How to determine a track bias

[DeltaLover]

Originally Posted by EMD4ME

Hey Delta,

I have date specific data which is what is needed, you know me.

4/17-rail was dead races 6-10. Race 6, the rail was the speed and the favorite, dueled inside and quit. Outside dueler won it. Race 7, PP 10 won with an outside trip. Race 8, yet again outside stalker won. Race 9-all outside PPs ran best. Race 10-chalk got to lead from pp 1, lost while racing inside.

Post 1 was 0/5 in that biased 1/2 day. (I had the rail slightly negative for races 1-5 FYI).

Track was biased not because PP 1 lost. The rail was dead because an intense, experienced Emerald player who knows all the horses inside out, realized that there too many incidents of solid horses who were racing inside, underperforming.

I can list the next 20-30 racing days but I don't want this post to run for 3 pages.

Those PP stats can be extremely misleading. First, we're talking about short term trends which are the most valuable to us. Why?

1) If we catch on to a bias before others, we can bet accordingly
2) If we catch these anamolous days and document properly, we can know more than Joe Schmo the next time each runner comes back. Who's performance was bias aided, who was killed by a bias.

One will never detect short term biases via looking at long term trends. (For example Post Position stats). Over the course of a meet, there will be gold rail days & dead rail days (and other types of biases) and the Post Position stats aren't definitive enough to point out which day had what bias. In the long run, they might even eachother out (but there are many variables to that-better horses in certain positions-lopsided amount of dead rail days etc.)

Finally, as most know, horses can break from any post and run in any paths. PP 1 could have no speed, be taken wide immediately and run 4 wide, 5 wide, 6 wide throughout. PP 10 can tuck and ride the rail for 75% of a race etc.

Not really talking to you as I know you know your stuff. Just replying in general to PA Nation.

good post and food for thought

[BCOURTNEY]

Originally Posted by EMD4ME

Maybe it's an omen, , but Courtney's Hero is running in the next race at CBY from PP 5.

Should we bet???

First time starter at long odds!

Statistically Horses with a name containing 'emd' are strongly negative this year so far:
bias
0.7207

Statistically Horses with a name containing 'courtney' are VERY strongly negative this year so far:
bias
0.0000 (no winners from 16 starts)

Update: It appears Courtney's Hero - was not.

[BCOURTNEY]

Originally Posted by Cratos

A good observation and you are correct, because banking of the turn changes the normal force on the horse by the cosine of the banking angle and this is significant given the inherent side force on the turn.

Reminds me of my early days in the late 80's and 90's with neural networks and greyhound racing, all about the banking turns. Now there is a racing paradigm with obvious bias.

[VigorsTheGrey]

Originally Posted by DeltaLover

good post and food for thought

Instead of post position bias maybe it should be path bias....like EMD said, any horse can break from any post and run in any path...the can break clean and drop to the rail immediately effectively make it the post position wise...so what use are post position stats really?

What is needed is running path stats and ways to index and cross reference the history of each horse with regard to the path it ran in for x-portion of the race....EMD's race notes on steroids is the prototype of this kind of record keeping...As Thaskalos says, that losers often know what to do, but don't apply what they know in order to reduce the negative impact of making losing wagers....

EMD's trip notes and record keeping, massive video time, and general overall familiarity with the current racing stock and jockey/ trainer colony, provide a gigantor-enormous edge compared with the regular player who might only do maybe one of these if at all...

...If EMD can convert his post position notes to path notes and somehow index the horses with the paths....all that may be needed is to have some clear idea of which horses can make it to which paths at this or that stage of the race....

EMD has already suggested that at Emerald Downs, for example, horses that make the front at the top of the stretch have a great chance for winning...understanding if any giving horse can make the front at this point in the race would necessarily involve a knowledge of which paths might become available and when, for which horses...

[DeltaLover]

Originally Posted by VigorsTheGrey

Instead of post position bias maybe it should be path bias....like you say any horse can break from any post and run in any path...the can break clean and drop to the rail immediately effectively make it the post position wise...so what use are post position stats really?

What is needed is running path stats and ways to index and cross reference the history of each horse with regard to the path it ran in for x-portion of the race....EMD's race notes on steroids is the prototype of this kind of record keeping...As Thaskalos says, that losers often know what to do, but don't apply what they know in order to reduce the negative impact of making losing wagers....

EMD's trip notes and record keeping, massive video time, and general overall familiarity with the current racing stock and jockey/ trainer colony, provide a gigantor-enormous edge compared with the regular player who might only do maybe one of these if at all...

...If EMD can convert his post position notes to path notes and somehow index the horses with the paths....all that may be needed is to have some clear idea of which horses can make it to which paths at this or that stage of the race....

EMD has already suggested that at Emerald Downs, for example, horses that make the front at the top of the stretch have a great chance for winning...understanding if any giving horse can make the front at this point in the race would necessarily involve a knowledge of which paths might become available and when, for which horses...

Based on what I have seen, when it comes to trip notes EMD is the best without a question.

[BCOURTNEY]

Originally Posted by DeltaLover

Based on what I have seen, when it comes to trip notes EMD is the best without a question.

Great if he is willing to post a month worth of notes I will figure out how to automate it, let's all get rich.

[Cratos]

Originally Posted by VigorsTheGrey

Instead of post position bias maybe it should be path bias....like EMD said, any horse can break from any post and run in any path...the can break clean and drop to the rail immediately effectively make it the post position wise...so what use are post position stats really?

What is needed is running path stats and ways to index and cross reference the history of each horse with regard to the path it ran in for x-portion of the race....EMD's race notes on steroids is the prototype of this kind of record keeping...As Thaskalos says, that losers often know what to do, but don't apply what they know in order to reduce the negative impact of making losing wagers....

EMD's trip notes and record keeping, massive video time, and general overall familiarity with the current racing stock and jockey/ trainer colony, provide a gigantor-enormous edge compared with the regular player who might only do maybe one of these if at all...

...If EMD can convert his post position notes to path notes and somehow index the horses with the paths....all that may be needed is to have some clear idea of which horses can make it to which paths at this or that stage of the race....

EMD has already suggested that at Emerald Downs, for example, horses that make the front at the top of the stretch have a great chance for winning...understanding if any giving horse can make the front at this point in the race would necessarily involve a knowledge of which paths might become available and when, for which horses...

We do what I think that you are suggesting.

We use Trakus data and use their given data points of distance from the rail and distance at each POC from the start for each horse in the race and this gives the displacement of the horse at point of intersection from point to point in the race.

Essentially we are conducting vector analysis with tangent vectors along the downward sloping nonlinear race curve.

However we only do this for the NYRA circuit.

[BCOURTNEY]

Originally Posted by Cratos

Everyone one on this forum I assume goes first with their learned or innate quantitative abilities when comes to their handicapping of racehorses; practicality is about understanding not about simplification.

Actually I was thinking most everyone posting in this thread goes first with their qualitative assessment abilities.

Is the problem we are trying to solve here is this a classification or regression problem? What is the intended output(s)?

Based on you posts I saw cleansed time as the intended output.

[Cratos]

Originally Posted by BCOURTNEY

Actually I was thinking most everyone posting in this thread goes first with their qualitative assessment abilities.

Is the problem we are trying to solve here is this a classification or regression problem? What is the intended output(s)?

Based on you posts I saw cleansed time as the intended output.

What we attempt to do is bring each horse back to its “clean final time” (as you put it) and ask the question: “If Horse A ran against Horse B under the same conditions (i.e., distance and class); what would be the outcome?”

[VigorsTheGrey]

Originally Posted by BCOURTNEY

Actually I was thinking most everyone posting in this thread goes first with their qualitative assessment abilities.

Is the problem we are trying to solve here is this a classification or regression problem? What is the intended output(s)?

Based on you posts I saw cleansed time as the intended output.

This idea of cleansed time is new to me but I'm sure it is not new for all you pros...put another way, is the output you are seeking "the best representative absolute (cleansed) race time for each runner superimposed onto one another to create moving races shapes that terminate at the finish line with a projected order of finish.". My quote.

[Tom]

My conclusions are Pandy is the only one who knows what he is doing.
He is watching how races are run and using the information to help him make selections.

Thanks for clarity, Pandy!

[mountainman]

Originally Posted by EMD4ME

I don't bet the Mountain often, I should but I want to give SUPER credit to Mark for this:

http://www.moreatmountaineer.com/racing/track-bias


Awesome work Mark, awesome.


Jason Blewitt compiles this list, good man!

https://www.nyra.com/saratoga/racing/track-trends

tx, pal..but, although the bias notes were my idea, nancy took them over (at my behest) many months ago when I almost died from the gall bladder thing.

I say this not only to give her credit, but because I would (and did) do them a bit differently, and would rather be associated with my own writing style. She's good, but we are very very different in approach, substance, and style.

And we often disagree on bias trends.

I'll pass along the props, and tell her they come from a very sharp guy.

[mountainman]

I've seen formulas for expressing track-bias-as pertains to running-style-in numeric fashion. But none of these methods take odds or race-flow into account. In other words, a 20-1 that wins after a prolonged speed-battle should be more assigned more significance than a chalk that clears easily and wires.

Just for fun, I've tinkered with devising my own formula, but prefer to use human judgement. And any such formula should have basis in voluminous research performed by geeks that establishes, for instance, what the percentage chance is of a front-running win in a two-turn mile involving an extended speed-duel between two chalks...etc...etc...etc....

[rsetup]

Originally Posted by mountainman

I've seen formulas for expressing track-bias-as pertains to running-style-in numeric fashion. But none of these methods take race-flow into account.

Setups are not captured by speed or pace numbers. Thought this would never catch on.

[cj]

Originally Posted by mountainman

I've seen formulas for expressing track-bias-as pertains to running-style-in numeric fashion. But none of these methods take odds or race-flow into account. In other words, a 20-1 that wins after a prolonged speed-battle should be more assigned more significance than a chalk that clears easily and wires.

The TimeformUS rating takes these into account...odds, pace, and race flow. We certainly don't count a 3-5 winning the same as a 15-1, and a horse setting dawdling fractions and winning doesn't count nearly as much as one setting fast fractions and hanging on, for example. Same goes for a dueler as opposed to a clear lead.