Saturday, May 8, 2010

Thursday's Sell Off

I don't think I have ever felt my heart sink into my stomach the way it did when I was watching the chaotic sell off at 2:45 on Thursday afternoon.  I literally thought that there had been another major terrorist attack, or a natural disaster orders of magnitude more severe than the oil spill in the Gulf (I do not mean to make light of the Gulf situation, but it's market impact thusfar has been relatively limited).  Many market commentators and participants have blamed a "fat finger" putting in a sell order for 6 billion shares of P&G instead of 6 million.  This may or may not be true, but it is ignoring the much more important issue.  Once P&G started tanking, the rest of the market waterfalled with it, with the Dow down nearly 1000 points in literally minutes.  The culprit is clearly algorithmic trade machines. 

For those unfamiliar with the basics of how such programs work, they effectively trade securities based on the historical correlations between them.  So if asset X drops historically in tandem with asset Y, and on a given day asset Y drops, algorithmic trading will exploit the 'mispricing' of asset X and short it based on the historical relationship between the prices of the assets.  Then assuming a third asset (Z) historically falls when asset X swoons, another (or perhaps the same) algorithmic trading program will sell asset Z.  These trading programs, combined with an erroneously large sell order which caused P&G to drop enormously (providing the spark), resulted in a positive feedback loop with enormous impacts on asset markets across the board.  In this particular case, cooler heads prevailed and markets closed within spitting distance of where they were before this unprecedented sell off, but with trading trending in an increasingly automated direction, the next situation may not end so happily.  In my opinion, the powers that be in financial markets want to distract attention away from the potentially enormous adverse impact of algorithmic trading, which provide very little economic value added to markets.  This is why we are hearing more about a fat fingered trader in P&G than about algorithmic trading when this sell off, which is unprecedented in combination scope and speed, is discussed.

I don't want this rant against algorithmic trading to be understood as opposition to the high-frequency trading we were hearing about last summer where programmers equipped with the most sophisticated computers were supposedly front running the orders of, and thereby ripping off, institutional investors.  These allegations are a complete misrepresentation of the reality of how such traders operate.

No comments:

Post a Comment