Exacta Win Rates

[NormanTD]

My database is showing me some positive roi's betting exactas but I'm not sure if my logic has been correct.

For example for factor A: 34 races, 20.6 win %, roi 1.30% (1 is break even) this looks reasonable however for factor B:34 races, 2.9 win %, 2.69 roi, this looks unreasonable. That's like winning one race and it pays like $500.

At this time I do not have the ability to pull out the exacta payoffs to find this really big one without going thru each one by hand.

Can this happen and do you think the factor B would be worth following and betting in the future?

Thanks in advance.

[formula_2002]

Ben Hogan said,"the harder I work the luckier I get”.

In the case of horse racing betting, the more data I review the less lucky I get. :)

[ryesteve]

Can this happen and do you think the factor B would be worth following and betting in the future?Yes, anything can happen in 34 races... hence I'd base no future wagers on a sample so small.

[bobphilo]

Definetly sounds like one giant score. At a 2.9% hit rate and such a small sample it could have just been a lucky score. You definitely need much more data before coming to any conclussions.

Even when you do, when dealing with this type of data with extreme values, traditional measures like averages (means) can be badly skewed by by these extreme values. I suspect you have a non-normal distribution. I'd suggest you use measures lake median (middle value with just as many pay-offs above and below it) to interprete your data. You can even try it on the limited data you have now just to see how different the result will be. I bet you will be surprised.
Using a much larger sample with median substituted for mean should answer your questions.

Good luck,
Bob

[NormanTD]

Thanks gang. Back to the drawing board.