Do you mean like that?
Hong Kong Speed Figures
- Thread starter The Hare
- Start date
Yes I mean like that.
what I’m referring to here is U.K. standard times in the public domain especially historic racing post standards that have been flawed and how difficult it is to compile accurate standards and that anyone that is just using medians or averages is not going to get close, also I’m doubting the likes of grok can take the data and know how to compile accurate standards unless you tell it exactly how to do it.
What you have here is something much more sophisticated obviously , this is the product of not just taking averages or medians, more as gone into this clearly, this is your answer to the problem, in the last day we have 4 people who study HK posting their standard times none match perfectly with each other, to me it’s clear all have been arrived at with sophisticated methods, but even then they differ.
What you have here for Happy Valley exactly what you need , you need times for every section to evaluate the race properly.
I have my version of this, I’m guessing everyone on this thread would have times like this for HK.
I look at the standard times published on the HKJC website and I wouldn’t want to be using those to create figures, I would happily use your times as I can see they make sense, but there is no substitute for using your own methods and knowing exactly how everything was calculated and having confidence in your own process.
That doesn’t mean mean that I personally think I can’t learn from others, quite the opposite.
Well, if you have the data, then I would not worry too much about what other people do.
You know what goes into your own, and that's the main thing.
Anyway, they may not match perfectly(the four lots posted here), but they are pretty close.
I have a way of checking how much any factor contributes to the overall.
No 1 is public odds this time
No 2 is public odds last race.
No 3 is numbers derived from my final times.
I don't think that the numbers from the sections add too much, certainly no more than just normalised ranking numbers do.
You know what goes into your own, and that's the main thing.
Anyway, they may not match perfectly(the four lots posted here), but they are pretty close.
I have a way of checking how much any factor contributes to the overall.
No 1 is public odds this time
No 2 is public odds last race.
No 3 is numbers derived from my final times.
I don't think that the numbers from the sections add too much, certainly no more than just normalised ranking numbers do.
Good Luck, last season you were very much on the mark with your comments
Your right on everything you said here, AI can only just make your coding quicker, but it can be a terrible debt to pay if you don't understand it's code, It's interesting that your sectional standards sum to your overall standard, which is very different to how I do it, again we all do things differently, but in saying that my Exp/act Track speed for sectionals will nearly always sum to the overall Exp/act track speed. I only use my sectional standards to get a raw S1-Par to a difference in comparsion with my S1-TV difference, and then multiplied by the a sectional weighting to get a final figure.
Woods 2IC sent me that many years ago.
It's not bad, and similar to how I would do it.
I can't remember why they sent to me, perhaps they were partying at the time!
They also sent me detailed diagrams of tracks, with every conceivable measurement, and even how deep the dirt was at various points.
It was here one tick and gone the next.
Speaking to myself now!
It's not bad, and similar to how I would do it.
I can't remember why they sent to me, perhaps they were partying at the time!
They also sent me detailed diagrams of tracks, with every conceivable measurement, and even how deep the dirt was at various points.
It was here one tick and gone the next.
Speaking to myself now!
StuartH
Filly
I removed them and any reference to your name,my mistake
I was just going to say the ole master Blume himself is here, when Stephen posted his ratio tables
StuartH
Filly
No problem Im made up you are here
You got the old bit right Grant, the rest..... I think people have gone right past whatever it is, I may have done once upon a long time ago.
no i can see a lot of basis from your work into a lot of other peoples work, as I say sometimes you need to stand on the shoulders of greatness to see the way forward, such as it is
When you are establishing standard times for sections my first thought was do them in the same way as you do the standard times for the overall race , you do this and they don’t add up to the overall standard time and I just thought immediately that wasn’t satisfactory, maybe I’m wrong ?
For the overall standards I adjust every time by every horse for class, weight and estimated going allowance with calibration to average 0 going on good going, if you iterate enough times the going allowance changes and the adjusted times group together and the going allowance gets more accurate and you are left with no fast outliers and a list of adjusted times for each CD, when I have these adjusted times I use a low percentile to calculate the standard time maybe 2.5%, then I calculate what the standard time would be if i used say 15%.
Then I look at all the adjusted times I used that were within this 15% (remember they are adjusted for class to it’s not just the fastest times it’s the fastest v expectation) so I now have the top 15% of times adjusted for all factors and I use those to calculate the optimum ratio for each section, from that I can then calculate the standard time for each section or combination of sections as ratio * 2.5% standard.
I don’t think there is a textbook describing how to calculate standard times, so just my way of doing it.
I have a python script now that does all this for me for U.K. tracks when I feed it data , it has all my calculations and does the iterations and reports back the standards times, when I have all the sectional data I will incorporate that into my UK script, but I only started looking at U.K. again recently , more interested in HK.
Last edited:
S StephenJB , you are very open and helpful with what you have. I assume you would have to have the same data structure for your code to be used ? Like I always say with
ARAZI91 and the stuff he posts you should write a book, although I’m not sure if many are interested like they were in the days or Beyer or Mordin now.This has motivated me to get chat gpt to write a python script for my HK standards method.
StuartH
Filly
Last edited: