Track Variant
drogba
Junior Member
Hello
What is the best model to compute the track variant for a given day / race.
I believe the highly publicised Beyer figures are good, because they predict those results we expect by intuition many times, but where is the scientific background behind them ?
What are the benchmarks ?
What is the best model to compute the track variant for a given day / race.
I believe the highly publicised Beyer figures are good, because they predict those results we expect by intuition many times, but where is the scientific background behind them ?
What are the benchmarks ?
Comments
Myself, i think in order to come up with a relaible track variant, you need to be familar with all the horses on the grounds, so that you can estimate how fast/slow each particular race should be run. The problem with using class groupings is that there are quite obviously horses off differing ability amongst each different class level. No two races are the same. So you can't just look at a NX2 allowance and say 'it should go THIS fast", because you need to know who's running. Once you're able to pin down a somewhat accurate guess on whatever races are being run, you can see how much (if any) the times deviate from that, and adjust your figures accordingly.
Welcome to the forum, btw...
Sounds like a tautology to me.
A pdf is a benchmark (probability distribution function).
So for 5 furlongs say you should obtain something like a figure of 20% in my reckonning. That is the average probability with which the race winners are predicted by the track variant model should be around 20%. The higher the figure the better.
I believe the class figures are the first approximation in an iteration process and you should have good class figures of course.
Also you need histogram analysis to refine it, when you try to assess each horse individually in relation to its past performences.
Has anyone published his method ?
I don't trust the experts big time, because horse racing is like show business really.
If somebody -anybody- nominates a horse as favourite we tend to believe him, if he is the owner of a publication and if someone is partially good we worship him.
In this game there are no university exams to "fry" the candidate expert - which what we should do to them really. It's the same in America-Australia-Japan-Canada-the world.
And half your post went right over my head...
Basically it's this:
T(new) + s(new) = T(old) + s(old)
The objective is to order the horse times in the future races as close as possible in the order those races will finish.
You can start with class pars but it's only a first approximation, because of the individuality of the horses in any given race.
If you don't go any further than class pars you will still predict some of the races correctly.
It's more to it than meets the eye, but I guess some useful books may exist (for me to see what others have done that is).
Anyone who fancies to work on this one, needs to have access to all the horses' records of course.
In my book if you are a tipster or bettor you are entitled to fail only when a) all the other people fail simultaneously or b) when there is a midnight job (a case where the horse shows vast improvement between races that is).
Does he tell you how to construct track variants (his method) ?
There was a book from English Timeform company, with a flow chart like diagram explaining how they were doing it. Unfortunately I lost it, I don't remember the title and I think it's out of print !!
The rest are apparent.
I mean if you go after jockes it's like when you go to ask for a bride. You notice the floor is unpolished in the bride's house and you walk out of the marriage. But your decision is only an educated guess and it is not based on real facts. Maybe the trainer who hires a low average jock is out there to pull the wool over your eyes. You never know.
So you ought to look for speed only and then exercise your judgment. You have to pray.
To evaluate speed properly you need to perform first histogram analysis (i.e. is the last performence excusable and to what probable extent ?). You also need to convert speed figures in one distance to speed figures in another distance. I mean if your horse runs 5 furlongs today and it's last four outings were over 7f what do you do ? You scrap form and use the 5f speed figure from Noah's ark time ?
You also need accurate measurements of track variant of course, because the racing surfaces are always uneven and because races are never ran in the same tempo.
This is very much different than any other sport. If you have a good 400 metres human sprinter he's going to perform the same in Los Angeles - Tokyo - Munich, when he is after the medals. But this is not the case with horses, whether in the same race course or traveling elsewhere,