Singapore International Commercial Court · Decided 14 March 2019
B2C2 Ltd v Quoine Pte Ltd
Feedback & support[2019] SGHC(I) 03 · Case no. Singapore International Commercial Court — Suit No 7 of 2017 Simon Thorley IJ 21–23, 26, 29 November 2018; 28 December 2018
✱ What happened, in plain English
There is no plain-English summary of this judgment yet.
Ask about this judgment →Laws this judgment relies on
Grounds of decision
Official text from eLitigation
Introduction
- [1]
[1] This is an action for breach of contract and breach of trust. The Defendant (“Quoine”) is a Singapore-registered company which operates a currency exchange platform (the “Platform”) enabling third parties to trade virtual currencies for other virtual currencies or for fiat currencies such as Singapore or US dollars. The two virtual currencies (sometimes referred to as cryptocurrencies) involved in this action are Bitcoin (“BTC”) and Ethereum (“ETH”).
- [2]
[2] The Plaintiff (“B2C2”) is company registered in England and Wales trading inter alia as an electronic market maker. As an electronic market maker, B2C2 provides liquidity on exchange platforms by actively buying or selling at the prices it quotes for virtual currency pairs, thereby generating trading revenue.
- [3]
[3] In recent years, there has been a significant growth in virtual currencies of which Bitcoin is perhaps the best known. They are not linked to any particular country nor regulated by any central monetary authority. They are traded for other virtual currencies or traditional currencies on computer networks such as the Platform.
- [4]
[4] B2C2’s claim arises out of an incident that occurred on the Platform on 19 April 2017 when, as a result of a series of events which I shall consider in more detail below, seven trades for the sale by the Plaintiff of ETH for BTC at an exchange rate of either 9.99999 or 10 BTC for 1 ETH were effected by the Platform. This was at a rate approximately 250 times the rate of about 0.04 BTC for 1 ETH which had been the previous going rate. The proceeds of the sale (3092.517116 BTC) were automatically credited to B2C2’s account by the platform and the corresponding amount of 309.2518 ETH was automatically debited from that account.
- [5]
[5] When Mr Mario Antonio Gomez Lozada (“Mr Lozada”), the Chief Technology Officer (“CTO”) of Quoine, became aware of the trades the following morning, he considered the exchange rate to be such a highly abnormal deviation from the previously going rate that the trades should be reversed. Accordingly, the seven trades were cancelled by Quoine and the debit and credit transactions were reversed.
- [6]
[6] B2C2 contends that Quoine had no contractual right unilaterally to cancel the trades once the orders had been effected and that its action in so doing was in breach of the version of the Terms and Conditions which regulated the trading relationship between Quoine and B2C2 at the material time. Additionally, B2C2 contends that Quoine holds the proceeds of B2C2’s account on trust for B2C2 and that the unilateral withdrawal of the BTC which had been credited to the account as a result of the trades was in breach of trust.
- [7]
[7] The writ was issued in the Singapore High Court on 18 May 2017, the Statement of Claim was served on that day and the Defence on 16 June 2017. With the agreement of the parties, the action was transferred to the Singapore International Commercial Court on 24 August 2017 and came on for trial before me on 21 November 2018.
Currency trading
- [9]
[9] There are a large number of trading platforms around the world and some of these provide for trading in virtual currencies. At a high level of generality, they operate in the same way. There are two types of traders, market makers and investor-traders. An investor-trader, as its name suggests, is a person who purchases a given commodity with a view to holding it for a period of time in the hope that its value increases so that it can be sold at a profit.
- [10]
[10] Market makers are more regular traders who provide liquidity to a given platform by quoting orders (both buy and sell orders) and seek to generate income by capturing what is known as the bid-ask spread. B2C2 is a market maker. Market makers will trade over a number of platforms (in B2C2’s case at the relevant time it was trading on 15) and do so by using sophisticated algorithmic trading software. At any one time a market maker may be seeking to sell a given currency on one exchange and buy it on another or, indeed, on the same exchange. The software monitors the trader’s position over all the platforms to regulate the exposure to any given currency and there will therefore be times when no buy quotes for a given currency are being placed on any of the platforms because the software has determined that the trader is over-exposed in that currency and is therefore, for the time being, only going to be a seller. The reverse can obviously also be the case.
- [11]
[11] Market makers can on occasions also be investor-traders but this action is concerned with B2C2’s activity as a market maker. Of more relevance is the fact that operators of currency exchange platforms can also trade as market makers on their own platform as Quoine did in the present case.
- [13]
[13] As Mr Maxime Boonen (“Mr Boonen”), a director of B2C2, explained in his affidavit of evidence-in-chief (“AEIC”), B2C2 acted as a market maker on the Platform in the following manner:
The Platform
- [14]
[14] Quoine was cofounded by Mr Lozada together with Mr Mike Kayamori in 2014. It operates platforms in Japan and in Singapore. The Platform functions in the following way:
- [15]
[15] At the relevant time, Quoine offered three types of trading: spot trading (where trades are settled instantly), margin trading (which allows trading with borrowed funds) and futures trading (which involves both the buyer and seller agreeing to sell at a future date at a given price). Quoine no longer offers futures trading.
- [16]
[16] Margin trading is central to this dispute and therefore needs to be explained more fully. A trader may trade on the Platform using borrowed funds, borrowed either from Quoine or from other users who offer their funds to be used for peer-to-peer loans. The assets in the margin trader’s account serve as collateral for the loans. In the event that the collateral in the account falls below a pre-determined percentage of the loan, the Platform’s system takes action by making a margin call and closing out all or part of the trader’s positions to prevent further loss and to seek to avoid a default.
The relationship between Quoine and B2C2
- [19]
[19] B2C2 was set up by Mr Boonen and a Mr Molendini. Mr Boonen was responsible for creating the vast majority of the algorithmic trading software which I accept is the life blood of B2C2’s business and therefore highly confidential. It has been necessary during the course of this action for steps to be taken to seek to ensure the maintenance of that confidentiality whilst also ensuring a fair and open trial. The way in which this was done and the reasons therefor are set out in an earlier judgment in this action: B2C2 Ltd v Quoine Pte Ltd [2018] 4 SLR 67.
- [20]
[20] B2C2 opened an account on the Platform on 28 June 2015. This was done electronically online by Mr Boonen and Mr Boonen was asked to confirm that B2C2 accepted Quoine’s Terms and Conditions which Mr Boonen did.
- [21]
[21] Mr Boonen gave evidence that initially he had some concerns about trading on the Platform but following some reassurance from Mr Lozada in June 2016 and the provision of a line of credit from Quoine to B2C2, B2C2 commenced full trading on the Platform in January 2017. A good deal of written evidence and some cross-examination was directed to the reasons underlying those concerns and how and why they were alleviated but I am satisfied that this has no bearing on the matters in dispute.
The Agreement and the Risk Disclosure Notice
- [22]
[22] The Terms and Conditions (“the Agreement”) which were on the Defendant’s website in June 2015 when B2C2 opened its account were stated to have been “Last updated April 2014”.
- [23]
[23] A great deal turns on the terms of the Agreement and its relationship with other documents that were on the website at the date of the incident. It is therefore included in full as Annex 1 to this judgment. The most important provisions are as follows:
- [24]
[24] Additionally, Quoine asserts that a Risk Disclosure Statement was uploaded onto the website on 22 March 2017. Whilst this was not admitted by B2C2, it was not seriously challenged and I accept the evidence demonstrating that it was. This document figures in one of Quoine’s proposed defences and, again, it is necessary to see the passages relied upon in context so this document forms Annex 2 to this Judgment. I shall revert to it later.
The incident on 19 April 2017
- [26]
[26] As indicated above, Quoine reversed these orders on 20 April 2017, the next day. The background to these trades and the way in which they came to be carried out is complex and has been the subject of both fact evidence and expert reports.
The Witnesses
- [27]
[27] Evidence was given by Mr Boonen for B2C2 and by Mr Lozada and Mr Tseung, a co-founder of Pulsar, for Quoine. Expert evidence was given by Mr Alexis Atkinson (“Mr Atkinson”) on behalf of B2C2 and Mr Vikram Kapoor (“Mr Kapoor”) on behalf of Quoine.
- [28]
[28] Mr Boonen received a Bachelor of Engineering degree from the University of Louvain in Belgium and then a Master’s degree in Financial Economics from the University of Oxford in England. He was first employed by Google where he learned about electronic trading and then worked in the fixed income department of Goldman Sachs in London. He left Goldman Sachs in May 2015 to found B2C2.
- [29]
[29] Mr Boonen was well-informed as to the trading environment in which B2C2 operated and was plainly a highly experienced computer programmer who had successfully designed B2C2’s trading software. In the witness box, he was both reasoned and precise and explained technical matters with clarity. It became apparent in cross-examination that an aspect of his AEIC was overstated but in the event nothing turned on that. I found him to be an honest and straightforward witness who gave reliable evidence, even if, on occasions, his desire to be precise led to overlong answers. His demeanour as a witness was not the subject of any general criticism in Quoine’s closing submissions but it was suggested that his evidence was unbelievable in certain respects. I shall assess the weight that can be placed on aspects of Mr Boonen’s evidence later.
- [30]
[30] Mr Lozada has both an undergraduate and Master’s Degree in Computer Sciences from the University of Kansas, USA. Initially he was employed in Tokyo by Fusions Systems Japan which provided financial software for banks. He then spent 10 years with Merrill Lynch ending up as CTO for the fixed income business in Japan. He then moved to Credit Suisse in 2009 or 2010 where he was both CTO for the fixed income business and Chief Information Officer for the Japan business. He left Credit Suisse in 2013 to cofound Quoine. Mr Lozada was well informed as to Quoine’s trading environment and was an experienced computer programmer having designed the Platform’s system.
- [31]
[31] In B2C2’s closing submissions it was submitted that Mr Lozada was very unconvincing and far from forthcoming so that his evidence should be viewed with extreme circumspection. A number of aspects of his evidence were brought to my attention to support this assertion. These related in part to evidence which displayed an element of uncertainty in his overall evidence and in part to evidence given at trial which was inconsistent with earlier evidence given on the application for summary judgment. I have considered carefully each of these assertions as, at the time he gave his evidence, I perceived him to be an honest and straightforward witness. I consider that, at best, it could be said that Mr Lozada did not investigate the circumstances surrounding the incident of 19 April in sufficient detail before giving the evidence he did on the summary judgment application. But I am satisfied that this was not done with any intention to mislead the Court. The facts are complex and it is not surprising that they became clearer in the course of time. Equally where there were passages in his evidence which constituted assertions as to the intentions of B2C2, these were, I consider, honestly held beliefs. The areas of uncertainty in his evidence were due to imperfect recollection rather than any intention to deceive. All in all, I see no reason to alter my original perception that he was an honest and straightforward witness who was particularly frank in identifying the areas where the Quoine system had proved to be inadequate and the steps that he had taken to rectify this.
- [32]
[32] The final witness of fact was Mr Tseung of Pulsar. Once certain passages of his evidence were ruled to be inadmissible, the Plaintiff no longer sought to cross-examine him. Notwithstanding this, in its written closing, B2C2 asserted that Mr Tseung’s evidence was self-serving. But this was a criticism directed to the evidence which had been struck out and to Mr Tseung’s statements relating to the validity of the seven trades, which is a matter for me and not him.
- [34]
[34] Mr Kapoor has a BSc in Engineering from the University of Pune, India, and an MBA from the State University of New York at Buffalo, USA. He is a senior managing director at Ankura in New York specialising in quantitative finance valuation and data analytics. His work includes analyses of algorithmic trading and trade routing by financial institutions. He has given expert evidence on a number of occasions both in the US courts and in arbitrations.
- [35]
[35] Both Mr Atkinson and Mr Kapoor are well qualified to give expert evidence on the defined issues. Both spent a considerable amount of time reviewing B2C2’s software and presented their reports on the issues in a focussed and helpful way. They were, however, very different witnesses. Mr Atkinson was significantly more diffident and less articulate than Mr Kapoor. He was more hesitant in giving his answers compared to the more polished approach of Mr Kapoor.
- [37]
[37] Secondly I was invited to place adverse weight on Mr Atkinson’s evidence because he had failed to sign a copy of a confidentiality undertaking supplied to him by B2C2. Plainly he should have done that and B2C2 should have chased him for a signature. But I do not see how this failure can be said to impact upon the evidence that he did give.
- [38]
[38] Finally, in the course of trial, when Mr Atkinson was in the witness box, he sought to introduce two further documents prepared by him. This was improper but understandable as he had had to deal with Mr Kapoor’s report in a short period of time. He explained that he was seeking to supplement his second report with these documents. I refused to allow one of them to be adduced in evidence but allowed the other, on the basis that, if necessary, Mr Atkinson could be recalled to be cross-examined on matters to which this document related later in the proceedings. In the event no application was made for Mr Atkinson to be recalled. I consider that this incident was but part of the normal vicissitudes of litigation and does not reflect adversely on Mr Atkinson’s evidence.
- [39]
[39] B2C2 levelled a number of criticisms at Mr Kapoor. Mr Kapoor was responding to Mr Atkinson’s first report which, in very simple terms, supported the assertion that B2C2’s software had been designed for wholly defensive purposes. It was plain that Mr Kapoor had been instructed to focus particularly on aspects of the software that might serve to identify the mindset of the writer as not being limited to defensive purposes. This was, as will become plain, a fundamental part of Quoine’s case. He cannot be criticised for doing this nor for highlighting aspects of the software which appeared to be indicative to him of an intention on B2C2’s part to program its software to take advantage of errors of others and, in the extreme, to manipulate the market. He raised technical matters which it was necessary for B2C2 to respond to and which needed to be resolved.
- [40]
[40] However, in two respects he strayed beyond permissible limits. First, he sought to draw his own conclusions as to the intentions of the writer of the software which is a matter for me and not for him and formed part of the evidence which I was asked to strike out. Conclusory assertions of this sort are not unusual in expert reports, particularly from experts more familiar in giving evidence in US courts and tribunals. It is no doubt what he was instructed to do and doing so does not add to or subtract from the technical evidence that he gave on the issues he was asked to address.
- [41]
[41] Secondly, he gave evidence on matters that were not within the scope of the issues on which expert evidence was to be adduced. Again, this was the subject of the application to strike out and I do not doubt that this is what he was asked to do. There is, to my mind, nothing untoward in this and it does not detract from the technical evidence that Mr Kapoor gave on the relevant issues.
- [42]
[42] All in all, I am satisfied that both experts were qualified to give the technical evidence that they did. They did so in support of their respective client’s cases but in an informed and coherent manner. At all times they were doing their best to assist the court and I am grateful to them for their patience in approaching the issues and in amplifying their reasoning in their oral evidence. It will be for me to assess the weight that I place on each expert’s evidence on these technical matters but I do not accept that either witness’ evidence should be discounted on the basis of the matters considered above.
Inadmissible evidence
- [43]
[43] Prior to trial, both parties served notices identifying passages in the AEICs and expert reports of the other party which were contended to be inadmissible. Sensibly, counsel agreed that the question of admissibility should not be considered before the witnesses gave evidence but should be left to closing submissions.
- [44]
[44] However, in advance of the trial, B2C2 did issue a summons to strike out certain matters arising in the affidavits of Mr Lozada and Mr Tseung and in Mr Kapoor’s expert report. This was initially heard in chambers but by agreement of the parties was adjourned into court to be heard at a convenient moment during the trial.
- [45]
[45] The Court directed that so far as the application sought to strike out parts of the affidavits of the witnesses of fact, this should be determined at the outset of the trial with the objections to Mr Kapoor’s report being determined at a later time.
- [46]
[46] I therefore heard argument on the fact witnesses. I indicated at the conclusion of that argument that three paragraphs would be struck out and that I would give my reasons for so doing as part of the Judgment after trial. These are my reasons.
- [47]
[47] The paragraphs in question were paragraph 28 of Mr Lozada’s first AEIC and paragraphs 9 and 10 of Mr Tseung’s AEIC. These state:
- [48]
[48] It is convenient to deal with paragraph 28 of Mr Lozada’s first AEIC and paragraph 9 of Mr Tseung’s AEIC together.
- [49]
[49] It will be seen that these paragraphs relate to an alleged common practice for trading exchanges to cancel clearly erroneous transactions. Mr Danny Ong for B2C2 contended that an investigation into how other exchanges approached the question of dealing with erroneous transactions was irrelevant to the issues arising on the pleadings in this action. He drew my attention to the fact that there is no express term in the Agreement providing for such cancellation and that the plea of an implied term to this effect contained in paragraph 19(d) of the Defence (Amendment No 4) (“the Defence”) was solely on the basis of giving business efficacy to the agreement and to give effect to the intentions of the parties. This did not, he submitted, provide for the implication of such a term on the basis that there was a common practice in the trade for this to be done. Further, he claimed that in any event the evidence constituted impermissible opinion evidence of what the witnesses perceived to be common practice. If the evidence was allowed to stand, it would be necessary for B2C2 to adduce further evidence to rebut the contention.
- [50]
[50] Mr Paul Ong for Quoine submitted, first, that the correct course was to deal with the question of admissibility by way of closing submissions as was being done in relation to the other disputed evidence. Secondly, whilst he accepted that there was no plea of common practice, he said that the evidence was relevant to demonstrate that this was a term to which the parties would have agreed had it been posed by the notional hypothetical onlooker at the time the Agreement was made. He said that if this was the policy of other trading platforms this would be an indication that it was more likely that the parties would have agreed. Thirdly, he contended that this was not impermissible opinion evidence.
- [51]
[51] Whilst I accept the third of these submissions, the agreements that are said to evidence such a common practice contain an express term to that effect; there is therefore no need for any implied term. There is no such express term in the Agreement. The pleading contends that one should be implied to give business efficacy to the Agreement. The mere fact that the parties might have agreed such a term had it been proposed is, of itself, insufficient for the term to be implied. If the needs of business efficacy do not require the implication of a term, it will not be implied even if, had it been proposed, it might have been accepted. I am therefore satisfied that the evidence contained in paragraph 28 of Mr Lozada’s first AEIC and paragraph 9 of Mr Tseung’s AEIC is irrelevant to any pleaded issue in this case.
- [52]
[52] Whilst there is often merit in adopting the course of leaving questions of admissibility to be discussed during closing submissions, equally there is merit in a clear case of striking out inadmissible evidence at an early stage if this is going to avoid the possibility of further evidence and/or cross-examination on the irrelevant material. I consider this to be a clear case.
- [53]
[53] Paragraph 10 of Mr Tseung’s AEIC falls into a different category. It is plainly speculation on the part of Mr Tseung as to why B2C2 programmed its trading systems in the way it did. Speculation by a witness of fact as to why another entity might have done something is of no assistance to the Court and time should not be spent at trial cross-examining Mr Tseung on the basis for his speculation.
- [55]
[55] At a later stage in the trial before the expert witnesses were to be cross-examined, I heard submissions on the admissibility of the passages in Mr Kapoor’s expert report to which objection was taken. There was an earlier hearing in this case on 20 February 2018 at which the question of expert evidence was discussed. I made it clear that in a case of this nature, expert evidence would only be allowed in relation to specific issues which were to be determined by the parties and approved by the Court. The parties proposed a number of possible avenues of expert evidence. In particular, the Defendant proposed that there should be expert evidence on market manipulation. Having heard submissions, I directed that expert reports should be directed to (and limited to) certain specific issues concerned with the working of the B2C2’s trading algorithm and that if the parties wished at a later date to adduce expert evidence on different issues, they should make an application for permission in advance of doing so.
- [56]
[56] The specific issues were drafted by the parties and settled by the Court. They were as follows:
- [57]
[57] It was directed that the expert reports should be prepared sequentially which resulted in the first report of Mr Atkinson being sworn on 28 August 2018 and Mr Kapoor’s in reply being sworn on 5 November 2018. Mr Atkinson’s report addresses the issues but Mr Kapoor’s report went somewhat further and sought to give expert evidence on two other matters. First, he concluded on the basis of his review of the B2C2’s computer programs that the “deep quotes” set by the algorithm (ie, the price of 10 BTC to 1 ETH) were not merely a defensive measure (of which more below) but constituted what are known as “stub quotes”, a practice which Mr Kapoor opined constituted impermissible market manipulation which was contrary to some regulatory requirements. Secondly, he reviewed the Platform’s software and the Quoter Program and gave evidence on how the Platform calculated the position of margin traders and expressed the view that if the Quoter Program had been working properly there would not have been a margin call.
- [58]
[58] As regards the first aspect (the stub quotes issue), Mr Danny Ong contended that the evidence should be struck out primarily on the basis that it was not directed to the questions that the experts had been asked to address and that no application had been made for different issues to be the subject of expert evidence. Secondly, he said that Mr Kapoor was not qualified to give evidence on what was essentially a regulatory matter. Thirdly, the evidence went to an unpleaded issue of stub quotes. Lastly, the issue of market manipulation was a matter for the Court and not for a witness. He indicated that if the evidence was admitted it would be necessary for the Plaintiff to instruct a witness to respond to Mr Kapoor’s assertion.
- [59]
[59] Mr Paul Ong contended that Mr Kapoor’s qualifications (or lack of them) were not a ground for striking out the evidence. His degree of competence to give the evidence went to weight, not to admissibility. To a certain extent I agree with this unless it is apparent from the witness’ CV that he has no relevant qualifications. Had this been the only objection I would have allowed him to be cross-examined on his credentials and then ruled on admissibility.
- [60]
[60] He went on to submit that the evidence was relevant to understanding the nature of the strategy underlying the trading software, to demonstrate that it was not merely a defensive measure and had, as he put it, the more sinister objective of market manipulation. This had been pleaded in general terms and, in particular, was relevant to the issue of unconscionability under unilateral mistake in equity.
- [61]
[61] On the second aspect (the Platform and Quoter Program issue), the primary argument related to the fact that no issue concerning those programs had been included in the list of issues and that the code for the programs had not been disclosed or considered by the Plaintiff. It was therefore wrong at this late stage for the issue to be canvassed any further than had been done in the AEICs of the witnesses of fact.
- [63]
[63] However, even if that was not the case, I concluded that it would be quite wrong to admit expert evidence on a new issue at this late stage. As I made clear at the earlier hearing, I consider that, in cases such as this, the expert evidence ought to be directed and only directed to specific defined issues. This enables the reports to be filed without unforeseen expert evidence being adduced which is likely to lead to delay and expense in dealing with that evidence in subsequent reports.
- [64]
[64] This is precisely what would happen in this case. Mr Paul Ong suggested that the stub quote issue could be hived off, B2C2 could adduce expert evidence in reply to Mr Kapoor and the matter could then be addressed at a subsequent hearing. I accept that Mr Kapoor had to prepare his report under certain time pressures and that his report contains material which he relies on as justifying the stub quote assertion, but this does not excuse the fact that Quoine could have sought leave to adduce expert evidence on the issue of market manipulation in advance of the service of his report.
- [66]
[66] On the second issue, again I was satisfied that justice could be done without going into detail on Quoine’s programs. There is no dispute that the Platform did determine that the Counterparties were in breach of their respective margin limits nor is there a dispute as to when this happened. The reason why this happened was explained by Mr Lozada and it is not necessary to go into any greater detail as to the steps that the code goes through to reach this conclusion.
- [67]
[67] Equally, there is no dispute that the Quoter Program was inoperative after 13 April. Whether or not the consequent lack of liquidity would have occurred had it been operative is irrelevant.
The facts
- [69]
[69] I propose to consider the facts under the following headings:
(a) The Quoter Program;
(b) The Platform and the Margin Calls;
(c) The B2C2 Trading Software; and
(d) The mindset of Mr Boonen.
- [70]
[70] Many of the facts are either common ground or not in dispute. In so far as any matter was the subject of dispute, I have reached an assessment on the basis of the balance of probabilities.
The Quoter Program
- [71]
[71] The causes of the incident on 19 April 2017 have their origin in some work done by Quoine on 13 April when changes were made to some login passwords for several critical systems on the Platform. This was done for security reasons. Due to an oversight, certain necessary changes to the Platform’s Quoter Program were not implemented.
- [72]
[72] The Quoter Program is responsible for retrieving external market prices from other exchanges which are then used to create new orders to be placed by Quoine on the Platform for market making purposes and to create liquidity. The Quoter Program is personal to Quoine and information generated by it is not available to other users of the Platform. The effect of the oversight was that it could not access data from those other exchanges and accordingly it became inoperative and stopped creating new ETH/BTC orders on the Platform. Unfortunately, this did not generate what is known as an exception message which would have alerted Quoine to the oversight as the notification flag for such messages had been disabled. Quoine did not become aware of the oversight until after the events of 19 April.
The Platform and the Margin Calls
- [74]
[74] So far as concerns the Platform, it had the effect of causing the mechanism for calculating the margin traders’ positions to detect that Pulsar’s and Mr Tomita’s accounts were in what is known as “Margin Sell-out Position” which served to trigger the margin calls resulting in the placement of the market orders to buy ETH at the best available market price. This occurred, in the case of Pulsar at 11.31pm and in the case of Mr Tomita at 11.53pm. This can be seen on a print out of the force-closed orders executed on 19 April 2017 (trader ID 5539 is Pulsar and trader ID 30742 is Mr Tomita). A copy of the print out forms Annex 3.
- [75]
[75] As Mr Lozada made clear in his evidence, the force-closed transactions were carried out notwithstanding that the available BTC balances in Pulsar and Tomita’s accounts were insufficient to meet B2C2’s orders at the limit price quoted by B2C2. This would not have happened if the Platform had a program to check whether there were sufficient available assets. There was however no such program at the time which Mr Lozada accepted in the course of cross-examination was the result of poor design:
- [76]
[76] Mr Tseung, who in the event was not cross-examined on his AEIC, gave evidence that at the relevant time the balance on Pulsar’s account on 19 April contained only 13.5256945100 BTC yet the Platform debited Pulsar’s account with over 3000 BTC. He stated as follows:
- [77]
[77] This evidence reflects the admission by Mr Lozada that the Platform should have had such a feature but did not. Had there been such a feature, the Platform would have traded only the 13.5256945100 BTC in Pulsar’s account, leaving any outstanding deficit on Pulsar’s account due to the lender, in this case Quoine, to be recovered outwith the operation of the Platform.
- [79]
[79] By 11.31pm, all the pre-existing orders by Quoine had been used up and the system had to search for the next lowest price available. As can be seen, these ranged from 0.05 to 1 BTC/ETH from traders other than Quoine including one from Pulsar itself at 0.07. It is thus apparent that the loss of liquidity was causing the trading systems of other traders to react by placing higher quotes, 1 BTC/ETH being some 25 times the 0.04 level. Exactly why this was so is unclear.
- [80]
[80] Then at 11.52pm, orders at 10 and 9.99999 from Trader 5776 (B2C2) were placed on the Platform and, since there were no lower priced orders, these were filled to complete the margin call on Pulsar. The position with regard to Mr Tomita is a little more complicated but it is not suggested that anything turns on this.
The B2C2 Trading Software
- [82]
[82] The B2C2 trading software (“the Software”) is largely deterministic; it produces the exact same output when provided with the same input. Mr Atkinson explained that this was a benefit as it makes testing easier and the behaviour of the system is easier to understand, reason about and validate. It was devised almost exclusively by Mr Boonen and, as might be expected, had been the subject of a number of changes during the course of time. When the software was originally written, Mr Boonen included a number of routines designed both to maintain the integrity of the system and to ensure that it continued to operate in circumstances when there was insufficient liquidity in a given market for the system to operate normally.
- [84]
[84] Mr Atkinson described this in paragraph 54 of his first report:
- [85]
[85] These prices have to be chosen in a manner such that on the Ask side it is sufficiently large that the prospect of a trade at that level is both unlikely and, if it occurs, will adequately protect B2C2 against the risk of any adverse consequences of such a trade. Conversely, the chosen price on the Bid side needs to be low. As at 19 April 2017, these prices were, respectively, for the Ask side 10 BTC/ETH and for the Bid side 0.00001 BTC/ETH. As Mr Boonen put it, “the prices had to be sufficiently advantageous so that B2C2 would not regret it if orders place at or near those price levels were filled”. In cross-examination, he said that it was his “role as a trader to try to find numbers that make sense in a situation of very [sic] uncertainty as to the future”.
- [86]
[86] He also indicated that the choice of price had to be at a level that was not so low or high that the trading platform might be programmed to reject it and thus act as an undesirable circuit breaker.
- [88]
[88] Mr Atkinson described the placing of these deep prices as being a defensive programming technique. Mr Boonen identified this as being the role of a programmer to think of potential, even if unlikely events, and design the code to cater for the worst case scenario that could occur . Mr Kapoor accepted that it was in part defensive but said there were aspects that were not necessary for defensive purposes and Mr Paul Ong relied on these aspects as indicating an adverse motive on the part of Mr Boonen to take advantage of a situation of low or no liquidity by triggering margin calls.
- [90]
[90] Mr Boonen accepted that this could have been done and that it had not been. He however drew attention to the fact that there was a provision whereby the code does not send out orders that exceed a certain USD amount. In cross-examination, he said:
- [93]
[93] The second factor Mr Kapoor drew attention to was that there had been changes in the deep prices that had been set in the software since it was first written. When Mr Boonen initially wrote the program, he configured the software to quote a deep price on the Ask side at a BTC/ETH ratio which was close to infinity and subsequently altered this to a price which was equivalent to a ratio of 50 BTC/ETHJHH. On 16 June 2016 Mr Boonen altered the deep price to the 10 BTC/ETH level. Mr Kapoor suggested that this change was indicative of an intention on the part of the programmer to increase the likelihood that such deep prices would be executed. The basis for this was Mr Kapoor’s assertion that higher prices were likely to be rejected by the Platform whereas rejection at the 10:1 level was less likely.
- [94]
[94] Mr Boonen explained the reasons behind this change in his third AEIC and was then cross-examined at some length on it. Mr Boonen explained that in June 2016 there was what he called a refactoring of the code with the intention of making it clearer and safer. Whilst the close to infinity level had previously been altered to place another somewhat smaller but still large number for the deep price, both were a “one size fits all” number in that it applied across the board to all currency pairs traded by B2C2. It so happened that at that date the effect was that the BTC/ETH rate was around 50:1.
- [95]
[95] By 2016, B2C2 were trading in an increased number of trading pairs including ETH/BTC (ETH having come into existence after the B2C2 trading software had first been written in 2014). Mr Boonen particularly drew attention to the fact that B2C2 had by then started trading in BTC/South African Rand, a significantly more volatile currency than the USD and that he had concerns that the generic deep price might not be deep enough to protect B2C2 in the event of a significant change in the value of the Rand.
- [96]
[96] He therefore concluded that the “one size fits all” number was no longer appropriate and he placed individual deep prices for each currency pair. Mr Boonen considered the 50BTC:1ETH ratio was too high and he selected the 10BTC:1ETH level. He accepted that one of the reasons for doing this was to ensure that any sell orders placed at that level would not be rejected by the Platform. He gave evidence that the price was chosen so as to be sufficiently advantageous to cover the risks to B2C2 of trading in a potentially illiquid market and again emphasised that he regarded ETH as being the greater risk at the time. He also explained that he used the same deep price to act as a circuit breaker to protect either against the risk of a significant change in the actual market price of a given currency pair or against the possibility that there might be bugs in other parts of the B2C2 trading system. He asserted that it was his role as a trader to find numbers which made sense to him in an uncertain market.
- [97]
[97] In re-examination, he amplified upon the circuit breaker aspect by giving an example of a case where with a different currency pair the actual market prices broke through the upper range and the deep price acted as a circuit breaker. He indicated that he felt that the 50:1 ratio was so high that it would be highly unlikely ever to be activated as a circuit breaker and thus would not adequately protect B2C2 and that he also chose the 10:1 ratio with this in mind. I accept this evidence.
- [98]
[98] The third factor to which Mr Kapoor drew particular attention was his assertion that certain portions of B2C2’s code were changed after the glitch occurred on 13 April 2017 which he suggested might have caused the software to quote more frequently or to quote more in total than it was quoting before the glitch. The underlying thrust behind this suggestion was that B2C2 had appreciated sometime between 13 and 19 April that something was causing illiquidity on the Platform and that the changes were implemented with a view to triggering force-closures.
- [99]
[99] As Mr Kapoor fairly acknowledges, he was unable to review the actual code for the suggested change, but he surmises that it did occur since the initial cycle frequency was set at 10 seconds and by 19 April the frequency was 5 seconds.
- [100]
[100] In his third AEIC, Mr Boonen rejected the implicit suggestion that any change to B2C2’s code configuration files was in response to or motivated by the glitch which he contended did not come to his notice until after the orders were fulfilled on 19 April. He went on to explain that the amendments made on 13 April were part of a routine change to the Inventory Strategy which adjusts prices to maintain a target balance of B2C2’s assets on the Platform.
- [101]
[101] Mr Atkinson also considered this suggestion in detail in his second report. He concluded that there was no evidence to suggest that there were any configuration changes made in April other than those which reflected configuration changes in inventory on the Platform. All the witnesses were cross-examined on this issue.
- [103]
[103] Mr Atkinson and Mr Kapoor metaphorically crossed swords on the technical aspects of this issue. Since I have accepted that Mr Boonen did not notice the drop in the volume of orders on 13 April, it is less important that I consider the technical aspects in detail. In brief, Mr Atkinson contended that Mr Kapoor’s analysis was flawed in that it sought to determine an undefined average cycle frequency but failed to distinguish between Bids and Asks and failed to consider the mode average. He pointed to the fact that that there was a large difference between the mean and the mode average which suggested that the approach of Mr Kapoor was not appropriate. He concluded that the data available to him, properly analysed, showed that the cycle frequency was set to 5 milliseconds before 13 April and remained the same through until 19 April.
- [104]
[104] In cross-examination, Mr Kapoor accepted that in the absence of full details of the changes implemented on 13 April he could not be certain that changes had been made which would reflect an intention to take advantage of the drop in orders but maintained that this was a possibility.
The mindset of Mr Boonen
- [106]
[106] For reasons which I shall amplify upon when considering the law of mistake, I have concluded that when the law is faced with a contention that a contract made by and between two computer systems acting as programmed but otherwise without human intervention is void or voidable for mistake, it is necessary to have regard to the mindset of the programmer when the relevant programs, or the relevant part of those programs, were written. The knowledge of the programmer in question is to be inferred by the Court from the evidence adduced and from all the surrounding circumstances.
- [107]
[107] In the case of the seven contracts made on 19 April 2017, the offer for sale at the 10 BTC/ETH price was made by the B2C2 trading system. It was accepted by the Platform in filling the market orders already placed on the Platform by the action of a separate program which had identified that the Counterparties’ collateral was insufficient. The Counterparties in their loan arrangement with Quoine had agreed that this could be done if the Platform determined that the collateral was insufficient. There was no human intervention in this process. Mr Boonen, Mr Lozada and the Counterparties were unaware what had happened until after the event.
- [109]
[109] The second aspect is pleaded in Particular (f) which reads as follows:
- [112]
[112] In his written evidence, Mr Boonen said that he was not aware that the Platform was affected by any glitch, that the trading software was not programmed to detect such glitches and that the orders were placed because of the deep pricing strategy referred to in [83] above. He accepted that there was always a possibility that such orders would be filled and that, if they were, B2C2 would be bound by them. It was for this reason that he pre-programmed the code so that the prices were sufficiently advantageous to cover the risk to B2C2 of uncertain liquidity and market conditions. He frankly stated that “the price levels had to be sufficiently advantageous so that B2C2 would not regret it if orders placed at or near those levels were filled”.
- [113]
[113] Early in his cross-examination, when considering the extent to which the deep prices were present to cover the risks of illiquidity Mr Boonen said this:
In its written closing, Quoine repeatedly referred to this passage as being an indication that Mr Boonen must have known that if the very unusual or unfathomable were to occur this was likely to be the result of a serious error or glitch.
- [114]
[114] Thereafter, Mr Boonen gave evidence that he was surprised when he first saw that the orders had been filled and that he considered the possibility that there was an error on the Platform or an error in the B2C2 system. Having checked that there were no apparent errors in either system he thought “Wow, there must have been something incredible happened [sic] in the market overnight”.
- [116]
[116] Matters came to a head in this passage of cross-examination:
- [117]
[117] Mr Boonen was affronted at the suggestion that he had designed his software with the specific purpose of taking advantage of the errors of others and, in the result, he did become over-loquacious but the thrust of his answers was always to the effect that Quoine were seeking to read far too much into the insertion of the deep prices than was warranted. His position can be illustrated by two further extracts of the oral evidence he gave on the second day of his trial:
- [118]
[118] What then was his mindset? At the time the original program was written, ETH did not exist. In June 2016 when the 10:1 ratio was added, it was a fledgling currency and I accept Mr Boonen’s evidence that he regarded it as representing the most risk. I also accept that his primary concern when writing the program was to protect the integrity of the B2C2 trading system so as to minimise the risk of any unwarranted exposure.
- [119]
[119] He was particularly careful because of the fact that the markets in virtual currencies were unregulated. He saw the need to protect B2C2 and anticipated that other traders and the platform operators would take similar steps. He appreciated that the way he had written the software, the deep prices could go on the Platform and that it was possible, although unlikely, that a trade would be executed at that level. It would only be, as he put it in the passage quoted above, when something very unusual happened or when the unfathomable actually crystallised. He frankly conceded that were this to happen he would hope to make a profit and that the price chosen was selected so as to maximise the possibility of not making a loss. His strategy was therefore not wholly defensive.
- [122]
[122] He might have appreciated that one contributing factor to illiquidity could be due to an oversight or error on the part of someone but there are no grounds for concluding that he considered that these would be the only circumstances that would lead to illiquidity. His attitude was well demonstrated by his reaction on seeing the outcome of the trades, when his first concern was that there might have been an error in the B2C2 code and then that there might have been an error in the Platform.
- [124]
[124] The evidence does not support the conclusion that when designing the code Mr Boonen perceived that no trader would ever agree to buy ETH at such a price unless he had made a mistake and included the deep prices with this in mind. More specifically, there are no grounds for concluding that he ever turned his mind to the relationship between the margin traders and those who had loaned them money, far less that he programmed the code with a view to exploiting that relationship by causing the deep prices to lead to a margin call and hence a force-close position. Equally there are no grounds for concluding that Mr Boonen wrote the code in the way that he did with an underlying intention of manipulating the market. I therefore reject Quoine’s assertions in those respects.
- [125]
[125] His mindset therefore, in summary, was that he intended the 10:1 ratio to act primarily to protect B2C2 against the risks of illiquidity but that he was aware of the unlikely possibility that trades might be executed at that level. He did not turn his mind in any detail to the circumstances that might lead to such trades being executed but he would have been aware that one possibility was that it could be the result, wholly or in part, of some error or omission on the part of someone, including himself. He intended that B2C2 should be best placed to make a profit, not a loss, if the unlikely became a reality, whatever the cause. In this respect his programming was opportunistic but it is in no respect sinister. It is not sinister to give yourself the opportunity of making a profit rather than a loss in the case of a wholly unlikely event occurring for whatever reason.
The Contractual Relationships
- [126]
[126] There is more than one contractual relationship that will exist when parties trade on a currency platform. First, all traders, whether buyers and sellers, will have a contract with the platform operator to regulate the relationship between that trader and the platform owner so that they can trade on the platform (“Platform contracts”). It is the Platform contract between B2C2 and Quoine that Quoine is alleged to have breached when it reversed the trades. Secondly, where there are margin traders, there will be a separate contract between the borrower and the lender (in this case Quoine on the one hand and the Counterparties on the other) which will regulate their lending relationship (“Margin contracts”). Finally, when a trade is executed, the buyer and seller will have some form of contractual relationship inter se (“Trading contracts”).
- [128]
[128] Mr Danny Ong for B2C2 submitted that the relationship was, as he put it, a spider’s web of contracts with Quoine in the middle. Quoine was thus a party to a contract with the buyer and a separate contract with the seller, acting as an intermediary. He contended that Mr Boonen had accepted in evidence that B2C2 always considered that the contractual counterparty to trade on the Platform was Quoine and that Quoine must necessarily act as a central counterparty to both customers since the Platform operated on an anonymous basis.
- [129]
[129] It is relevant to decide this issue as it will be necessary to identify what the mistake was, if there was one, and who the mistaken party was. It should, however, be noted that the answer to this question has no bearing on whether or not Quoine is liable to B2C2 for breach of the Platform contract.
And the third paragraph reads:
The judgment continues past this page. Read the rest at eLitigation.