Bailout

The weirdly rational flash crash

By Felix Salmon

As the Dow started falling dramatically this morning, my colleague Frank Tantillo and I put together a Google spreadsheet comparing the Dow’s current level to what it looked like at the low point on the flash crash day of May 6.

It’s not all that easy to understand, especially now that the Dow seems to be rising again, but basically what I was trying to do was get a feeling for just how crazy the markets got on May 6. My intuition was that if the Dow today, near its May 6 lows, looked very much like the Dow actually looked on May 6, then the crash would be much more rational than if the Dow components today were trading at wildly different levels from where they were trading at the lows on May 6.

There’s a few weird bits and pieces in the spreadsheet. For one thing, because the Dow components’ low points didn’t all happen at the same time, the Dow didn’t fall quite as much as the spreadsheet makes it look. (The actual low point is in cell B2; the fake calculated one is in cell B34.) And what’s more, because there’s still a significant gap between the Dow’s actual level (in column C) and its low point (in column B), I created a new column (column F) which simply takes the current price and shrinks it so that the overall Dow works out to the May 6 low point.

The interesting stuff starts happening when you look at the difference between rational, real-world depressed stock prices today, and the crazy, flash-crash depressed stock prices of May 6: it turns out that there isn’t much of a difference at all. In fact, with the exception of 3M and Procter & Gamble, it seems that the Dow fell in a pretty rational manner on May 6. Those two companies overshot a lot, but everything else looks like it’s pretty much in the same ballpark today as it was at the height of the craziness.

If the flash crash was caused by liquidity completely drying up and computers doing insane things which make no rational sense, then it seems to me that the relative prices of the Dow’s 30 components wouldn’t have stayed as tightly correlated as they did: I don’t think there was a lot of relative-value Dow component arbitrage going on during those 15 minutes.

And clearly the absolute level that the Dow reached on May 6 wasn’t all that irrational either, since we got pretty close to that level today, just a couple of weeks later.

So while the volatility we saw on May 6 was crazy, and the speed of the drop was unprecedented, it seems to me that the stock market didn’t break, during those 15 minutes, quite as much as conventional wisdom says that it did. There were a few genuinely crazy individual prints, but overall, something seemed to be working.

Full post as published by Macroprudential on May 20, 2010 (boomark / email).