Who Really Did Cause The Flash Crash?


It’s hard to believe that it has been almost three years since the flash crash.  Although this seems like a long time, the flash crash was an event that will be forever burned in every investors mind.  While there have been some folks like Jim Simons of Renaissance and even Warren Buffet who have claimed that the flash crash didn’t really matter, most investors still view this as a major confidence shaking event.  As you all know by now, the official government report on the flash crash blamed an e-mini futures algorithm initiated by an institutional investor as the cause of the flash crash.  But many folks in the investment community have questioned this government report and do not believe its findings.

Well, almost three years later, a new academic paper has been published by Albert Menkveld titled “Anatomy Of The Flash Crash”  Before getting into the specifics of the paper, it’s important to note that the author has stated “There are no competing financial interests that might be perceived to influence the results and discussion of this article.”

Let’s skip right to the conclusion of the paper:

“This paper complements earlier empirical work on the Flash Crash with new evidence. It adds analysis based on a proprietary dataset provided by the large seller whose trading, reportedly, was a key part of the crash.  By and large, the evidence reveals that her trading did not cause any of the steep price declines ahead of the E-mini halt in a direct manner. Her trades in this period however are followed with a 300 millisecond delay by strong net sells by others and, at that same time, a sharp price decline. The evidence further reveals that large seller strategy was driven by a volume ratio target, but she did not follow it blindly. For example, she timed her trades so as to benefit from transitory price increases. Finally, a simple Grossman-Miller type calibration exercise reveals that the ‘price pressure’ she paid was extraordinarily large.

One reading of the new results is that the crash cannot be attributed to a single agent but really is the product of agent interaction.” 

Agent interaction?  But we thought that the high frequency trading was exonerated?  It said so right there in the government report, right?

Isn’t it amazing how much truth comes out when research uses actual data tracked to the millisecond.  The data for this paper was supplied by our friends at Nanex and the actual institutional investor who was blamed for the flash crash.  Professor Menkveld’s paper is also highly consistent with the “hot potato” volume concept that was introduced by Andrei Kirilenko of the CFTC after the flash crash:

“Still lacking sufficient demand from fundamental buyers or cross-market arbitrageurs, HFTs began to quickly buy and then resell contracts to each other – generating a “hot-potato” volume effect as the same positions were rapidly passed back and forth…the interaction between automated execution programs and algorithmic trading strategies can quickly erode liquidity and result in disorderly markets.”

Sure sounds like the “agent interaction” during the flash crash had nothing to do with market making.