Woof43,
How do you account for field strength in these calculations? If you had a stronger/weaker field running within a certain class, then by the process you outlined the difference between winning time/par time would be attributed to track speed, when a component of the difference would/could be due to the relative field strength of the class in question.

Here we all are again. The horses are machines and follow a certain "precise" set of mathematical laws, scientifically calculatable (??) which makes even a Swiss clock look "haphazard" by comparison!!! Past performance, as has been pointed out (not least, by every Fund Manager/New Stock Float PDS in the world!!) is not a guarantee of FUTURE performance. Yesterday is only a "rough" guide fellas,at best:an approximation only able to tell us WHAT IT TELLS US: that if these horse from Melbourne, Wangaratta,Kyneton and Murtoa had met, at Geelong, over 1500m YESTERDAY, when it was pouring, we could've made a killing!! In fact we could've got the "first fifteen" no sweat. Average times? Perfect so long as no horse IMPROVES!! Nor ONE blunders and messes the others up. Nor the "best" horse has been kept awake by howling cats next to the stables all night. I see something "unexpected" in nearly every race I watch,yet there appears to be no place anywhere in YOUR calculations for such surprises.

This is simply a rehash of the idea that the fastest horse,from yesterday, ("the right time", in Woofspeak) wins. Who should we have backed in the Olympic men's 100m final last year? We all knew the best times for each runner AND there was nothing like "pace" or interference or "the bend" or Barrier Draw or "going" to worry about. The bookies had the odds at just about "perfect", statistically. Pity they couldn't pick the winner!! As Ron Clarke once noted (most ironically, when you think of it) after losing AGAIN in a big 5000m Final "But he (Keino) had to set a world record to do it!!" Or when Roger Moens was nutted by Peter Snell in the 1960 800m: "who is he???" By the way, Snell is the quintessential problem for "statistical" analysts such as you Woofy. He was (in retrospect) both totally predictable AND totally unpredictable: he NEVER lost at the Olympics or Commonwealth Games where "statistically" he was most vulnerable {on times,only marginally superior or,sometimes, inferior to his opponents) In smaller races where he was FAR BETTER on times (and "performance" to boot) he often didn't "get there". We know that NOW!! Just as we know that a "slow",class 2 performer from the boondocks COULD win the Group 1 S.A. Derby this year. Just as we know that the fastest 800m runner (Coe) of his, and most generations, never DID win the 800m at a Major championship, despite "the clock" telling us what SHOULD have been. Both his 1500s at the Olympics were, "statistically seen", the WRONG races, (try telling that to the beaten brigade!!),though NOT "psychologically seen".
Why do I mention all this? Because horseracing is FAR MORE uncertain than human racing, and since we can't even begin to have them "pegged" (the humans, I mean) what folly it is to waste braincells on quantifying that which is not quantifiable (the horse) to the LAST SPLIT OF THE HAIR

A case of "I can't make it work, therefore it can't be done", perhaps?
Various ways of determining relative ability within a race-field can't be used to profitable outcomes?
Fine, tell that to my bank balance.

Hi Punter,
Maybe you must have misunderstood the intent of this thread, what i described in my posts were how I went about identifying the Class runners in each race.
My Class analysis provides only one of 4 crosshairs in the overall ability matrix.
Sorry i didnt explain, this isnt my Speed nor Pace method.

Shoto, what you have mentioned is so important, the method i outlined earlier measures the combined Class strength an the physical Track configuration which means its messy.

I turned to greyhound racing for some answers, as greyhounds perform more often than horses an go backward an forward from track to track.I built a large database of performances, what i was interested in was in quantifying the track to track adjustments then the grade to grade adjustments from track to track.

I built a whole set of Grade Par Times based on Grades, then i looked at Track to Track adjustments based on Track records an using a combination of MPS, linear regression an centripetal forces based on the physical layouts
.
Then i did a number of tests based on looking at greyhounds performances based on track to track avg's, single performance figures track to track (that is dogs only racing once at one of the tracks), performances that occured in 7 days , 14 days right up to 200 days between each performance, then looking at a 3rd performance or a return visit to the first track raced, then age analysis, as age degradation also impacts on greyhounds but in a shorter time span and finally Male an female comparrisons.

With that done if one was to look at the two major dog tracks in Vic the Class component from one track to the other is around .06 of a total of say .19 and this is where the Best of the best race from the state.

No need to say its valuable to know the class strenghts from track to track apart from the measuring the physical component of the track to track.

Fascinating Woof, and obviously a very thorough approach. I trust you are duly rewarded. Through different means I aim at a similar outcome with the GG's, where the class (ie Grade) system has become so diluted that the official Class of the race is quite often meaningless, and one must find other ways of determining the real field strength.

Yes, but punter57 does bring up a point which I have found very valid.

One cannot really quantify unknown potential, increased performance, interference nor anticipate that a horse will perform as expected. This is where many ratings systems fall apart a little.

Not saying it doesn't work and definitely not poo pooing anyone's methods, just throwing an electricians spanner in the works.

Consider this....

Horses rising in class have a better strike rate than horses dropping in class.

Horses rising in weight have a better strike rate than horses dropping in weight.

Horses running very fast race times and very fast sectionals do not reproduce this as often as expected. Some horses put in an enormous run and never reproduce this in their career again.

How do we get around all this?

__________________
RaceCensus - powerful system testing software.
Now with over 400,000 Metropolitan, Provincial and Country races!
http://www.propun.com.au/horse_raci...ng_systems.html
*RaceCensus now updated to 30/04/2024
Video overview of RaceCensus here:
http://www.youtube.com/watch?v=W821YP_b0Pg

Hi Chrome,

If you rate each horses individual performance an make no adjustments for trouble or whatever and then find the Standard Deviation of its performances then do this for hundreds of horses from the Best to the Worst the youngest to the oldest you will find the answers your seeking.

If you use the Probability function within Excel an try manipulating the Stdev an check the effect it has on the probabilities I'm sure you will agree the importance isn't on the avg rating figure but its to be found within the stdev figure.

The main focus of ones attention should be on how to tighten or widen each horses Stdev.
Then find Variables/Factors that allow you to make minor adjustments to the Stdev based on either the individuals past performances or by analysis of all runners for that same Factor eg Horses starting from say inside barriers over X distance may have a very tight Stdev based just on those past performances, so a Horse starting from that position today may need to have slight adjustment to its Stdev based on this analysis, or say horses returning from a very lenghty spell may have a bigger/wider stdev so we might make that adjustment then to that same horse.
You can also use it to isolate races which have a tighter overall average stdev which means a more stable result, an this is what in fact will happen. A step further is to find in those above races where your top horses have a bigger advantage over its rivals, then your Golden

Hi woof43,

My questions are not meant to put down what you and others do, you clearly have a really good grasp on how you go about it. I must say I'm impressed by many of your posts, as you give detailed info. As a stats punter, I'm trying to come to grips with a lot of your stuff. At this point I'm far from anywhere near your level, but have made some amazing dscoveries along the way.

Many of my systems just don't provide enough action and hence the profit is not enough to employ a large amount of time on it. Having said that, my systems continue to produce profitable results.

I'm really looking to rate races to come up with some more profitable action.

Thanks for your input.

__________________
RaceCensus - powerful system testing software.
Now with over 400,000 Metropolitan, Provincial and Country races!
http://www.propun.com.au/horse_raci...ng_systems.html
*RaceCensus now updated to 30/04/2024
Video overview of RaceCensus here:
http://www.youtube.com/watch?v=W821YP_b0Pg

Shoto,
In reply to your question about excepionally fast or slow times effecting the daily track variant the method I use is to simply ignore any unsually fast or slow times on a particular day.
In this way a horse that has run a particularly fast time on a (say) 'slow' day will be rewarded when it's adjusted time is calculated.

to Punter69
We all acknowledge that there is no way to calculate all the variables in any one race but if you choose to ignore times, beaten margins, etc just how do YOU select the best horse in the race??????

Morning Zorro.I've got almost 15 minutes to my first possible bet of the day (AR1) so I'll have to be brief. in selecting the bettable nags I'm looking for the psychological factor. That is.... the trainer's plans /motivations, and am always asking myself NOT which is the "best" horse but WHY each horse has been PLACED in such and such a race.
I don't bother with the favoured horses (near the jump) as they ARE probably rated (by the money) as good things and need no further analysis. It's the runners over 20-1 who have me intrigued, as I can't understand why a trainer would even bother turning up if his/her horse is as "hopeless" as 20 or 50 or 100 to one suggests. Right this minute I'm considering Number 8 (about 50-1)> This one did very poorly at Mildura last time but the trainer has brought him all the way from Normanville for a MUCH harder race. Why? Let's see!!
P.S. I'm Punter 57 NOT P69: only my BEST FRIENDS can call me that!!!