The problem with arbitrage of that kind (indeed, most arbs) is they are based upon small but relatively persistent inefficiencies — which have a tendency to get pretty efficient pretty fast.

But it was good learning experience, using networked PCs (almost but not quite neural networks) to find and exploit small price differences between markets.

Now, I’m just another jackass with an opinion on pretty much everything . . .

“Subsequently, Mr. Ritholtz developed and implemented proprietary analytical criteria for automated equities trading. This led to the design of a computerized arbitrage program to take advantages of price discrepancies between Regional Exchanges and the New York Stock Exchange. The program was successful enough that it forced the Regional Exchanges to modify their trading and membership rules so as to thwart any future programmed arbitrage trading.”

This struck me as kind of funny, because some friends of mine have been involved, off and on, in a project to do computerized foreign exchange trading (see http://www.moneyhackers.com/ ).