question on a database payoff

[jasperson]

I am currently tracking a selection system that out of 205 races it has 54.4% winner. Roi to win is $.20 to place is $.22. Saturday it had a 2nd place horse that paid $27.40 to place and $13.60 to show. Would you include this abnormal pay off in the database or would you think it is distorting the payoffs to place and show? I think it would make the place and show bets look better than they reallly are on an average. I didn't have a dime on this horse because I was in a handicapping tournament and spending all my time on that.

[kenwoodall2]

If you would count the real money you count the theoretical cash too! To not do so would distort your data. It's not your responsibility to throw out winnings because other bettors are more stupid than you and did not bet the horse, causing the big overlay! In any individual race, the finish is how the finish is supposed to be, it all averages out. This overlay covers all the cheating, stiffing, bad trips, bumping, scratches, and long layoffs happening when your horse should be winning the race and, in this case, still leaves a profit! I allow 5% for the above and you found a horse that takes care of all that! Congrats!!!

[shouldacoulda]

Why wouldn't it count? If you had money on it it would count. There will always be horses the public misses on. The selection process is just that.At a 54% win rate and a .20 ROI what happens to the correlation if you restrict plays by an odds minimum?

[vegasone]

I think unless you normally get those kind of prices, a big price like that would skew any kind of test. best to leave it out in such a small sample.

[markgoldie]

I think unless you normally get those kind of prices, a big price like that would skew any kind of test. best to leave it out in such a small sample.
Agreed. If the sample were 2-3k races, then that would be different. My advice would be to count it as twice the av. payoff you have found so far and continue on.

[garyoz]

How many standard deviations is it from your mean average of payoffs? Also look at it relative to the median for both categories. Certainly sounds like an outlier and I would toss with only a sample of 200. You are doing it for your own good if you plan to use the dbase. BTW, shouldn't this be in the handicapping software thread??

[jasperson]

Agreed. If the sample were 2-3k races, then that would be different. My advice would be to count it as twice the av. payoff you have found so far and continue on.
Thanks to everybody for their input it has helped me decide what to do with this race. In looking at my data my best payoff to place was $5.80 and show was $3.80. There I am going to enter the horse as $6.00 the place and $4.00 to show. This race (12th at CRC) was a race that all in the in money horses were longshots.

[gm10]

Agreed. If the sample were 2-3k races, then that would be different. My advice would be to count it as twice the av. payoff you have found so far and continue on.

This seems like the most sensible suggestion to me.