Hi all,
I have been running some automation that I put together, I have run it on every NH, AW and US race over the past 3 weeks, totaling a few thousand races. I'll put the results below. It picks out one horse to back in-play.
Stats and such like are not my strong point, what sample size would be needed to say that this is likely to be a profitable strategy rather than luck? How many races would it take to say that these results are not just random? It is a good profit, but I can see that the 1st week was the strongest, with the most recent week being the weakest, perhaps an edge existed that is diminishing. Is there a way of knowing these things or is it a case of constantly tracking and evaluating as you go along?
Many thanks
Week 1
NH 89 points
AW -19.59 Points
US 40.4 Points
Week 2
NH 37.9 Points
AW 11.12 Points
US 12.4 Points
Week 3
NH -10.9 Points
AW 4.4 Points
US 9 Points
Profitable or lucky ? What is the required sample size?
NP, mathematically it's possible to calculate the exact sample size, but I think it misses the point. So I've resorted to this method to understand if I am on the right track as it became more accurate as time elapses, which is more or less what you want.