How would you reconcile your picks with the tote-board?

[Capper Al]

Originally Posted by whodoyoulike

The way I was understanding your OP was you had handicapped this race with these 4 contenders. I take it C and D have gone up in odds. Why would you get off if they still appear to be contenders? Again, I don't bet all my contenders.

This might uncover a hidden reason when to play value horses. If the tote-board selections are understandable and we get our value is the only time to play. If we don't understand the tote-board then even with high odds, our horse probably doesn't have any real value.

[Capper Al]

Originally Posted by cashmachine

I agree with Overlay: I would stick with my selections since I know from experience that my program has positive expectation. I do make adjustment of my probabilities using public odds but that's all. The very point of handicapping is to find scenarios when your prediction is different from public. If you chicken when it happens you can't make any money even theoretically: if you always go with public you won't overcome track take.

I also agree with DeltaLover that toteboard itself provides valuable clues. I have more than 10 indicators based only on the public odds that help me to incorporate into my prediction various patterns of toteboard behaviour. If after all adjustments I compute my probabilities and according to them horse is overlay I bet on him no matter what.

Even with a positive long run scenario with sticking to your selections, might you not be maximizing your profits play these races. This scenario might itself be a source of losses for you. It still might be better to skip.

[Capper Al]

Originally Posted by Overlay

I have enough confidence in my line (based not only on established probabilities, but also on the memory of horses such as you describe failing to live up to the expectations of those who are driving its odds down (usually to levels that I consider underlaid anyway), combined with the elevated payoffs on the horse's competitors as a result) that I almost never change my opinion or wagers. If I would somehow have reason to genuinely suspect/conclude that something illicit or completely imperceptible was not only accounting for the action on the horse, but could also potentially have a material effect on the outcome of the race, my fallback position would be to sit it out.

Might be the most sensible approach.

[Capper Al]

Originally Posted by traynor

I agree that it is both doable and simple. In essence, rather than modeling the attributes of the race winners, one models the attributes of the betting favorite. From there, it is a small step to calculate the probabilities of each profile actually winning. Comparing those profiles to entries in a given race would automate the process of "uncovering false favorites." It is also relatively simple to extend those profiles to place and show positions.

Accept that there is no way for us to understand the public's favorite outside of the tote-board. Either we miss the logic that the public sees or something is going on that smart money might know about with the information not available to handicap on.

[Capper Al]

Originally Posted by cashmachine

When do you compute your probabilities? Why you don't do it in the very last few seconds before the race start? I compute my probabilities 45 seconds before the race start and no new information can appear before the start, all toteboard action already taken into account.

I struggle at 2 minutes to post to make my mind up and get my bet in.

[Capper Al]

Originally Posted by imofe

I spend a lot of time trying to eliminate horses in the top 2 or 3 that I expect that the public will bet. In your example, if I expect E to get bet, then I have a playable race. Maybe A to win and an AB exacta or box depending on what the exacta payouts are. If I did not expect E to get bet, but he is at something like 5/2, then I have to worry about wise guy action on this horse. If there are unknowns about this horse ( shipper, layoff, recent barn change) then these horses are usually live.

This doesn't work here in most cases, and you are not the only one doing it. What I am talking about is the false favorite. Most false favorites should figure into your top four selections. A typical false favorite will have some good numbers in it's past performance like a high speed figure or dropping in class. We just can't assume the best handicapper, the public, doesn't know what they are doing everytime this scenario comes up.

[Robert Fischer]

Originally Posted by Capper Al

Let's say your top 4 picks are A-B-C-D and watching the tote-board you note the action is B-E-A-F. How would you reconcile this difference and make a wager? Your A-B are in but you are not seeing E-F as contenders.

example 1
My morning-line = B-E-A-C-D-F-G-H
My top-4-picks = A-B-C-D-E-G-F-H
I notice the tote action is = B-E-A-F-D-C-G-H

Next I compare My morning line (B-E-A-C) to the Tote Action (B-E-A-F).
The public is betting "C" as the sixth choice, instead of the fourth choice, and is betting "F" as the fourth choice instead of the sixth choice.
I did not anticipate this!
Since "E" is a big underlay here, I'm going to make money over the long-term with "A" and "B" in this race. If I want to use vertical exotics as well, I will use "F" along with "C" and "D".



example 2
My morning-line = B-C-A-D-E-F-G-H
My top-4-picks = A-B-C-D-E-F-G-H
I notice the tote action is = B-E-A-F-C-D-G-H

Next I compare My morning line (B-C-A-D) to the Tote Action (B-E-A-F).
The public is betting "E" as a top contender, and "C" is ice-cold on the board.
I did not anticipate this!
I don't know why the public is behaving in this fashion. I will pass this race entirely except when linking to a high-value multi race series, where I will now be forced to include the "E" (A-B-E-C).

[traynor]

Originally Posted by Capper Al

Accept that there is no way for us to understand the public's favorite outside of the tote-board. Either we miss the logic that the public sees or something is going on that smart money might know about with the information not available to handicap on.

The point is, that is not true. It ONLY appears so when one is looking at the world of racing as a one-time event, with no connection to anything that has gone before. In reality, the behavior of the betting public is as predictable as any other factor.

It isn't complex, complicated, unknowable, or hidden in any way. It just requires a bit more astute analysis of the available information than knee-jerk response to tote board movement. Specifically, it is useful to know when the tote board indicators usually associated with "insider betting" are based on insider betting and when they are based on (usually skewed and unrealistic) public opinion regarding some factor or combination of factors.

A simplistic example--speed on the rail wins three in a row. E and F--in your scenario--have inner posts and a bit of early speed. D and C are outside, and slow starters. What a myopic view might conclude was "insider betting" is perfectly normal under the circumstances. While the reasons for the tote action may not be as blatant (and noticeable to the uncritical observer) as that indicated above, in most cases they are both knowable and apparent. If one is using appropriately sophisticated computer models.

[traynor]

I have a real problem with the notion of analyzing a race, making my choices, then looking for external validation of those choices on the tote board before wagering. It seems that would virtually guarantee wagers restricted to crowd favorites, while passing races in which one's selections were actually overlays.

[tophatmert]

Originally Posted by Tom

I read that wrong - If I make a horse a contender, I don't care that that it is not in the top 4. If I don't make a horse a contender, love to see it getting action. I go by my handicapping.

I agree especially in claiming and ALW races for older horses. 2yos and lightly raced maidens are not my cup of tea because there are more variables that I don't care to analyze. Long term I will take my opinion over the public and barns trying to put one over . I have taken notes on every race from BEL SAR and AQU since 1989. I use good speed figures . I have studied a great deal over the years why would I let myself be influenced by a guy who heard from a guy. Most of the time I believe I know something that the public doesn't know. I get that information from hard work and paying attention and I won't let that me marginalized by fear of the tote.

[DeltaLover]

Originally Posted by traynor

I agree that it is both doable and simple. In essence, rather than modeling the attributes of the race winners, one models the attributes of the betting favorite. From there, it is a small step to calculate the probabilities of each profile actually winning. Comparing those profiles to entries in a given race would automate the process of "uncovering false favorites." It is also relatively simple to extend those profiles to place and show positions.

The problem is that in the vast majority of the races, the calculated probabilities of each profile (as you put it very accurately and correctly) will be aligned with crowd's selection. Sure, there might be some discrepancies but the huge takeout is enough to convert them to losing propositions.

What is needed, is a way to estimate not only the win probability of each horse but the probability of the crowd committing an error of a specific magnitude in its final decision, regardless of the odds of each individual starter.


I think that the key to understand what I am trying to say here, is the expression specific magnitude and refers to the probability of the crowd to commit a very large error when forming its opinion..

The question that needs to be addresses here is the following:

Given a race profile (encapsulating of all publicly known data, including past performances, recent connection behavior, well known names etc) what is the probability that the crowd will commit a LARGE ERROR in its final assessment?


Of course I need to define what I mean by LARGE ERROR, which is probably not the main topic in this thread. From a high level, we can consider a LARGE ERROR cases where for example a 10-1 is expected to run as a 2-1 shot..

BUT this is not exactly what I mean here...

Instead of a specific horse that might be misjudged by the crowd, I am referring to the WHOLE RACE as an entity.


Let's say that we use some measurement to express the accuracy of the crowd (it can be R^2 or any other similar metric)... Following what I have just described, our objective becomes to detect if we can find a cluster of races similar to the current one, where the accuracy of the crowd is systematically less than the rest of the racing universe...

If we are able to accurately detect this kind of clusterization, out betting strategy becomes much simpler as we now in advance what races we should be playing and the next question we should answer has to do with the profile of the horse to bet, which can be again discovered by analyzing where the crowd's error is coming from.

[traynor]

Originally Posted by DeltaLover

The problem is that in the vast majority of the races, the calculated probabilities of each profile (as you put it very accurately and correctly) will be aligned with crowd's selection. Sure, there might be some discrepancies but the huge takeout is enough to convert them to losing propositions.

What is needed, is a way to estimate not only the win probability of each horse but the probability of the crowd committing an error of a specific magnitude in its final decision, regardless of the odds of each individual starter.


I think that the key to understand what I am trying to say here, is the expression specific magnitude and refers to the probability of the crowd to commit a very large error when forming its opinion..

The question that needs to be addresses here is the following:

Given a race profile (encapsulating of all publicly known data, including past performances, recent connection behavior, well known names etc) what is the probability that the crowd will commit a LARGE ERROR in its final assessment?


Of course I need to define what I mean by LARGE ERROR, which is probably not the main topic in this thread. From a high level, we can consider a LARGE ERROR cases where for example a 10-1 is expected to run as a 2-1 shot..

BUT this is not exactly what I mean here...

Instead of a specific horse that might be misjudged by the crowd, I am referring to the WHOLE RACE as an entity.


Let's say that we use some measurement to express the accuracy of the crowd (it can be R^2 or any other similar metric)... Following what I have just described, our objective becomes to detect if we can find a cluster of races similar to the current one, where the accuracy of the crowd is systematically less than the rest of the racing universe...

If we are able to accurately detect this kind of clusterization, out betting strategy becomes much simpler as we now in advance what races we should be playing and the next question we should answer has to do with the profile of the horse to bet, which can be again discovered by analyzing where the crowd's error is coming from.

That is essentially what (in part) I am doing. Consider 100 variables, 10 of which are "considered" by the crowd as "indicators." The other 90 function as confounding variables. Find clusters in (what many ignore as "unknowable") those 90 (or whatever) variables that indicate when the 10 variables considered indicative by the crowd are negated or seriously diminished by their existence.

Short version would be, "Don't look only what what wins. Look at what loses." and merge that with the "what-wins" categories, so that discrepancies are readily apparent. That makes detection of races with "false favorites" quite simple.

[Robert Fischer]

Originally Posted by classhandicapper

I think about odds 2 ways.

1. What do I think the odds will be? (like my own morning line)

2. What do I think the fair odds are?

If the actual odds disagree with my predicted odds significantly, then I think there's always possibility that I am missing something or someone knows something I don't. So I might double check those horses (trainer pattern, trip, etc..that I missed)

If the odds look around where I expected them to be, then the only thing that matters to me is #2.

Yea, this is very close to what I do.

[DeltaLover]

Originally Posted by traynor

That is essentially what (in part) I am doing. Consider 100 variables, 10 of which are "considered" by the crowd as "indicators." The other 90 function as confounding variables. Find clusters in (what many ignore as "unknowable") those 90 (or whatever) variables that indicate when the 10 variables considered indicative by the crowd are negated or seriously diminished by their existence.

Short version would be, "Don't look only what what wins. Look at what loses." and merge that with the "what-wins" categories, so that discrepancies are readily apparent. That makes detection of races with "false favorites" quite simple.

I completely agree and even push your statement a little higher.

As a gambler, one of your worst attitudes is to be result oriented. The outcome of a specific event or even a (relatively short) sequence of them is completely a matter of luck and there is nothing you can do to predict it.

The best you can do, is to try to detect favorable betting situations knowing that you are actually making money while you are placing your bets and not when you are cashing out!

Horse bettors who still think of the pools as some form of an ATM, that can be used for regular daily withdraws, is a very amusing situation. People do not seem to realize that, despite the fact that without a doubt, horse racing is highly predictable, it is still gambling with the fullest meaning of the world!

[whodoyoulike]

Originally Posted by Capper Al

This doesn't work here in most cases, and you are not the only one doing it. What I am talking about is the false favorite. Most false favorites should figure into your top four selections. A typical false favorite will have some good numbers in it's past performance like a high speed figure or dropping in class. We just can't assume the best handicapper, the public, doesn't know what they are doing everytime this scenario comes up.

I haven't really thought this through but isn't a false favorite from a handicapper's or your perspective a favorite which is not your selection?

And, why do you consider the public is the "best handicapper"? Is it because they are consistently able to identify the winners 33% +/- of the time?