Slave to the Algorithm: Burford and the Importance of Maintaining Confidence in a Broken System

On 1 October 2019 I wrote an article about Burford in which I said that their application to obtain, from the LSE, details of how trades in their shares were placed and then apparently cancelled last August was a rare opportunity to discover whether market manipulation is endemic on the LSE. It is important to highlight that this case is not about Burford, which has become a divisive company. It is nothing to do with whether the claims made by Muddy Waters in respect of Burford are true. It was about an opportunity to rebuild trust in the stock market.

In April, the court case took place. ShareSoc wrote a letter in support of the application. We did this for the simple reason that in our view, many investors do not believe that identifying and prosecuting market manipulation is a priority for either the FCA or the LSE.

It was therefore hugely disappointing that Burford’s court application failed. The reasons for the failure were numerous. One of the more significant ones was that Burford had suffered no obvious loss: while shareholders may have done, the company itself had not directly suffered any financial loss.

The real loss was one of faith in the system. For those who noticed the trading frenzy as Muddy Waters announced one afternoon that it was short an “arguably insolvent” company whose identity would be revealed the next day, who saw the stream of tweets proving that the identity of the stock was widely known by third parties in advance of publication, and then watched the share price of Burford lose 30%, gain most of it back, and then drop 75% as Muddy Waters published their report, it was hard not to conclude that something wasn’t right. For those who were not aware of it already, the rules of the market certainly changed that day.

The court’s key finding was that Burford had failed to show that there was a good arguable case that the trading on the days in question was market manipulation. Burford had argued that, prima facie, it appeared that extensive spoofing and layering (effectively, the rapid placing and cancellation of multiple orders never intended to be executed) had taken place. They were seeking access to the trade data that would reveal whether those orders were coming from the same place.

But the court, relying heavily on evidence provided by the LSE, found that such orders could be the product of legitimate activity. The judgment noted that Burford’s case was “speculative”:

“Burford’s and Prof Mitts’ inability to see fully what went on, because the data available to them is incomplete and in general anonymised, may mean they feel they cannot independently rule out all possibility of market manipulation, i.e. independently of what they are told by the Stock Exchange and/or the FCA” (para 123 of judgment)

As long as you trust the LSE and FCA to fully investigate market manipulation, this seems, at first sight, a satisfactory outcome. But there is a problem. Anyone who was watching closely could see that what happened to Burford – at that time one of the biggest companies on AIM – was anything but usual. If the FCA and LSE ruled that there was no market manipulation, what exactly did happen?

Anonymity is more important than transparency

Fortunately, the judge’s reasoning sheds more light on the matter.

“… it is quite typical for legitimate execution algorithms (programmed to execute orders) to be set so as … not to “cross the spread” and aggressively execute against an outstanding buy order. This dynamic … is often confused with the appearance of spoofing by those without an operational knowledge of trade execution” (para 98)

Or in other words, high frequency trading algorithms will generate vast numbers of trades, cancelling and varying them all within milliseconds, looking to all intents and purposes like illegal market manipulation, but actually, being totally “legitimate”. One might wonder what “legitimate” means in that context.

And then later,

It is a significant and important feature of the AIM market, as operated by the Stock Exchange and regulated by the FCA, that generally trading activity is anonymous. Providing Participant Identity details in this case to a third party other than as part of the regulatory regime would be a serious invasion into the confidential and commercially sensitive trading activities and strategies of, it could be, a large number of entirely innocent parties. (para 195)

An ordinary investor may wonder what strategies could be revealed by releasing the trading data in relation to a single company on two days. It appears that there is a genuine fear that “algorithmic” trading companies would have their secrets exposed by this information.

To those of us who believe that investing in the stock market should be a matter of researching a company and buying the ones that you believe are likely to increase in value and selling those which you believe will decline, this revelation may come as a surprise. It is clear from these comments that a major, if not chief, part of the LSE’s function is to enable computers to place and cancel countless trades every millisecond.

This may look like market manipulation, but the LSE, who have a commercial interest in maintaining the income stream provided by high frequency, algorithmic trading, say it is “legitimate”. Indeed, it is a noticeable feature of the judgment that the judge appeared unconcerned by the clear conflict that the LSE has in being a commercial business seeking to maximise revenues with its role in identifying market manipulation. A cynic might say that the difference between “legitimate” and “illegitimate” market manipulation depends on the level of fees you pay to the gatekeeper.

In 2010, high frequency trading (HFT) was estimated to make up over half of all trades placed on global stock markets. The figure is likely to be materially higher now. They are the LSE’s key clients: the average time period a share was held for in 2018 was below 8 seconds. No wonder the LSE is keen to ensure that their right to secrecy trumps our right to have confidence that the markets are fair. And the judge agrees. So there, the matter ends.

My friend Michael Robert Taylor, who writes for the Investment Chronicle and blogs under the name @shiftingshares, referred me to a recent book, Flash Crash, by Liam Vaughan. Flash Crash tells the story of Navinder Singh Sarao, the trader who caused a 1,000 point flash crash in the US and was tried there earlier this year. The book makes clear that Singh only got into the murky world of HFT after complaining many times to regulators about HFT practices.

Michael has long been frustrated by high frequency trading. In his experience, he commonly places trades which get “jumped” by others. In other words, algorithmic traders are often able to see what trade you have placed, then interpose their own trades before yours is executed. As a result, the price you thought you were getting is suddenly withdrawn. The reality of this is shocking to those who believe the market is fair. It looks very much like front-running to many.

A large proportion of serious investors – I hesitate to say all, but I have never found one who didn’t – believe that the FCA and LSE are happy to let this practice continue without scrutiny. It goes back to the question of what is “legitimate ”. The FCA and LSE are of the view that it is legitimate for anonymous people to see the orders that genuine investors are seeking to place in the market, and then to overtake those investors in order to buy or sell shares before those orders are executed, forcing bona fide investors to deal on worse terms and allowing the algorithmic traders to pocket the difference. Most of us would regard such activity as obviously both wrong and undesirable. But if you do it enough, and pay enough in execution fees, it not only becomes legitimate, it becomes the main function of the market.

Of course, there is the counterargument that this multibillion pound a year industry run entirely for the benefit of anonymous entities is nothing but beneficial for the markets, providing liquidity and helping long term investors. When you make such profits you can get anyone to write anything, but anyone who has read Michael Lewis’s masterly Flash Boys may take some convincing that HFT and algorithmic trading are anything other than cheating.

The Sharesoc and UKSA Letter

The judge was not impressed by the letter from Sharesoc. The letter referred to the following Twitter poll I carried out in January:

“If you saw some suspicious trading activity take place on AIM, are you confident that the authorities (the FCA and/or LSE) would investigate and prosecute it?”

Of 142 responses, 140 – over 98% – said they were not confident in the FCA and/or LSE. Just pause and think about that. This was a poll answered by serious private investors. And yet, 98% had no confidence in the regulators.

The judge’s view was clear:

“the Stock Exchange and the FCA both said they had conducted a comprehensive review of the fully reconstituted order book and were satisfied, independently of each other, that there was no indication of any unlawful activity. If (which I could not find in any event) the “admittedly small [Twitter poll] sample” were representative of some a priori concern that was at all widespread, the full evidence in this case should have served as reassurance that the concern was unfounded.”

The judge here misses the point entirely. The poll was not in the context of Burford. It was a general question asked without reference to any specific company. As investors, we see activity all the time which reeks of market manipulation: share prices that fall just before a placing or rise or fall just before good or bad news is announced. Holdings announcements which are filed late, if at all. The sort of price manipulation that had Navinder Singh Sarao complaining to regulators until he decided, out of frustration, to take matters into his own hands. The only thing these activities have in common is a failure of relevant regulators to investigate or prosecute.

The notion that the way to rebuild confidence in a system that is widely regarded as biased against the interests of private investors is to tell such investors that the system is in fact working very well is rather patrician, even by judicial standards. But, if you accept that the system is, in fact, aimed at supporting HFT, rather than the interests of legitimate investors, such a view makes perfect sense.

The judge took particular umbrage to one view set out in the letter:

The UKSA/ShareSoc letter was transparently partisan and built up to an unwelcome in terrorem conclusion that, unless Burford’s claim succeeded, “the perception of private investors is likely to be that the preference of the authorities is to ignore, rather than investigate, market manipulation”. The thought that the court could and should be trusted to assess the case for itself, independently and impartially, appears not to have occurred to the authors.

The thought that the investing community might not be convinced by the court, FCA and LSE closing rank does not appear to have occurred to the judge. Had he looked at Twitter, or any number of bulletin boards, he would have seen that the perception of private investors is exactly as the letter had predicted.

The truth is that rather that strengthen confidence in the FCA and LSE, the judgment weakened confidence in the court.

The Most Important Thing

Prior to the case I felt the most important thing was to rebuild confidence in the markets. The judgment makes it clear that I was wrong:

“a conclusion by the court that there was a good arguable case of unlawful market manipulation in a case where the Stock Exchange and the FCA had each, independently, concluded that there was no arguable case at all, could be perceived as an authoritative finding that the systems in place for detecting and/or acting upon possible market abuse are weak. That might be said to be an encouragement, not a deterrent, to those who might be minded to engage in unlawful market manipulation.”

By now we are in the realms of Kafka. I would paraphrase that paragraph as saying that a claim that the FCA and LSE do not investigate market manipulation properly cannot be allowed to succeed because if it did it would suggest that the FCA and LSE do not investigate market manipulation properly.

It didn’t take too long for bulletin boards to discern the echo of Lord Denning’s judgment in relation to the Birmingham 6: that their claim could not proceed because if it did the damage to the police would be too great. The thought that the investing public may have discerned that echo did not appear to have occurred to the judge.

Indeed, the judge even went so far as to conclude, in paragraph 216, that even if Burford had shown a good arguable case that wrongdoing had taken place, he would still have declined to offer relief because of “a risk of damage to public confidence in the FCA as regulator” and “there would be a finding that if wrongdoing were ultimately established, it might then be seen that there had been fault on the part of the Stock Exchange in relation to its facilitation”.

So, in conclusion, it seems that the most important thing is to ensure that the reputation of the FCA and the LSE is preserved. And faced with evidence that the investing public already has already lost faith in such bodies, the court’s response is that they are doing a very good job, and no, we can’t see the evidence that would show whether that is true or not. Preserving the anonymity of those engaged in activity that looks illegal, is regarded as unethical, but is of sufficient commercial importance to be called “legitimate” is also a high priority.

It was hard to imagine how confidence in the markets could be further eroded. This judgment has managed to do so.

Paul de Gruchy, Director, ShareSoc
The author is a shareholder of Burford Capital.