Proposing a Profitable and Useful Horserace Betting Gizmo

[Tater10]

Excuse this bump, I just came across this thread.

I've been doing just what is suggested (simulating race outcomes using an avg and std) for years. I, however, have been using dog track data. You can obtain it online for free, and there is plenty of it. Most importantly, there is a dog track 15 miles from my house, and they have $1 beers and hotdogs every Friday. I do this only for entertainment.

I have written software to do the sims, but lately the program I use is excel. I only simulate 2k-5k races, not millions. Using " =normsinv(rand())*std " will simulate 1 running of a particular dog, go from there. There is no point in getting data to the .001 if you just look on the board and see an overlay at 5:2.

Using my math skills, this is only valuable for win bets. I believe I've won more than my share of 13:1 shots that I thought should be 5:1 over the years because the 'crowd' doesn't see the volatility in the dogs. I guess they see crappy times in previous races and X a dog out.

Where I have had problems, is the correlation between dogs. This severly hinders quiniella, exacta, and trifecta wagering. I don't have the expertise to explain perfectly, but looking at previous data, I can say that if the #3 dog wins, the #4 dog has a much better chance of placing than if the #8 dog wins. Look at historical quiniella charts, each one does not have a probability of 1/28th. Implementing this is difficult (again, with my crazy math skillz)

Trifectas are a different story. Most importantly, they don't post odds for individual trifectas. I know of no way to make an overlay bet if you dont know the 'Over' of the 'Lay'.

This may work better for horses, as the races are much "cleaner" - you don't have to worry as much about the #8 cutting off the #3.