Baseball - Converting my total prediction to a winning percentage
Stat Experts,
Another question I have been struggling with. My Baseball model projects a total that I can compare to the total given by the sportsbook. However, I simply have no idea how to convert that into a winning percentage. It seems like it should be easy but I have really been struggling with this. I assume it would have something to do with the scoring distribution which I have - just not sure how to apply it.
For example - If my model predicts the total as say 7.72 and the total is at 8.5 - what is my edge on a -105 line? When the number is closer I need to know if I even have an edge at all.
Agree with flsaders. I find the Monte Carlo Simulation to be the easiest method for this kind of thing.
Try this: simulate like a 100.000 games with the 'projected' total of 7.72. Then check the results to see what % goes below 8.5.
For example: I don't cap bases and don't know how the score is distributed. But given your numbers, I made 1 million simulations with poisson distribution. The chance that the score goes below 8.5 is ~ 63.13%.
If you know a little of pyhton that can be done with a few lines of code.
Thank you - Looking into this now. I see that you can do a custom discrete distribution with Monte Carlo - Not sure how to work my prediction into it quite yet but will keep plugging away. Thanks for pointing me in the right direction!
Baseball totals are tough since it is heavily weighted on 7 and 9 with a dip in the middle at 8. Seems to throw everything off.
Last edited by Dead__Red__777; 01-20-15 at 12:05 AM.