No one could say for sure what caused the flash crash – for the same reason no one could prove that high-frequency traders were front-running the orders of ordinary investors. The data didnt exist.
– Chapter 3 of Flash Boys: A Wall Street Revolt by Michael Lewis
It may take a while to unpack these three conclusions that Michael Lewis links together. Let us put aside the fact that, in interviews subsequent to the publication of Flash Boys, Lewis and his protagonist, IEX co-founder Brad Katsuyama all but admitted that this alleged front-running couldnt affect the orders of ordinary retail investors. (This also contradicts the middle conclusion). Let us also grant that, although there are a lot of pretty good explanations for the flash crash, no one will ever be able to say with 100 percent certainty what caused it, if for no other reason than the event involved the intentions and decisions of thousands of individuals, whose minds we cannot read. We are still left with the question: does the data exist?
Lewis is correct in saying that the data to support his claim does not exist. However, he is utterly incorrect when he says that fine-grained trading data doesnt exist.
For example, Lewis writes, The unit of trading was now the microsecond, but the records kept by the exchanges were by the second. This is completely false. Had Lewis really interviewed anyone at an exchange, he would have been disabused of this notion immediately. Every event that occurs in any given stock at an exchange occurs in a precise order that can be replayed. If order A to buy Microsoft arrives one microsecond before order B to buy Microsoft, it is processed before order B. You can call the exchange and ask what happened, and they can replay the precise sequence, in the exact order that everything occurred. Even 10 years ago – when most traders considered the unit of trading to be the second – we would call an exchange and ask what happened at a particular millisecond and get an immediate answer.
But even this misses the point completely. A paragraph earlier, Katsuyama complained that you cant see a single time-stamped sheet of every trade. This does exist – everyone trading in the market possesses this data. I dont know why he didnt. The SEC pored over many time-stamped sheets of every trade in their investigation of the flash crash. The SEC has also pored over many time-stamped sheets of every trade in probes of high-frequency traders, as well as their source code, emails, and other data. The data is there, and it doesnt support the front-running theory.
This leads to one of the most surprising revelations of Flash Boys: none of its allegations of front-running are substantiated by a single shred of market data.
Flash Boys tells us that Katsuyama spent years honing his sales-call pitch of a high-frequency front-running conspiracy – yet during all that time he apparently didnt find any supporting market data evidence worth mentioning in this book. Instead, here he tries to sell us the bogus excuse that such data doesnt — and cant — exist.
At every trading firm Ive seen, when a trade goes wrong the trader replays the market data to see what happened. They walk through every trade and every order in the market data and figure out what happened. In Flash Boys, the reader is simply told that bids or offers disappeared. If Katsuyama were being front-run as he described, he would see trades occurring immediately ahead of his, at exactly the prices and quantities he was trying to obtain.
If no such trades are found, hes not being front-run. He would also see these trades being flipped soon thereafter as the alleged front-runners take their profits and exit their position. Any Wall Street intern with a Bloomberg account could figure this out in 30 seconds. So why is the reader not given the answer here, and instead fed the excuse that the data doesnt exist?
When someone investigating a conspiracy theory cannot corroborate the theory, the last-ditch defense of the conspiracy theorist is to claim that the evidence supporting their conspiracy is now impossible to find.
The fact of the matter is that the amount of data available is voluminous: not just trading records, but order records, market data, computer code, e-mails, instant messages. Regulators can demand this information from any broker at any time – no lengthy court process or subpoena is required. The data does exist, and lots of it.
Oddly, in later chapters Lewis reveals some knowledge of the vast amounts of data available to investigators: John Schwall complains about his brokerage firm recording his e-mail; Sergey Aleynikov talks about using a source control system called subversion, which records each and every change to the computer code used for trading. In fact, every single evening, every single broker-dealer registered with FINRA uploads a record of every single stock order they placed that day, complete with the millisecond that they sent it. Thats worth repeating, because its rather remarkable. Every night, every broker has to submit to the regulators a record of every order they touched that day. Theres no chance of destroying or falsifying incriminating evidence: the orders are all cross-checked against the exchange records as soon as they are uploaded.
In short, this means that every piece of evidence necessary to track down a HFT who is manipulating the markets would be a few clicks away from any regulator. The trades and orders document what happened, the e-mails and IMs document why it happened, and the computer code demonstrates how it happened. Consider the wealth of evidence available: in the past, numerous insider trading cases against hedge funds stalled trying to prove the why and how. If any high-frequency trader breaks the law, the computer code demonstrates irrefutably how it was done and exactly why. The e-mails and IMs provide the icing on the cake at trial.
Every day, tens of millions of trades occur. If Lewiss front-running theory is even partially right, it implies millions of violations of securities laws every day. It would be almost impossible to avoid stumbling across evidence of the conspiracy. One would expect to see hundreds of cases against high-frequency front-runners. But we dont. Why?
The most reasonable explanation would be that if these crimes are being committed, they are quite rare. For the inveterate conspiracy theorists, however, the only way to dodge this conclusion is by insinuating that the regulators must also be in on the front-running conspiracy – which, if you are keeping count, would now have to include high-frequency traders, big banks, all the stock exchanges, and the regulators. Are we really to believe that every one of these groups has sworn to some code of silence? A more reasonable explanation is that (a) front-running is extremely hard to do in todays markets, and (b) the massive amount of surveillance data available deters criminal activity.
As detailed earlier in this chapter, its extremely difficult to front-run in our modern markets. Every would-be explanation of front-running relies on some imagined method for divining the mere existence of a customer order, and Lewis doesnt even try to explain how somebody figures out the number of shares in the order – the one piece of information that is actually needed to make the scam plausible. Even with that information, the economics would be entirely skewed against the front-runner due to the markets rules on the price and prioritization of trades. Lastly, todays markets are made up of a huge, diverse set of participants. Its absurd to think that a predictable front-running scheme wouldnt either be easily avoided, or be a target itself. Any front-running scam would have to overcome each and every one of these obstacles to subsist – and then, it would have to evade a massive surveillance system built to track every piece of trading data available.
This market surveillance serves as a significant deterrent. When a city installs red light cameras that automatically track drivers who violate a red light, fewer people run red lights. Now, imagine one hundred times more surveillance. Imagine that your cars computer tracked every detail of your driving habits – your speed every millisecond on every street, when and where you stopped, whether or not you used your turn signals, with video of who was in the crosswalk, too. Now imagine that the police can stop you at any time and review your cars data to see you if ever exceeded the speed limit or failed to stop at a stop sign. Most people would become much more careful about their driving. In fact, Id even pay extra for a new car if it could prevent me from speeding or running a red light.
So it is in todays electronic stock markets. On any day, the regulators can show up at your door and demand to see every detail of your trading. If they find anything amiss, you dont just get a ticket, you lose your drivers license – you are no longer allowed to trade. In other words, if you so much as run a red light, you run the risk of shutting down your company and permanently ending your career. Its no surprise that most electronic trading firms spend prodigiously on compliance systems that monitor their own trading to stop somebody who is about to go for it on a yellow light. The risk is too high.
The use of computer software itself is also another deterrent. Electronic trading strategies are executed by computer programs. They are the recipes for these strategies. Computers do what they are told to do, nothing more and nothing less. Never before in the history of the markets have any traders had to write down the recipes that state how they are trading and why. Yet electronic trading requires exactly this. If some trading firm really is front-running the markets electronically, it had to write a computer program to do so – and that program is now available to the regulators. How many would-be criminals do you think keep a diary of their exploits, complete with instructions on how its done? Given the amount of surveillance, the odds are heavily against anyone who tries to manipulate the markets electronically.
Is it possible that some drivers will still run a red light, even if they know that they are guaranteed to be convicted if they get caught? Yes. Sadly, in any walk of life there will be those who break the rules. Some of those people will break the rules the old fashioned way, and some will now use computers to do their dirty work. Fortunately, in the surveillance state that is todays stock market, such a violation would have to be pretty rare and pretty insignificant to avoid detection.
Peter Kovac is the former COO of EWT, the market making firm. “Flash Boys: Not So Fast” is available on Amazon.
