Role of ML in age of CAW

[Aerocraft67]

We all know two separate critical assessments of a race. What's the horses' chances, and how will they get bet.

We've established that CAWs get the last actionable look at the tote and effectively determine how they're bet. This can be a much different value than the last actionable look from regular bettors.

Not only do regular bettors have to asses the chances of runners, they have to assess what their odds will be. They have to make a guess at values that CAWs will know better.

At face value, this is a job for the morning line. Do today's ML makers use today's tools to assess how the runners will be bet? If not, what good is another human assessment of that when computers are driving the result?

Most of us respect ML makers. Not that it ever was this easy, but we're way past seeing blinkers on or whatever and surmising that Joe Blow is going to overbet that. I hate to think that track management hands its ML employees pencil and paper to compete with CAWs on which the same track management is lavishing all the advantages.

I'm not demanding that the ML beat the CAWs at their own game. But after all our handwringing about what's to be done about the scandalously close and preferential relationship between track management and CAWs at the expense of the majority of human customers, providing a ML aided by similar methods as the CAWs would seem to be a very practical way to restore the gap in playability, not to mention fairness.

CAW apologists tout the kind service they do for us in making the pools efficient. If that's truly the case, then they should apply the golden rule and share that efficient assessment before the race is run. At least an entry level assessment, one that could be aided by human interpolation.

Beyond a hapless demand of the industry, my more practical question is, how much does the ML incorporate the same factors that drive CAWs to bet how they do to determine the final odds? How effective is today's ML?

[Tom]

I would guess the ML would be far too erratic track to track to be a value. Some lines are plain garbage.

Be intereste to hear other views.

[Twin Double]

Originally Posted by Aerocraft67

We all know two separate critical assessments of a race. What's the horses' chances, and how will they get bet.

We've established that CAWs get the last actionable look at the tote and effectively determine how they're bet. This can be a much different value than the last actionable look from regular bettors.

Not only do regular bettors have to asses the chances of runners, they have to assess what their odds will be. They have to make a guess at values that CAWs will know better.

At face value, this is a job for the morning line. Do today's ML makers use today's tools to assess how the runners will be bet? If not, what good is another human assessment of that when computers are driving the result?

Most of us respect ML makers. Not that it ever was this easy, but we're way past seeing blinkers on or whatever and surmising that Joe Blow is going to overbet that. I hate to think that track management hands its ML employees pencil and paper to compete with CAWs on which the same track management is lavishing all the advantages.

I'm not demanding that the ML beat the CAWs at their own game. But after all our handwringing about what's to be done about the scandalously close and preferential relationship between track management and CAWs at the expense of the majority of human customers, providing a ML aided by similar methods as the CAWs would seem to be a very practical way to restore the gap in playability, not to mention fairness.

CAW apologists tout the kind service they do for us in making the pools efficient. If that's truly the case, then they should apply the golden rule and share that efficient assessment before the race is run. At least an entry level assessment, one that could be aided by human interpolation.

Beyond a hapless demand of the industry, my more practical question is, how much does the ML incorporate the same factors that drive CAWs to bet how they do to determine the final odds? How effective is today's ML?

I see this statement a lot in articles and TV talking heads. Always remember whoever says this either A - Doesn't understand professional advantage gambling or B - is intentionally trying to mislead.

If pools were always 100% efficient it would be stupid to even play the game as a retail player. All gambling comes down to a basic (or complex) EV equation. Efficient pools mean your expectation will always end up around the track takeout. In the long run of course.

On the flip side, efficient pools would help the very casual player from having an expectation worse than the track takeout which could easily happen with inefficient pools. They are destined to lose either way so this protects them a bit.

[Dave Schwartz]

I've built a rudimentary model using a handful of factors available in the HDW system and some that are proprietary to HSH.

They are (in order of correlation to final odds):

1. Morning Line
2. PSR (Cramer's Projected Speed Rating)
3. CPWR (Older version of PSR)
4. RTG (My proprietary Rating)
5. cFT (Composite Final Time - lots of Speed Ratings)
6. TrTrk (Trainer Standings at the track)

#2, #3 and #4 are all similar to BRIS Prime Power.
(All get 30-32% winners on horses ranked 1st.)

The whales certainly have stuff like this.
They also have better weighting than I use because I am using a simple Fibonacci sequence.

This Object does an excellent job of pointing to the horses that should be bet.

So much so that I've been able to fashion strategies from and around it.

In addition, the whales also have tote information to utilize.
They have both historic tote movement and minute-by-minute in the current race.

To be clear, while this object is an excellent predictor of the tote, there are several gyrations that must be done to find a useful strategy from this.

That's because knowing who SHOULD BE BET is far different than knowing who WILL BE BET.

Hope this helps someone.

[thaskalos]

Originally Posted by Tom

I would guess the ML would be far too erratic track to track to be a value. Some lines are plain garbage.

Be intereste to hear other views.

This mirrors my experience exactly.

[Nitro]

I am personally so tired hearing about all the negativity surrounding the CAW type player and how they might impact the game. So, I decided to check it out further and away from all the bias and apparent inaccuracies being made about them here on this forum.

For those of you who might want some real clarification to better understand the realities of this type of player I would highly recommend reading “Horseracing’s true lifeline”.
https://pastthewire.com/horseracings-true-lifeline/

You just might learn something. I know I did.
,
.

[Aerocraft67]

Originally Posted by Aerocraft67

I hate to think that track management hands its ML employees pencil and paper to compete with CAWs on which the same track management is lavishing all the [digital] advantages.

Just an attempt to keep this on topic. That quote [and edit] is really the thrust of my concern and inquiry about the morning line (ML).

I know there are publicly-provided models like BRIS Prime Power that serve as proxies for a morning line, or at least what a basic computer model will point to. Tom and Thaskalos pointed out the inconsistency of lines, so that's an advantage of a Prime Power, which is at least consistent across tracks, putting aside its accuracy.

If you get to know a linemaker, that can lend some consistency, and some ways to "translate" the line into something more accurate. Better yet, the linemaker also serves as an analyst. But are they just applying their formidable wisdom to the line, or are they weighing properly the digital portion with digital tools, with forthright assistance from their employers?

I have a basic grasp of the bias in a morning line—no one wants to admit a highly likely 3/5 favorite lest it discourage handle. No need for pinpoint accuracy above 20-1 or so. That's a little different than my inquiry, but they're relevant ML factors. Aren't we just past that now?

I also understand the track isn't privy to all the tools and methods of CAWs. But tracks absolutely have all the wagering data. Do tracks allow the linemaker to use the same data that the CAWs get? Shouldn't that data point to a pretty good morning line, even with modest digital tools?

Anyone care to share an accounting of ML accuracy and whether its accuracy has changed over time corresponding with CAW participation?

[Nitro]

Anyone can reach into their bag of skepticism and belittle the Morning lines and those who produce them. These are the same people who claim to be able to generate their own lines based on subjective interpretations of PP data and perhaps statistical information to boot. So, if you don’t understand the reasoning behind any M/L just use your own line if you think its superior.

That’s obviously what the CAW’s are doing. Do you think for a minute that they care about what the original M/L might be?

Apparently, what many can’t seem to acknowledge is the fact that they have the tools and resources to provide a much higher degree of predictability than the average handicapper. They better have or they won’t be playing at the CAW level very long. No, what they do is (what everyone else who creates a line should do) make comparisons between what they have and what the tote presents during a typical betting cycle. Why? Obviously to discover flaws in the actual betting and even more importantly to see if there’s real value to take advantage of among those possible short comings.

BTW the idea that the CAW’s have the last look at the tote information is another misnomer. As even confirmed by the CAW being interviewed in article I posted (above). I get my tote information every 30 seconds until the final accounting is complete. They certainly don’t have any advantage there. When utilizing a proficient computerized tool like a tote analysis the betting patterns (regardless of the odds) are of primary interest.

As far as the basic M/L’s go I’ve found over the years that the more accurate ones seemed to be derived from the tracks with the larger handles. But no matter which track you’re playing if there’s a single or multiple scratches in a race you can consider both the M/L and your own line to be compromised.

[biggestal99]

which bettor pays attention to human morning lines nowadays.

we live in the age of AI

i like AI morning lines much better than a biased human morning line.

AI morning line WAaaaaaay better.

Allan

[cj]

Originally Posted by biggestal99

which bettor pays attention to human morning lines nowadays.

we live in the age of AI

i like AI morning lines much better than a biased human morning line.

AI morning line WAaaaaaay better.

Allan

They post those somewhere?

[turfeyejoe]

Are there tracks where the pools are so small the CAWs don't bother with them? Places like Fonner, Grants Pass, Rillito? If so, and they are available on online platforms, would it be worth playing?

[biggestal99]

Originally Posted by cj

They post those somewhere?

subscription to Equinedge (80 bucks American per month)

well worth the price of seeing what the actual prices should be.

Allan

[cj]

Originally Posted by biggestal99

subscription to Equinedge (80 bucks American per month)

well worth the price of seeing what the actual prices should be.

Allan

We do something similar in Formulator with Projected Odds that comes with the PPs.

[ranchwest]

Quick Grid has an odds line. It is basically a measure of the "best" horse, so it should not be considered a selection. It's usually less extreme than the off odds, higher on the low end and lower on the high end.

[Parkview_Pirate]

Originally Posted by Nitro

... Do you think for a minute that they care about what the original M/L might be?

Yes, I do. Wouldn't part of the whales/CAW approach be to analyze the pools and differentiate the "dumb" public money with money coming in late?

Looking at Daily Double probables early in the betting cycle, at least at some tracks, the influence of the M/L appears pretty obvious. In a more contentious race, that influence fades as post time approaches, but if the whales are half as smart as we think they are, surely they track the amounts of square and sharp bets to predict the final odds.

As for the article you linked, it reads rather suspiciously like a measured response from the CAW PR department who've been keeping tabs on complaints about the wagering "racket", as I would describe it.

"The Thoroughbred Racing Protective Bureau, or TRPB, has a policy to not be quoted as part of any article or by any media sources."

Oh really? Interesting. The folks in charge of ensuring the integrity of racing remain hidden in the shadows, with no obvious transparency? Not exactly reassuring on the topic of racing's image problems....

There's no way for me to know what's stated in the article is B.S., just as there's no way for you to know what's stated is true. I didn't learn anything from it.