Value of Modeling

[For The Lead]

I took the last 97 races ( it just worked out that way) at PEN and did some checking.

My #1 TPR horse won at a rate of 27% and returned $2.68 for each $2.00 wagered, an ROI of 34%. Not bad for single horse betting. And that is an ‘overall view’ of how PEN is playing since TPR was the best factor. Now if you are happy with that, you don’t need to do anything else. But, because it is my usual “MO”, I took a closer look, which means I broke it down further.

The first thing I did was look at “females” vs. “males”. Here’s what I found.
#1 TPR for females returned $2.45 for each $2.00 wagered, still not too bad, but not as good as the overall.
#1 TPR for males returned $2.90 for each $2.00 wagered. Now this is MUCH BETTER than the overall. But let’s take an even closer look.

Any maiden race ( maiden claiming or maiden special weight) returned $0.86 for each $2.00 wagered. A negative -57% ROI. WOW! I guess you wouldn’t want to bet those races, huh? There were 25 of these races, so let’s take them out and see what happens.

Now, 37 races for females that are not maidens returned $3.08, a +54% ROI and a 32% win rate.
Now, 35 races for males that are not maidens returned $3.55, a +77.5% ROI and a 31% win rate.
And all I did was a little research and found 25 races that were dragging down profits.

This is the power of breaking down models. And if you are playing a track with a long meet you can break them down even further, as an example, by distance, restrictions, etc. Doing so really helps the bottom line!

Now just in case you were wondering, the #1 TPR at MNR, over the same period of time, returned a PALTRY $ 1.22 for every $2.00 wagered. A negative ROI of -39%. Although there are good factors at MNR, as you can see, #1 TPR isn’t one of them.

As Tim Y said in a previous post, "each track has its' own footprint".
In my opinion, it's just a matter of how deep that footprint goes.

[mikesal57]

Excellent FTL!!

This is what modeling is all about...if you have the patients and time..it will payoff in the long run..and for a track that is open all but 2 weeks of the year you can have a nice little profit. I'm sure you can break it down even further , maybe distance ,track condition , class etc. etc. Also there might be a certain category that gives you longshot plays too :rolleyes:
The only question that I have is what paceline criteria do u use, best of last 3?

thxs
mike

[gl45]

The modeling is based on pace lines selection. How sure we are that our selection will run based on the model at hand. Trying to model using the result chart, will give you better results.

[mikesal57]

True Pino...
is there a way to incorporate the actual pace of the race into the model?

take for example
ARL 8.5 furs

you find out the best LPR rating is winning...

but the average pace of the race is something like 47.3----110.4

suppose you predict a pace of 48.1----112.1

the best LPR horse is no where in site now

How do we handle something like this?

mike

[tom]

As Tim Y said in a previous post, "each track has its' own footprint".
In my opinion, it's just a matter of how deep that footprint goes.

Looks like you stumbled onto Bigfoot!
Excellent use of a model!
Not only do you know where your money is coming from, you know where you need to do more work.

Thanks for sharing.

[For The Lead]

Excellent FTL!!


The only question that I have is what paceline criteria do u use, best of last 3?

thxs
mike

No, I do not use EXACTLY “best of last 3 distance/surface”. I guess the first thing I should tell you is, all my lines are computer generated. Naturally, since I wrote the program, it generates lines base on what I told the program to look for and what not to look for and, to me, CONSISTENCY is the key. The lines are ALWAYS generated the same EXACT way! Let me put it this way, here is what I DON’T do.
I don’t mix turf with dirt/poly
I don’t mix dirt/poly with turf
I don’t mix sprint with route
I don’t mix route with sprint
I don’t scan the speed ratings and take the horses lifetime best and use that line
I don’t use lines over 90 days old
I don’t use lines if the horse showed no “sign of life”.( i.e.- NOT 1st or 2nd at the first or second call or NOT in the money finish) I don’t want to use a line where the horse was out for exercise or was not competitive at some point in the race.

Having said all that and taking all that into consideration, what I use is the MOST RECENT line where the horse DID SHOW “signs of life” at the distance/surface it is running at today. It doesn’t necessarily have to be the horses’ BEST race.
So am I close to “best of last 3 distance/surface”? Sort of.:)

[Tim Y]

Modeling has never shown any differences in male/female md clm/MSW

Energy distribution is a physiological aspect of the animals IT JUST IS without over-loading it with some artificial MAN-MADE categorizations.

[Ted Craven]

Modeling has never shown any differences in male/female md clm/MSW

How do you account for the stats described in post #1?

Ted

[For The Lead]

Modeling has never shown any differences in male/female md clm/MSW

Energy distribution is a physiological aspect of the animals IT JUST IS without over-loading it with some artificial MAN-MADE categorizations.

In the case I posted here the males and females were pretty much equal. However, my findings show this is not always the case. In fact, many times they are considerably different.

[Tim Y]

How do you account for the stats described in post #1?

Ted

Apples Oranges as TPR is not energy distribution... NEVER was

[For The Lead]

Apples Oranges as TPR is not energy distribution... NEVER was

Ok, let me take those same 97 races and look at #1 EPR.

Maidens male= 14 races return $0.63. A negative ROI of 66.5%.
Maidens females = 11 races return $0.78. A negative ROI of 61%.
NON maiden males = 35 races return $1.46. A negative ROI of 27%.
NON maiden females = 37 races return $4.15. A POSITIVE ROI of 107.5%

Quite a difference!

[tom]

Apples Oranges as TPR is not energy distribution... NEVER was

He is modeling TPR...you are the only one who mentioned energy distribution.

[mikesal57]

Having said all that and taking all that into consideration, what I use is the MOST RECENT line where the horse DID SHOW “signs of life” at the distance/surface it is running at today. It doesn’t necessarily have to be the horses’ BEST race.
So am I close to “best of last 3 distance/surface”? Sort of.:)

thxs FTL...understood what your saying...

now suppose your program picks a paceline ,say 3 back...do u examine the PP to see if the horse has gone off form...

This is a question that I've thrown around with not much answers , how do u determine if the horse is going off form or coming back to a good race.

mike

[Tim Y]

Ok, let me take those same 97 races and look at #1 EPR.

Maidens male= 14 races return $0.63. A negative ROI of 66.5%.
Maidens females = 11 races return $0.78. A negative ROI of 61%.
NON maiden males = 35 races return $1.46. A negative ROI of 27%.
NON maiden females = 37 races return $4.15. A POSITIVE ROI of 107.5%

Quite a difference!

Energy distribution is, for young horses, is A LEARNED aspect. With babies one cannot use same standards for adults.

Follow the career of ANY horse, sans at the top of the heap, and they always get a bit later as they mature. In one of the first Sartin manuals I ever read, they followed the %median of Secretariat from his first to later races and made it clear that it was a common fact that thye learn to distribute energy better as they LEARN how to run.

[For The Lead]

thxs FTL...understood what your saying...

now suppose your program picks a paceline ,say 3 back...do u examine the PP to see if the horse has gone off form...

This is a question that I've thrown around with not much answers , how do u determine if the horse is going off form or coming back to a good race.

mike

Yes, I check the line. Since I did the programming on this, I am always checking my work to make sure I am getting what I was hoping to get when I wrote the program. :)
Beyond that, if I get a paceline 3 back it is generally because (as an example) today is a 6f dirt and neither the last line nor two back were similar in distance/surface. If I get a paceline 3 back and the last line and the line two back were similar to today's distance/surface, I do check the last two lines. Again, mostly to check my work. Built into my program are some parameters that will skip over races because of differences in class compared to today. Understand that all this programming was in an effort to replicate in an automated way what I would do manually. I had to ask myself, "what would I forgive" in order to go back a line (or two). Granted, this becomes a matter of an individual's observations and opinions.
So, you have a horse with a good race 3 back. At that point the horse was obviously in good shape. In its' last two races, what happened that it did not continue that good form? If the answer is a change in distance and surface, I will forgive it, with the consideration that those races did not hurt the horses form, it just wasn't the type of race at which the horse performs best. Likewise, if the answer is, the horse raced over its' head and is now coming back to a similar class race, I will forgive it. And then there is all the class variations. Here's an example.

Last race = 5k clm OPEN (ran bad)
2 back = 6500 clm NW3L (ran bad)
3 back = 5k clm NW2L (WON race)

In this case I will accept the line 3 back. Keep in mind, that when I say the horse "ran bad", I don't mean that the horse trailed the field. In my opinion, an "in form" horse generally shouldn't trail the field, but if there is a surface change and the horse really doesn't like the surface it is possible. I hope this helps.

[jms62]

I took the last 97 races ( it just worked out that way) at PEN and did some checking.

My #1 TPR horse won at a rate of 27% and returned $2.68 for each $2.00 wagered, an ROI of 34%. Not bad for single horse betting. And that is an ‘overall view’ of how PEN is playing since TPR was the best factor. Now if you are happy with that, you don’t need to do anything else. But, because it is my usual “MO”, I took a closer look, which means I broke it down further.

The first thing I did was look at “females” vs. “males”. Here’s what I found.
#1 TPR for females returned $2.45 for each $2.00 wagered, still not too bad, but not as good as the overall.
#1 TPR for males returned $2.90 for each $2.00 wagered. Now this is MUCH BETTER than the overall. But let’s take an even closer look.

Any maiden race ( maiden claiming or maiden special weight) returned $0.86 for each $2.00 wagered. A negative -57% ROI. WOW! I guess you wouldn’t want to bet those races, huh? There were 25 of these races, so let’s take them out and see what happens.

Now, 37 races for females that are not maidens returned $3.08, a +54% ROI and a 32% win rate.
Now, 35 races for males that are not maidens returned $3.55, a +77.5% ROI and a 31% win rate.
And all I did was a little research and found 25 races that were dragging down profits.

This is the power of breaking down models. And if you are playing a track with a long meet you can break them down even further, as an example, by distance, restrictions, etc. Doing so really helps the bottom line!

Now just in case you were wondering, the #1 TPR at MNR, over the same period of time, returned a PALTRY $ 1.22 for every $2.00 wagered. A negative ROI of -39%. Although there are good factors at MNR, as you can see, #1 TPR isn’t one of them.

As Tim Y said in a previous post, "each track has its' own footprint".
In my opinion, it's just a matter of how deep that footprint goes.


Excellent work and the key to profits..