testing

[Capper Al]

Soon the first phase of my new program will be tested. It is always difficult to determine if an individual attribute has value. Using only strike rate or only ROI doesn't quite get it. The problem, in this complex game, is that an attribute on its own may not be profitable but could be supportive in a comprehensive mix. Frederick Davis instead of tracking the top rated attribute's hit rate only, would track any of the top three ranked hit rates. If the hit rate for any of the top three came in at over 62%, it was considered good. This makes sense. Consider you found a factor that wasn't time or speed based with the top three having a strike rate of 70%. Now you have your contenders in your new found factor.

[cashmachine]

Originally Posted by Capper Al

Using only strike rate or only ROI doesn't quite get it.

Why ROI is not good? After all it is ROI you are interested in. Hit rate of top three choices doesn't take into account field size, nor odds of those three horses. I think ROI is less misleading.

[Capper Al]

Originally Posted by cashmachine

Why ROI is not good? After all it is ROI you are interested in. Hit rate of top three choices doesn't take into account field size, nor odds of those three horses. I think ROI is less misleading.

Most cappers who test use strike rate (win percentage) or ROI or impact value. Probably most don't test. But those that do test very few succeed.

[Capper Al]

What I'm suggesting is that there might be needed other ways to look at results. Given the idea that in a comprehensive mix that the sum is expected to be greater than the whole, an isolated factor may not be needed to be profitable on it's own.

[whodoyoulike]

Originally Posted by Capper Al

Soon the first phase of my new program will be tested. It is always difficult to determine if an individual attribute has value. Using only strike rate or only ROI doesn't quite get it. The problem, in this complex game, is that an attribute on its own may not be profitable but could be supportive in a comprehensive mix. Frederick Davis instead of tracking the top rated attribute's hit rate only, would track any of the top three ranked hit rates. If the hit rate for any of the top three came in at over 62%, it was considered good. This makes sense. Consider you found a factor that wasn't time or speed based with the top three having a strike rate of 70%. Now you have your contenders in your new found factor.

Who is Fredrick Davis?

Are you blindly betting all or some of the top three (or xx number of) rated attributes or those over 62% or whatever? You aren't considering the distance, pace or speed for the race? Or, is it incorporated in the attributes?

Are you going to provide real time examples of handicapping with your new program?

I'm always interested in different (or new) approaches to handicapping. It's the way one improves.

Thanks,

[Capper Al]

Originally Posted by whodoyoulike

Who is Fredrick Davis?

Are you blindly betting all or some of the top three (or xx number of) rated attributes or those over 62% or whatever? You aren't considering the distance, pace or speed for the race? Or, is it incorporated in the attributes?

Are you going to provide real time examples of handicapping with your new program?

I'm always interested in different (or new) approaches to handicapping. It's the way one improves.

Thanks,

Betting on nothing. I'm actually going to test on paper so to speak. I'm not testing a selection method but only one factor, class ratings. I will be comparing mine to BRIS'. BRIS does do an excellent job with their race ratings (RR) and their class ratings (CR). Eventually, I will be providing real time examples. If you are looking for new approaches and are not using BRIS' class ratings, I suggest that you look into them.

[Capper Al]

Fred Davis wrote Percentages & Probabilities back in the day, a long time ago. It is a classic handicapping book.

[classhandicapper]

Originally Posted by Capper Al

I'm not testing a selection method but only one factor, class ratings. I will be comparing mine to BRIS'. BRIS does do an excellent job with their race ratings (RR) and their class ratings (CR).

Interested in the results.

[Capper Al]

Am surprised that more developers haven't jumped into this conversation. Testing methods might be as propriety as formulas.

[traynor]

Originally Posted by Capper Al

Am surprised that more developers haven't jumped into this conversation. Testing methods might be as propriety as formulas.

Yuh. Especially the ones that work.

[dkithore]

Originally Posted by Capper Al

Betting on nothing. I'm actually going to test on paper so to speak. I'm not testing a selection method but only one factor, class ratings. I will be comparing mine to BRIS'. BRIS does do an excellent job with their race ratings (RR) and their class ratings (CR). Eventually, I will be providing real time examples. If you are looking for new approaches and are not using BRIS' class ratings, I suggest that you look into them.

AL,

I use BRIS CR regularly but have not figured out use of RR. Does it mean theoretically, higher RR and CR are better? Is that it?

Do you mind telling me how You use it? Thx.

[Billnewman]

I think RR means the average class of that race

[Capper Al]

Originally Posted by dkithore

AL,

I use BRIS CR regularly but have not figured out use of RR. Does it mean theoretically, higher RR and CR are better? Is that it?

Do you mind telling me how You use it? Thx.

It does indirectly because because CR is based off RR or vise versa. They are related, and yes higher is better.

[Capper Al]

Originally Posted by Billnewman

I think RR means the average class of that race

Agree.

[dkithore]

Thanks Capper Al.