Anyone know how the 'Percentage Projections' work?

[Dave Schwartz]

Originally Posted by classhandicapper

Dave,

We can ask the same question about the Win odds, the ML or any handicapper's odds lines.

In fact, I'd argue that most people's odds lines are wrong more often than the real odds and sometimes ridiculously wrong. That was one of the points of my other thread about how people determine when they actually have a good value bet.

Exactly right.

The only way to validate probabilities is in the performance over a large sample.

To do it right, you'd look at all the probabilities from 0 to 100% and look at the error at each range. Then do a regression to determine how accurate each of those ranges were.

Of course, that would take a massive sampling.

They are still important, of course.
After all, they're the basis of predicting $Net (or ROI, if you prefer).

But without some validation process, they are just numbers.

And doing without them - just saying "I think #4 is a good bet" - is certainly much worse. (But even that can be calibrated if someone cares to do the work.)

[mike_123_ca]

I suspect they are displaying AI analytics (such as AWS) during the race. You see this during football games where they show the probability of success as the game progresses. Wouldn't it be great to have access to that data while betting is still open?

[thaskalos]

Originally Posted by mike_123_ca

I suspect they are displaying AI analytics (such as AWS) during the race. You see this during football games where they show the probability of success as the game progresses. Wouldn't it be great to have access to that data while betting is still open?

I agree with you, these projection ratings have come about because of the live betting that is taking place with team sports. The difference is that no sports team intends to be 10 points behind during a game...whereas certain horses fully intend to be 10 lengths behind during a race. So...I don't know how accurate these projection methods can be when applied to horse racing. Of course, this is a moot point...because we can't bet into these inner-race projections anyway.

[Robert Fischer]

Originally Posted by thaskalos

I agree with you, these projection ratings have come about because of the live betting that is taking place with team sports. The difference is that no sports team intends to be 10 points behind during a game...whereas certain horses fully intend to be 10 lengths behind during a race. So...I don't know how accurate these projection methods can be when applied to horse racing. Of course, this is a moot point...because we can't bet into these inner-race projections anyway.

1.)


2.)



Horse from the OP Continentalcongres ran 1/14 Sunday (yesterday)...
Not his best performance.

The My Flicker was best and got a good ride and trip.

Credit to 'AI' for realizing that the won the race once he got through on the rail.

I posted 2 screenshots because in the first screenshot; 'A.I' had as most likely, & in second screenshot you can see A.I. adjust and basically award the victory... for trip handicappers you can also see the (Green Cap, Center) happen to enter the stretch a step or two before switching right lead.
So now at this point there's a small sample but we can see that the tends to peak a bit prematurely.

[Dave Schwartz]

Thask,

So, how COULD AI-based analytics be helpful - assuming they were available before the race began?

I don't mean how could it be ACCOMPLISHED. I mean what should it look like?

I've already produced such an engine.
Working on extracting things from it now.
(And adding new things for it to discover and understand.)

[denniswilliams]

I see this working one of two ways:


1) use past races to analyze running style and speed/pace for each horse and then run race simulations accounting for every possible combination of speed/pace/flow -- you have percentages for each race flow possibility (and there'd be many)



2) train a model on race replays -- it'd be able to assess win probabilities at different points in the race given the flow/position/ride, etc. -- not deliverable before the race commences

[thaskalos]

Originally Posted by Dave Schwartz

Thask,

So, how COULD AI-based analytics be helpful - assuming they were available before the race began?

I don't mean how could it be ACCOMPLISHED. I mean what should it look like?

I've already produced such an engine.
Working on extracting things from it now.
(And adding new things for it to discover and understand.)

Before the race begins, the projected winning percentages must be in accordance to whatever pre-race prognostications the AI engine has come up with by way of its "handicapping" of the race. And once the gate opens, these projections would fluctuate to reflect the influence that the dynamics of the given race has on the individual horses in the field. If a front runner secures a comfortable early lead, then this horse's winning projection would be elevated...whereas its winning chances would drastically decline if early speed competition were to materialize at the half-mile marker, etc.

The problem is that this sort of technological advancement is ill-suited to horse racing, because the game moves too fast for these percentage projections to be acted upon, even if they were somehow made available to the betting public.

[Dave Schwartz]

Originally Posted by thaskalos

Before the race begins, the projected winning percentages must be in accordance to whatever pre-race prognostications the AI engine has come up with by way of its "handicapping" of the race. And once the gate opens, these projections would fluctuate to reflect the influence that the dynamics of the given race has on the individual horses in the field. If a front runner secures a comfortable early lead, then this horse's winning projection would be elevated...whereas its winning chances would drastically decline if early speed competition were to materialize at the half-mile marker, etc.

The problem is that this sort of technological advancement is ill-suited to horse racing, because the game moves too fast for these percentage projections to be acted upon, even if they were somehow made available to the betting public.

I just don't see any advantage in this beyond entertainment value.

It would seem that anything AFTER THE GATE has no value.

What am I missing?

[thaskalos]

Originally Posted by Dave Schwartz

I just don't see any advantage in this beyond entertainment value.

It would seem that anything AFTER THE GATE has no value.

What am I missing?

Here's what I think. Laurel has seen that these projection percentages are used in team sporting events, and they have implemented them in horse racing too, hoping that this will stimulate some interest among their younger audience...even if these projections can only offer slight "entertainment value" to the viewer in this context. When you have nothing more substantial to offer...you start clutching at straws.

[mountainman]

Originally Posted by Robert Fischer

48-32!

I'm often better at a Postmortem Analysis, than I am at Projection Forecast/Projection.

Me too. And I more enjoy summarizing a horserace than previewing it. I've occasionally taken criticism for expending significant time on recaps, but while pre-race analysis sometimes falls into familiar patterns, an actual contest often can lead commentary in unexpected and challenging directions. Multiple directions.

Underscoring this, I'm nearly always certain when the camera comes up about the tone and content a given pre-race will take. But many times after a field crosses the wire, an overflow of post-race thoughts I'm irresistibly compelled to get across completely floods my mind.

Also, I just never know when enough is enough. In any aspect of life. So there's that.

Mahoning cancelled for tomorrow , btw.

[Jeff P]

Originally Posted by mountainman

Me too. And I more enjoy summarizing a horserace than previewing it. I've occasionally taken criticism for expending significant time on recaps, but while pre-race analysis sometimes falls into familiar patterns, an actual contest often can lead commentary in unexpected and challenging directions. Multiple directions.

Underscoring this, I'm nearly always certain when the camera comes up about the tone and content a given pre-race will take. But many times after a field crosses the wire, an overflow of post-race thoughts I'm irresistibly compelled to get across completely floods my mind.

Also, I just never know when enough is enough. In any aspect of life. So there's that.

Mahoning cancelled for tomorrow , btw.

Re: the bolded parts --

I played Mountaineer this past season as much as free time allowed.

Want you to know I very much enjoyed hearing your thoughts - both pre-race and post-race.

Being on track four or five days a week like I was some 30 years ago gives one a certain perspective.

Hearing you talk brought back memories of that. Especially when you touched on certain areas of the game a lot of on air hosts never acknowledge.

I think you being real played a part in me making an effort to play your track whenever possible.

I can't be the only one out there.

Ignore your critics. Especially anyone criticizing the pool table story!


-jp
.

[mountainman]

Originally Posted by Jeff P

Re: the bolded parts --

I played Mountaineer this past season as much as free time allowed.

Want you to know I very much enjoyed hearing your thoughts - both pre-race and post-race.

Being on track four or five days a week like I was some 30 years ago gives one a certain perspective.

Hearing you talk brought back memories of that. Especially when you touched on certain areas of the game a lot of on air hosts never acknowledge.

I think you being real played a part in me making an effort to play your track whenever possible.

I can't be the only one out there.

Ignore your critics. Especially anyone criticizing the pool table story!


-jp
.

Tx, Jeff. This means so much coming from you.

As to the trackside experience, on most of their racing days I drive 55 min to Mahoning rather than simply betting from the simulcast theatre less than 10 min away.

Snow..frigid temps..bracing winds encountered along the way only sweeten the feeling of deja vu. And hooking up again with old pals I hung out with decades ago helps remind me of who I am and somehow makes me whole. Heck..it's even BETTER than rediscovering a long lost love.

Believe me, Thomas Wolfe was wrong. Sometimes you really CAN go home again. When I'm amongst guys that refer to me as "Markie," I know I am there.

[classhandicapper]

If you start with current odds, current position, point in race you are at now I think you'll already be pretty close to a good answer.

If you add in a projected running style/position you should be very close.

For example, if a horse is 1/5 and is 10th of 10 after the 1st quarter, it's probably not still 1/5, but you wouldn't want to downgrade it too much yet. It's too early in the race.

If the horse was projected to be on or near the lead early, you'd downgrade it more.

If the horse is Zenyatta, she would still be 1/5 after the first quarter.

Then as the race develops position gets more and more important in the metric and odds less.

To take it further, you might want to look at the other contenders and compare their relative positions. The 2nd choice in solid position looks a lot better when the favorite is being outrun early instead of right there with him.

[rastajenk]

I think I've read everything here in this topic...I'm surprised it's been stepped around but not mentioned...isn't this a test for in-race wagering? Sports betting has in-game action and cash out options; this is a response, is it not? Somehow. Personally, it holds no appeal for me, and I hope it doesn't get legs. But that's what it's all about, isn't it? I don't think it's just another metric to fine-tune your pace-based analyses.

On further review, Onesome mentions in-race betting in post 2.

[ubercapper]

I received this from TPD


TPD In Play Odds.jpg