Creating an Oddsline

[Outlander]

I think the thread has gone away from my initial description. Its perfectly valid to talk about how one would aggregate and in some way get a spread of possible ratings given the last X ratings of a horse but this is not what I am describing. The ML algorithm produces one probability (I will stop using rating) for the race. The question is what is the possible range, say 90% range of probabilities it may have. The residual errors on past training data is used to create a range

about creating an oddsline from ratings. Essentially it revolved around formulating a normal distribution of ratings for a horse,

Using these derived ranges we can then perform a monte carlo simulation, thats the part that tests it all thousands of times, to get an oddsline.

So if horse horse A is a 65 horse with a range of 60 to 70 and randomly 70 is picked whilst randomly 65 is picked for horse A then horse B is predicted to win. Do this many thousands of times and use the win scores to create a probability for each horse.

Silly me got the complete wrong eng of the stick for some reason I thought it was a thread about using ratings to create an oddsline. How I veered off into this territory I will never know when it’s so clear the thread has absolutely nothing to do with that, what a brain fart I had.

 

[woof43]

Silly me got the complete wrong eng of the stick for some reason I thought it was a thread about using ratings to create an oddsline. How I veered off into this territory I will never know when it’s so clear the thread has absolutely nothing to do with that, what a brain fart I had.

I have been looking at what you posted, Mike, it's good for a discussion as I predominantly use MCsims, looking at your race 1 data, it would need a little fine tuning, and some decisions to be made , if it's HK related the avg runners have a stdev of around 3.5-4.5 , so that is a start point, identify runners that sit outside those points, 3 and 8 do , then you have the 5, 1 and 7 lesser degrees, 8 having the largest stdev at around 7.09 or a MADev of about 7.84 which is a stdev based on Medians. any sort of rolling avg would pick o up that runner r1-r6 are different populations to the last four races so the choice one would make is run with the r1-r6 data then you have an r4 data point that is an outlier so remove that figure so now that runner has a 97.1 for r1-r6 but your still going to use the original Stdev of 7.01 . Now you why use that Stdev because you know the runner has had ability prior so you need to make sure the MCsims are using its upper ratings, because when you run the MCsims if you using NORMINV it will make sure each interation is spread around the mean, but you know a runner has a physical limit to what it can run time wise so you would make that runner Upper limit either the 109.15 or just use 97.1 + 1stdev = 7.01 = 104.2 as its expected best rating so just doing that would move that runners Probs from around the $6.00 to $14.00 from my reckoning , I'd would work thru each of those other runners like #3 r8 is an outlier here is a jpg of that number 8 and 9 i didnt limit the 8 to 104 etc but let it ride