that it was not clear that the dark market was, by Professor Putnins’ standards, more reflective of traders’ true intentions. Professor Putnins gave oral evidence that Select Vantage’s cancellation rates in the dark market were not meaningfully different from those of the rest of the market on the relevant stock/days. The mere fact that Select Vantage’s cancellation rates were higher in the dark than lit market was “nothing surprising. It’s a different market”. There were technical reasons for there being higher cancellation rates in the dark market. On his evidence it was like comparing the Australian and Canadian (or American) equity markets.

354. Accepting that to be so, at that point it is difficult to see how the quoting imbalance in the dark market could be significantly informative about the true trading intentions of Select Vantage traders in the lit market. It would be entirely consistent with Professor Putnins’ evidence that traders using the dark market differed systematically in their trading strategies or sophistication from those placing orders only in the lit market. As was already observed, he conceded that the eighth characteristic could have been generated by aggregation of the trading data of traders acting legitimately and independently, and that it might therefore be less useful than the other metrics in identifying manipulative behaviour.

355. The tenth characteristic is the most difficult to explain innocently. Dr Carr noted in reply that orders placed during the last half hour were consistent with inventory reversion and suggested that the statistic was an artefact of the inclusion of orders placed earlier and left on the books during the last half hour. That only underlines the question why those orders were left on the books, rather than cancelled, when their execution would not have been desirable. It suggests as one possible answer that the orders were not then intended to execute because they were never intended to execute.

356. In oral evidence Dr Carr made two points about this characteristic. The first was that during the last half hour of continuous trading, cancelling bids left on the books would not be traders’ primary concern, and would be necessary only if there were a real risk of the bids executing. Presumably in all the circumstances that risk was low. His second point was that traders would have another option to close out a long position in the closing auction, and that it was not unheard of for traders to have “actual strategies based on what happens at the closing auction”. What those strategies might be, and whether it was likely that any number of Select Vantage traders were pursuing them, was not explained.

357. Stepping back to consider the suggested layering strategy in general terms, there remain questions about both its appeal and its efficacy. The profit-sharing arrangements required to give an incentive for cooperation between traders or pods who were not yet long would be complex and possibly difficult to implement. In the absence of such arrangements there was no reason for a trader to engage in layering behaviour until he or she, or at least his or her pod, had a long position.

358. For the strategy as described by Professor Putnins to be sensible, it was necessary for Select Vantage to amass a very considerable long position at the beginning of the day and to elevate prices considerably while it was unwound. Once the initial position was unwound, unless and until the manipulative bids were cancelled or amended, or supervening and presumably unpredicted events pushed prices back down, any further trades would simply be trades around the spread (but centred on a higher trading price). At that point there would be no reason, other than to escape detection for layering, consistently to maintain a high volume of non-genuine bids over the course of the day. Given the prescriptiveness of the layering detection software monitoring traders’ behaviour, it would still have been possible to cancel bids after the initial position had been unwound and to repeat the same strategy over a shorter time horizon.

359. There was evidence neither of the size of the initial long position Select Vantage traders would accumulate nor of whether it was acquired after, before or simultaneously with the allegedly manipulative conduct commencing. Nor was there any evidence of whether that position was in fact slowly wound down over the course of the day, or whether there was continual churn as traders sold and then rebought shares in order to continue to profit from the spread. That Select Vantage was always or almost always long, a fact referable to the short constraint, does not distinguish between those two possibilities. If it was the latter, the suggested layering strategy was at best implausibly unambitious.

360. Further, there was no evidence that the suggested strategy was successfully executed, in the sense that on all or some of the stock/days Select Vantage managed first to acquire a large long position and then to unwind it over the course of the day at elevated (or elevated and rising) prices. As noted earlier in this judgment, there was a reference in the email concerning ASIC’s preliminary analysis to a “positive correlation between the change in [Select Vantage’s] contribution to the order book imbalance and the mid-point price of the relevant securities”, but that evidence was not admitted for the purpose of proving the truth defence (a point made in the grounds of appeal).

361. Dr Carr’s report included some evidence that he had been unable to discern any price effects on a relatively short timescale. That evidence was persuasively criticised by Professor Putnins on the grounds that it was extremely difficult to isolate any causal link, given the number of factors which might plausibly have an effect on market pricing. Nevertheless, some evidence that Select Vantage’s trading patterns had price effects, or even just that there were price increases from which Select Vantage benefited over the course of the relevant stock/days, would have been of assistance. I appreciate that the market manipulation provisions do not require proof of actual price effects; it would be sufficient that there were acts or transactions likely to have that effect. But it would be easier to conclude that there was a reason for traders to engage in a manipulative strategy if there was evidence suggesting that it had been (and hence could be) successfully executed.

362. That Select Vantage’s trading patterns did not reflect other cases of layering is not, as has been observed, decisive. Professor Putnins gave evidence that layering strategies might differ between countries and markets: in less liquid equity markets, there is less depth at the best quoted bid price, so that the execution probabilities of bids at that price may be relatively high (and undesirably so for a trader engaged in layering). Lower liquidity also makes it easier to induce price movements. For those reasons, layering through lower priced bids left on the books for an extended period, rather than through bids at or close to the best quoted bid cancelled shortly after entry, might be more feasible in thinly traded stocks on Australian equity markets. There was also the speculative possibility that Select Vantage traders had adopted a novel form of layering after regulatory action led to the imposition of stricter internal monitoring, making standard forms of layering less viable.

363. Accepting those possibilities, this Court is left with suspicious trading patterns that were only partially explained; one metric (the tenth) which demands and did not receive any satisfactory potential explanation, from Dr Carr or otherwise, and Professor Putnins’ opinion that in his view Select Vantage traders were likely engaged in a form of layering. ASIC did not even attempt to demonstrate that Select Vantage’s trading patterns actually reflected the whole-day manipulative strategy it maintained Select Vantage traders were pursuing, even though such evidence should have been available. Manipulation in the form alleged has never previously been observed. There are at least some questions about how rewarding such a strategy would be. The level of coordination required would be considerable, but there was no independent evidence of any such coordination.

364. In those circumstances, approaching the evidence in the general way in which it was approached by the primary judge and by the parties, the finding that Select Vantage was likely engaged in manipulative trading contrary to ss 1041A and/or 1041B on each of the relevant stock/days cannot stand. Although that finding was not in terms a finding of a contravention of the statutory prohibitions, and no question of civil penalty arose in the proceedings, the allegations are extremely serious. Having regard to their gravity and the indirect and ultimately somewhat speculative evidence offered for them, in my view it is not possible affirmatively to conclude that Select Vantage or Merlito were engaged in market manipulation. While I have not addressed every subpart of grounds 13 and 14 individually, had it been necessary to determine those grounds, I would have upheld Mr Schlaepfer’s challenge to the primary judge’s conclusion concerning the truth defence.

Conclusion

369. In any event, as I would uphold the primary judge’s conclusion concerning the defence of qualified privilege at common law, the order I propose (subject to what follows) is that the appeal be dismissed.

370. As to costs, unless either party wishes to be heard, the order I propose is that Mr Schlaepfer pay one third of ASIC’s costs of the appeal and of the proceedings at first instance. Although Mr Schlaepfer has been unsuccessful in the outcome of the appeal, he has been successful on most issues including the defence of truth, which occupied a substantial portion of the proceedings. That success has achieved what was said to be an important outcome of the appeal, namely, the vindication of Mr Schlaepfer’s reputation. Although ASIC has succeeded in establishing the defence of qualified privilege at common law, that is a defence of confession and avoidance. To put the matter another way, Mr Schlaepfer has established in the appeal that he was defamed, but defensibly so. For those reasons, while costs ordinarily follow the event, I do not think he should be liable for the whole of ASIC’s costs.

371. Accordingly, the orders I propose are:

1. that the primary judge’s order as to costs be vacated;

2. that the appeal otherwise be dismissed;

3. that Mr Schlaepfer pay one third of the costs of the proceedings below and of the appeal.

**********

Endnotes

Amendments

01 July 2021 - Corrected typographical error on coversheet

Decision last updated: 01 July 2021