I'm not at all convinced that the entire story of the computer players is being revealed. As TGJB has asked, without answer, "Okay, explain to me that little part where the computer finds the overlays...".

For one thing, I've done considerable work looking at the exacta pool. I can say this with certainty, that the VAST MAJORITY of winning exacta payoffs in NY are statistical UNDERLAY's. If that is true, then the whole theory of what the geeks are doing is in error. No one is going to bet an underlay, looking for an edge.

So , if that's true ,they have somehow calculated the bets using information and data that differs from the raw exacta par calculation. Could be they are excellent handicappers. Could be something else.

Another thing, to test the theory, all you have to do is plot how often the WINNING combination in the exacta is bet down at the end. If the geeks are doing what is claimed, and winning with great frequency , you should see that. I suspect you will NOT.

In order the get back 95% of your money, you'd have to beat the exotic takout by something like 15%. Consistently. Doing that entirely by 'value' mining, with no other input , would be extremely difficult, if not impossible.