Creating Factors

[Big Russ]

Looks like the 1,3 and 7 are scratches.

[mikesal57]

Originally Posted by headhawg

If you're talking about CC, PS, and Power I added those to Handifast to create my version (HHX). The X was for eXperimental as I wanted users to know that it was different from the original. I wasn't sure that they had any value, but it was something that I wanted to try.

Those factors are not coded exactly like Mitchell's, btw. The EPS number is based on Quirin's research so it's slightly different, and PS does not use fps. It does use the 60-100 indexing formula from Mitchell's book. IIRC, Mitchell wrote that you could change the range if you so desired but I stuck with his suggestion. Power is weighted the same way as in the book.

Glad -- and somewhat surprised -- that those are holding up as being usable factors.

THEY ARE !!!

need to find some more....

[mikesal57]

Originally Posted by Big Russ

Looks like the 1,3 and 7 are scratches.

crap...there goes that one...

[CincyHorseplayer]

Originally Posted by mikesal57

CINCY...

Off the top of my head....and not looking at your approach

Your right on this giving an ability rating ...but its not giving the early horses ability once you add speed fig and divide by 3...

Why I say...cause horses are creatures of habit...
they will run where they are comfortable and in a position where they win from...also, most important....they will win when the fractions are in their favor.
Too fast , they lose...slow...they win..

I'm trying to figure out how to compare winning situations compared to losing situations using E1 + E2...
Maybe theres not a way but maybe someone can help

Mike

You are right. My EP ranking is E1 + E2 straight up. But My ability rating is as described. The former was developed among and for some sad performers. The latter for horses that have actually won. That rating is still relevant in the most important races where pace is devoid. Like I said it is a staple in my handicapping.

[CincyHorseplayer]

Originally Posted by headhawg

Great questions, Mike, and likely one of the most important for a software developer -- what factors really work? If you have a large database then you could test things yourself but you would need some knowledge of statistical methods (probably some form of regression) to know if the factor is a real predictor or just some random blip. Statistics is a fascinating topic but it's mostly over my head except in it's most basic forms.

In lieu of that you could use things that others have researched. A good place to start would be Dave Schwartz's Percentages and Probabilities. It's an updated version of Quirin's research from the 70s. I am working on my own software and using some of the things that I learned from P&P as a base. And it's currently on sale. (Dave, you can send me the commission check later. ) P&P is a few years old now but I doubt that the IVs and $Nets have changed substantially. My feeling is that the high IV factors have gotten even stronger but the $Nets have probably decreased somewhat.

And this thread might provide some good ideas to test, and I emphasize test. Some ideas might seem logical and that they will work but won't hold up under scrutiny. And it's not likely that someone will give away their secret factors.

A large database only solves some problems. Which while statisticians and mathematicians give us the tools they can't grasp what the numbers mean in the game in many circumstances. I developed my theories from small samples and intuition. Mostly because that is how meets are conducted and change from year to year. I am baffled plenty! A track plays a certain way or horses perform a certain way on a certain surface.You try to express that through a formula derived from pace numbers. It is an ongoing theoretical exercise. In my opinion the big database is the generic collective psychology. But the intuition based ratings are much more the Jungian emphasis on individual psychology. We are dealing with individual horses and individual tracks and imo individual theoretical experience. No database is going to unconvince me of what I am doing. Many players what I am talking about.

[mikesal57]

Originally Posted by CincyHorseplayer

A large database only solves some problems. Which while statisticians and mathematicians give us the tools they can't grasp what the numbers mean in the game in many circumstances. I developed my theories from small samples and intuition. Mostly because that is how meets are conducted and change from year to year. I am baffled plenty! A track plays a certain way or horses perform a certain way on a certain surface.You try to express that through a formula derived from pace numbers. It is an ongoing theoretical exercise. In my opinion the big database is the generic collective psychology. But the intuition based ratings are much more the Jungian emphasis on individual psychology. We are dealing with individual horses and individual tracks and imo individual theoretical experience. No database is going to unconvince me of what I am doing. Many players what I am talking about.

Yes....and you have only 3 pace figures to work with...How much can you really get from that and that others have manipulated over the years???

And what if , these figs are off , like CJ said a long time ago, then your screwed from the start....

I was thinking about Daves new Par Book......its in csv format and can easily be imported into a program...it has actual numbers from all tracks that you can use and make your own figs...

Has anyone used it in that sense?

Mike

[headhawg]

Several years ago I bought Dave's pars with the express purpose of doing just that, Mike. I never got around to doing anything with them, the whole "life got in the way" scenario. Now that I have more time I am considering getting back to it, but probably not until later on this year. Currently working on other stuff.

[mikesal57]

Originally Posted by headhawg

Several years ago I bought Dave's pars with the express purpose of doing just that, Mike. I never got around to doing anything with them, the whole "life got in the way" scenario. Now that I have more time I am considering getting back to it, but probably not until later on this year. Currently working on other stuff.

I hear ya man...

When i have days off from work and even at work , I would say I think of things 80% of the time...(yes, my boring life at 61)....lol
then when its time to actually put it down in code......something else pops up
So I say , take one thing at a time and FINISH it before life changes again...

Mike

[Tom]

I am one of Dave's loyal customers. Here's what I do:

1. Download the races into Excel. (I use HTR and get both charts and a track profile I can use for this.)

2. Using Beyer's chart from his book, I first verify the Beyer equivalent in time - not all distances/track follow the chart exactly. I take a years of data and check out how each distance lines and make a correct if I find any out of line. Aqueduct, for example, is off 6 points at 6.5 furlongs.

3. I then add a column for Beyer/Quiring, where I convert each Beyer to a Quirn speed figure, based on a Beyer of 80 = a Qurrin 100.

4. I then use Dave's pars t convert the raw times to pace figures, all three calls.

5. Now, the par for the race is the Beyer/Quirrin number. So the race variant is the difference between that number and the raw Qurrin number I get from Dave, and the pace variant are the difference between the race Quirrins and the Beyer/Quirin.

6. I can now look a the whole day and see if Beyer has any numbers he broke out, re-adjusted for some reason, whatever.

7. Any races out of the norm are marked so I know later when I used them that this number is not to be taken as set in stone.

8. I then calculate a pace variant for each call.

9. Last three columns are the race pace and speed figs.

10. If I have a par for the race, I then add the race shape from Quirrin's books, AF, AA, FF, or for routes, AAF, SAA, etc.


I check the Winner's Books every month and see if Beyer has changed any numbers since I first got them - this happens more than occasionally. I also make this check annually to get any track to track adjustment he might make. I just did Finger Lakes for 2017 this week in preparation for the opening today. Any changes are put into the sheet and the variants are re-calculated for those days.

I get to see the whole day doing it this way - I see if any races were exceptionally fast or slow in the raw times.
It is not unusual to see some VERY slow first fractions at NYRA.

It is not just entering Dave's stuff into a program a lot of hand-entry is need to set up the spreadsheet, but then it is just add new race data and dragging the formulas to get Quirrin numbers into place.

[jay68802]

Originally Posted by Tom

I am one of Dave's loyal customers. Here's what I do:

Quiring, Quirn, Quirrin

Are you using 4 different figure's, or just posting under the influence?

I do something very similar with Briss and Quiring, Quirn, Quirrin figures. This way I can look at both the pace and final figures for comparison.

[Tom]

When you use 4 figures, you can say you had the all 9 winners in the top 4! I can box a super using only the top fig horse!


You convert BRIS to Quirin - I have a chart I set up for that - not too bad actually.

[jay68802]

Originally Posted by Tom

When you use 4 figures, you can say you had the all 9 winners in the top 4! I can box a super using only the top fig horse!


You convert BRIS to Quirin - I have a chart I set up for that - not too bad actually.

http://www.youtube.com/watch?v=sShMA85pv8M

[CincyHorseplayer]

Originally Posted by mikesal57

Yes....and you have only 3 pace figures to work with...How much can you really get from that and that others have manipulated over the years???

And what if , these figs are off , like CJ said a long time ago, then your screwed from the start....

I was thinking about Daves new Par Book......its in csv format and can easily be imported into a program...it has actual numbers from all tracks that you can use and make your own figs...

Has anyone used it in that sense?

Mike

There are 3 fractions and E2. I use sabermetric concepts with these numbers. Much of it derived from reading Brohamer's book. Some of their ratings I disagreed with so I started altering them. Then reading Beyer's discussion on the difference in nature of different racing surfaces I started to apply what I think are the general laws of those surfaces to pace figures. You'd be surprised how far this can go. From 2005 to 2014 it evolved in 9 ratings and the formulas changed. I haven't changed them since. Once you start testing these statistically and see when and where they work they are validated. To be involved in this scrutiny of pace numbers is better than to not be. It's better than simply using speed figures.