More on Flash Crash Report
I have a column in yesterday’s Financial Express with a more detailed discussion about the flash crash report that I blogged about (here / here) last week.
The Commodity Futures Trading Commission (CFTC) and Securities and Exchange Commission (SEC), which regulate the US equity markets and equity futures markets respectively, have released a hundred page report on the flash crash of May 6, 2010. On the afternoon of May 6, the broad market index in the US dropped by over 5% in the space of less than five minutes only to bounce back in the next five minutes. Then, even as the broad index was recovering, several stocks crashed to near zero. For example, Accenture fell from $30 to $0.01 in the space of seven seconds and then snapped back to the old level within two minutes.
The worst sufferers were retail investors who found their orders executing at absurd prices. A retail sell order (possibly a stop-loss order) might have been triggered when Accenture was trading at $30, but the order might have ended up being executed at $0.01. Some, but not all, of the damage was undone when the exchanges cancelled all trades that were more than 60% away from the pre-crash prices.
The CFTC-SEC report claims that the crash was triggered when a mutual fund sold $4.1 billion worth of index futures contracts very rapidly. The mutual fund’s strategy was to sell one contract for every ten contracts being sold by other traders, so that it would account for 9% of all trades during each minute until the entire order was executed. The large order allegedly confused the high frequency traders (HFTs) who having bought from the mutual fund, found themselves holding a hot potato, and then tried to pass the potato around by trading rapidly with each other. The result was a sharp rise in the HFTs’ trading volume, and this higher volume fooled the mutual fund’s algorithm into selling even faster to maintain the desired 9% participation rate. This set up a vicious circle of sharp price declines.
This story makes for a great movie plot but in an official investigative report, one expects to see evidence. Sadly, the report provides no econometric tests like vector auto regressions or Granger causality tests on tick-by-tick data to substantiate its story. Nor are there any computer simulations (using agent-based models) to show that the popular HFT algorithms would exhibit the alleged behaviour when confronted by a large price-insensitive seller.
Moreover, the data in the report itself casts doubt on the story. More than half of the mutual fund’s $4.1 billion trade was executed after prices began to recover. And the report suggests that the hot potato trading was set off by the selling of a mere 3,300 contracts ($180 million notional value). In one of the most liquid futures markets in the world, $180 million is not an outlandishly large trade. Surely, there must have been such episodes in the past and if there is a hot potato effect, it must have been observed. The report is silent on this. Further, the one thing that HFTs are good at is analysing past high frequency data to improve their algorithms. Would they not then have observed the hot potato effect in the past data and modified their algorithms to cope with that? Finally, during the most intense period of the alleged hot potato trading, the HFTs were net buyers and not net sellers. This suggests that perhaps the potato was not so hot after all.
The analysis in the report is even more flawed when it comes to the issue that concerns retail investors most—the carnage of individual stocks that began two or three minutes after the index began its recovery. The report bases its conclusions, on this issue, almost entirely on extensive interviews with the big Wall Street firms—market makers, HFTs and other brokers.
Astonishingly, the regulator did not find it necessary to interview retail investors at all. This is like a policeman investigating a theft without talking to the victim. There is no discussion of whether retail investors were confused, misled or exploited. For example, the report dismisses concerns about delays in the public price dissemination because all the big firms subscribe to premium data services that did not suffer delays. If delayed data led to wrong decisions by retail traders, that apparently is of no concern to the regulators.
In the post-crisis, post-Madoff world, we expect two things from regulatory investigations. First, we expect regulators to have the capability to investigate complex situations using state-of-the-art analytical tools. Second we expect them to carry out an unbiased investigation without giving high-profile regulated firms undue importance.
On both counts, the CFTC-SEC report is disappointing. After five months of effort, they do not seem to have come to grips with the terabytes of data that are available. The analysis does not seem to go beyond presenting an array of impressive graphs. Most importantly, the regulators appear to still be cognitively captured by the big securities firms and are, therefore, reluctant to question current market structures and practices.
Posted at 11:33 am IST on Sat, 16 Oct 2010 permanent link
Comments