my understanding of the historical development of these things is that DE Shaw and O'Connor originally started off by trying to data-mine stock prices to come up with money-making machines. They ended up finding that the quant models they come up with kept on advising them to put on huge more-or-less-offsetting positions in order to exploit small but robust arbitrages. This meant that they had a gruesome problem of slippage, and so they ended up inventing the kind of algorithmic trading systems I was talking about, in order to get their positions on as efficiently as possible. And thus was algorithmic trading invented.
But there certainly are algorithms used to pick stocks - AHL, probably the biggest and most successful commodities fund in the world, is almost entirely formula-driven, and a lot of BGI's hedge fund money is too. The thing is though that in a lot of these things, the actual trading rules are quite laughably simple (moving average crossovers, four week breakouts and such; about the most complicated thing you'll see is a cointegrating regression for trading pairs of stocks), but the system for putting the actual positions on is seriously complicated. If AHL sent you a CD with all of their trading system on it, it would most likely be totally useless to you as you would not be able to execute anything like the trades they do. best dd -----Original Message----- From: PEN-L list [mailto:[EMAIL PROTECTED] Behalf Of Perelman, Michael Sent: 25 November 2006 16:20 To: [email protected] Subject: Re: Formula for disaster This came just before the part I cited. It suggests that algorithms are used to pick stocks. I was very surprised by the estimate of one-third. David's explanation makes it seem sensible. Thanks. "Complicated stock picking methods are nothing new. For decades, Wall Street firms and hedge funds like D. E. Shaw have snapped up math and engineering Ph.D.s and assigned them to find hidden market patterns. When these analysts discover subtle relationships, like similarities in the price movements of Microsoft and I.B.M., investors seek profits by buying one stock and selling the other when their prices diverge, betting historical patterns will eventually push them back into synchronicity." Michael Perelman Economics Department California State University michael at ecst.csuchico.edu Chico, CA 95929 530-898-5321 fax 530-898-5901 www.michaelperelman.wordpress.com -----Original Message----- From: PEN-L list [mailto:[EMAIL PROTECTED] On Behalf Of Daniel Davies Sent: Saturday, November 25, 2006 7:11 AM To: [email protected] Subject: Re: [PEN-L] Formula for disaster aaaargh! algorithmic trading is completely misunderstood and articles claiming it's "a computer to pick stocks" are the problem. It's algorithmic *trading*. The algorithms aren't used to pick the stocks, they're used to minimise the "slippage" in placing the trades. The point is that if you decide you want to buy $100million of Google and it's trading at $499, then normally, by the time your trades have gone through, you've pushed the price up to $505, meaning that the average price you paid is somewhere around $502. If you have a smart trader working for you, he will break your order up into chunks, and time the placing of the chunks for when he thinks that there are biggish Sell orders coming across the market. If he can do this reasonably well, then he might get your average price down to $501, which is obviously a big saving. The "algorithmic trading" models that this article is talking about are just a way of automating that process. Some of them are quite tricksy in forecasting the flow of orders, but the main gain from an algorithmic trading system is that a computer can watch a lot more broken-up orders than a human being and never forgets about any of them. There are some automated stock-picking systems out there, but they're nowhere near a third of the market (and most of them are basically used by index funds). best dd -----Original Message----- From: PEN-L list [mailto:[EMAIL PROTECTED] Behalf Of Perelman, Michael Sent: 25 November 2006 02:31 To: [email protected] Subject: Formula for disaster The New York Times discusses the growth of trading financial assets by mathematical formulae. This type of trading can potentially overwhelm markets -- something the article does not mention. Duhigg, Charles. 2006. "A Smarter Computer to Pick Stocks." New York Times (24 November). "Studies estimate that a third of all stock trades in the United States were driven by automatic algorithms last year, contributing to an explosion in stock market activity. Between 1995 and 2005, the average daily volume of shares traded on the New York Stock Exchange increased to 1.6 billion from 346 million. But in recent years, as algorithms and traditional quantitative techniques have multiplied, their successes have slowed." Michael Perelman Economics Department California State University michael at ecst.csuchico.edu Chico, CA 95929 530-898-5321 fax 530-898-5901 www.michaelperelman.wordpress.com
