The game is almost over. The days of picking off crosses on the ECNs have almost evaporated, since there are people that are running very fast automated systems to do this, not to mention inter-ECN linkages. Trying to arb against or among exchanges (NYSE and AMEX) is very treacherous, primarily because it is almost impossible to change or cancel an order or get an execution from them in less than about 15-30 seconds, and sometimes much longer, even in very liquid stocks like ETFs. Compared to ECN speeds, this might as well be forever. Your only hope is to be a great tape reader, and to pick off orders on the ECNs that you "know" will be profitable based on the way the exchange market is about to move. While most of my trades could be considered arbitrage, it's a lot more complicated than that, and requires a lot of manual work and experience. There are very, very few riskless arb opportunities available in equities any more. As someone else said, if there were such opportunities, people aren't going to tell you about them. Heck, I might even be lying
I've been doing some semi-riskless arbitrage for some time , but it requires precision entry and exit, trader judgement, and would be very difficult to do with a computer, but probably not impossible. The win rate is near 100% (small sums , so don't be too jealous) and I can't reveal it so don't ask. As far as arbing between ecn's , I doubt there are many opportunities to do that, but if a computer was scanning 1000 stocks at once it might find a few a day. I suspect the best arbs come from knowledge of interstock/market relationships that get out of whack due to supply/demand imbalaces, and not crosses. In essence a cross is evidence of some traders ineptness, error, or intent to manipulate a market, and is hard to come by, though I have seen them.