Question on extrapolating Odds

[andicap]

I am wondering how to extrapolate odds from smaller to larger fields and vice versa.
I am taking the odds of the top three horses in a particular category (its related to CJ's figures) and need to know how to allow for field sizes say by converting them all to the equivalent of 8-horse fields.

For example, if the odds of my top three total 6 in a 5 horse field. (e.g., a 2-1, 3-1 and 1-1 horse) how would I convert that figure to an 8 horse field where the total might be 7? Obviously the larger the field sized the higher the odds are likely to be.

Simiilarly if my three horses add up to 15 in a 10 horse field, how could I convert that to my benchmark?

Secondly, although I am no math whiz I do feel I should cap the odds somewhere so I don't distort the result. (e.g., if I have a 2-1,3-1 and 50-1 horse, averaging them would come to 18-1, a real distortion. Using the median of 3-1 doesn't seem right either. Seems like if I use a cap that might help.)

I know nothing of standard deviations, variables, etc.

Thanks in advance.

[Overlay]

Why not determine the winning percentages associated with the rank order position of horses in fields of various sizes using cj's figures, and go with that to determine your odds? Alternatively, you might calculate the overall impact values associated with each of those ranks in general (regardless of field size), and use them to extrapolate from one field size to another. For instance, an impact value of 1.50 for a given rank would translate to a 30% winning probability (or odds of 2.33-1) in a five-horse field (50% greater than the random winning probability of 20%). Similarly, a value of 1.50 in an eight-horse field would translate to 50% greater than a 7-1 (12.5%) random probability, or 18.75%, representing 9-2 odds. (I'm assuming that you're dealing only with the one type of figure or performance measure, and that it allows ranking of each horse in the field, preferably with a smooth winning-percentage or impact-value progression from top to bottom.)

[GameTheory]

What are you doing with this number? As Overlay says, it sounds like you should be dealing with percentages or IVs instead of odds. Or just keep each field size separate. There are often real differences there anyway...

[Tom]

I thinl percentages is what you need to look at.
Each odds has a number of percentage points, and the total for any race is alway 100, so you could take you 33 percentage points in a 6 horse race and convert it to a 7 horse race rather easily, depending on what you are trying to do.

[BetHorses!]

(I'm assuming that you're dealing only with the one type of figure or performance measure, and that it allows ranking of each horse in the field, preferably with a smooth winning-percentage or impact-value progression from top to bottom.)


What if you are dealing with multiple type figures such as a final speed figure, pace figure, class figure and a recency fig?

[K9Pup]

I am wondering how to extrapolate odds from smaller to larger fields and vice versa.
I am taking the odds of the top three horses in a particular category (its related to CJ's figures) and need to know how to allow for field sizes say by converting them all to the equivalent of 8-horse fields.

For example, if the odds of my top three total 6 in a 5 horse field. (e.g., a 2-1, 3-1 and 1-1 horse) how would I convert that figure to an 8 horse field where the total might be 7? Obviously the larger the field sized the higher the odds are likely to be.

Simiilarly if my three horses add up to 15 in a 10 horse field, how could I convert that to my benchmark?

Secondly, although I am no math whiz I do feel I should cap the odds somewhere so I don't distort the result. (e.g., if I have a 2-1,3-1 and 50-1 horse, averaging them would come to 18-1, a real distortion. Using the median of 3-1 doesn't seem right either. Seems like if I use a cap that might help.)

I know nothing of standard deviations, variables, etc.

Thanks in advance.

So you are making the assumption that these horses would NOT be wagered in the same ratio if more entries exist?

Otherwise just convert those odds to percentages.

prob%= 1/(odds+1)

So a horse with odds of 1-1 has 50% chance to win, 2-1 has a 33% chance, etc. This doesn't change no matter if the race has 5,8,10 horses.

[Overlay]

What if you are dealing with multiple type figures such as a final speed figure, pace figure, class figure and a recency fig?

That's why I mentioned the stipulation about dealing with just that one kind of figure. Trying to combine diverse performance measures into an odds line requires taking account of possible dependent relationships among them, and their various weights with regard to influence on the outcome of a race.