On-Track Observation

[GameTheory]

Originally Posted by traynor

That would be as foolish and one-dimensional as relying exclusively on crunching Equibase PPs to make selections.

To make profit (maybe), but it could certainly establish whether it was a useful practice or not, especially if it could be quantified into numerical factors. I have a book around here somewhere that attempted to do something like that -- has pictures of various ear positions, etc.

I once posted in this forum about a program that I had made that made selections based *solely* on the last race chart comments for each horse. It had a winning hit rate of around 19% and was usually profitable (I say usually because profit came from huge longshots, so it did badly in most small samples but ok on large ones). Not the greatest way to actually play in real life, but certainly showed that it could be a valuable factor. (Of course I was mocked by a few for posting such a thing, which shows why it could be valuable -- it was both predictive and handicappers did not want to take it seriously.)

[DeltaLover]

Originally Posted by GameTheory

To make profit (maybe), but it could certainly establish whether it was a useful practice or not, especially if it could be quantified into numerical factors. I have a book around here somewhere that attempted to do something like that -- has pictures of various ear positions, etc.

I once posted in this forum about a program that I had made that made selections based *solely* on the last race chart comments for each horse.

I have gone through the same direction in the past, using Markov chains to analyze past performance chains, although I have to admit with not much success (if memory serves, marginally beating a completely random bettor)...

[johnhannibalsmith]

Originally Posted by thaskalos

It is often said that the horse who appears to be high-strung in the paddock is a bad bet...because he is using up his energy before the race. By the same logic...shouldn't the horse who is lethargic-looking in the paddock be a DESIRABLE wager...since it spends as little energy as possible before the race?

I'll try to make a comparison that might work and put it into perspective a bit - there's a look of controlled energy that is probably the ideal point. Like a boxer making his entrance and appearing utterly focused and reeking of adrenaline, but every action is measured and made with confidence.

Then there's two wiry drunks in the parking lot about to come to blows but seem intent on exerting themselves with looping haymakers that serve to exhaust them more than actually harm the intended target. By the time the dance commences, they've run out of both true physical energy and any mental advantage.

Of course, there's the boxer, Geraldo Rivera or Tonya Harding, that certainly can't be accused of wasting any energy - but the mental element, the intensity and confidence, is simply non-existent.

There's the middle ground. Controlled confidence and radiant energy despite a sense of complete calm.

[traynor]

Originally Posted by GameTheory

To make profit (maybe), but it could certainly establish whether it was a useful practice or not, especially if it could be quantified into numerical factors. I have a book around here somewhere that attempted to do something like that -- has pictures of various ear positions, etc.

I once posted in this forum about a program that I had made that made selections based *solely* on the last race chart comments for each horse. It had a winning hit rate of around 19% and was usually profitable (I say usually because profit came from huge longshots, so it did badly in most small samples but ok on large ones). Not the greatest way to actually play in real life, but certainly showed that it could be a valuable factor. (Of course I was mocked by a few for posting such a thing, which shows why it could be valuable -- it was both predictive and handicappers did not want to take it seriously.)

Triangulation--a synthesis of quantitative and qualitative approaches--is often useful. It could be argued that reducing qualitative analysis to a set of numbers would be self-defeating, and little more than a different form of quantitative method.

Business analysts (among many others) discovered quite awhile back that when the quantitative approach indicates A is best, the qualitative approach indicates B is best, the optimal choice is often C--none of the above.

The example of a slow horse that looks great (mentioned above, by Hoofless-Wonder) is a good example. So would be the inverse--a horse with big numbers that looks and acts like Eeyore. In such cases (and many others with much less drastic differences) it is not especially useful to revert to quantifying by degrees (which puts too much emphasis on largely cosmetic differences in sets of numbers, most of which are little more than imperfect descriptors of some segment of a past event).

Passing such races (despite occasional lost opportunities that one will tend to remember much more strongly than the much greater number of losing wagers saved) is usually a good idea.

[Maximillion]

Originally Posted by traynor

Triangulation--a synthesis of quantitative and qualitative approaches--is often useful. It could be argued that reducing qualitative analysis to a set of numbers would be self-defeating, and little more than a different form of quantitative method.

Business analysts (among many others) discovered quite awhile back that when the quantitative approach indicates A is best, the qualitative approach indicates B is best, the optimal choice is often C--none of the above.

The example of a slow horse that looks great (mentioned above, by Hoofless-Wonder) is a good example. So would be the inverse--a horse with big numbers that looks and acts like Eeyore. In such cases (and many others with much less drastic differences) it is not especially useful to revert to quantifying by degrees (which puts too much emphasis on largely cosmetic differences in sets of numbers, most of which are little more than imperfect descriptors of some segment of a past event).

Passing such races (despite occasional lost opportunities that one will tend to remember much more strongly than the much greater number of losing wagers saved) is usually a good idea.

I know you have been firm on this from day one, and I believe you when you say that you use visual analysis effectively.
Do you have to actually be at the track though to do this to your satisfaction?

[thaskalos]

Originally Posted by traynor

Triangulation--a synthesis of quantitative and qualitative approaches--is often useful. It could be argued that reducing qualitative analysis to a set of numbers would be self-defeating, and little more than a different form of quantitative method.

Business analysts (among many others) discovered quite awhile back that when the quantitative approach indicates A is best, the qualitative approach indicates B is best, the optimal choice is often C--none of the above.

The example of a slow horse that looks great (mentioned above, by Hoofless-Wonder) is a good example. So would be the inverse--a horse with big numbers that looks and acts like Eeyore. In such cases (and many others with much less drastic differences) it is not especially useful to revert to quantifying by degrees (which puts too much emphasis on largely cosmetic differences in sets of numbers, most of which are little more than imperfect descriptors of some segment of a past event).

Passing such races (despite occasional lost opportunities that one will tend to remember much more strongly than the much greater number of losing wagers saved) is usually a good idea.

Reading this, I find myself wondering if you've ever authored anything for mass readership.

And if not...why not?

[GameTheory]

Originally Posted by traynor

Triangulation--a synthesis of quantitative and qualitative approaches--is often useful. It could be argued that reducing qualitative analysis to a set of numbers would be self-defeating, and little more than a different form of quantitative method.

Business analysts (among many others) discovered quite awhile back that when the quantitative approach indicates A is best, the qualitative approach indicates B is best, the optimal choice is often C--none of the above.

Yes, maybe, but that assumes that both A & B have already proved their merit and are both available to you. The same could be said of any two factors whether both are quantitative or qualitative or one of each. If one thing says good and another says bad, then you've got a decision to make, what else is new? Should we just never look at one of the factors to make it easier on ourselves?

But getting back to the point, if you want to find out whether something is worth doing, then some sort of structured test where you isolate it needs to be done (either by looking at it as a sole factor or by somehow making "all else equal" if it is to be used as minor modifying factor that doesn't make sense in total isolation).

And "reducing" qualitative analysis to numbers is not self-defeating, it is the only way to use it (in the general sense), naturally if you overdo it looking for phantom distinctions that's no good, but even if all you have is "this is positive" or "this is negative" that is the same as "reducing" it to 1/0. Just depends how fine a granularity is useful for that particular factor. Is it little more than than a different form of quantitative method? Of course -- all factors are as soon as you bring them to bear in any structured, disciplined (i.e. repeatable) way. That's not a criticism though. Ultimately, there is no such thing as "hard" data at the source of all our factors, they've all undergone this transformation into numbers or categories. We put things in bins, we create (ultimately) arbitrary thresholds, we order and rank things, etc. That's how we make sense of the world, not just in handicapping but in all things.

[traynor]

Originally Posted by Maximillion

I know you have been firm on this from day one, and I believe you when you say that you use visual analysis effectively.
Do you have to actually be at the track though to do this to your satisfaction?

Yes.

[traynor]

Originally Posted by GameTheory

Yes, maybe, but that assumes that both A & B have already proved their merit and are both available to you. The same could be said of any two factors whether both are quantitative or qualitative or one of each. If one thing says good and another says bad, then you've got a decision to make, what else is new? Should we just never look at one of the factors to make it easier on ourselves?

But getting back to the point, if you want to find out whether something is worth doing, then some sort of structured test where you isolate it needs to be done (either by looking at it as a sole factor or by somehow making "all else equal" if it is to be used as minor modifying factor that doesn't make sense in total isolation).

And "reducing" qualitative analysis to numbers is not self-defeating, it is the only way to use it (in the general sense), naturally if you overdo it looking for phantom distinctions that's no good, but even if all you have is "this is positive" or "this is negative" that is the same as "reducing" it to 1/0. Just depends how fine a granularity is useful for that particular factor. Is it little more than than a different form of quantitative method? Of course -- all factors are as soon as you bring them to bear in any structured, disciplined (i.e. repeatable) way. That's not a criticism though. Ultimately, there is no such thing as "hard" data at the source of all our factors, they've all undergone this transformation into numbers or categories. We put things in bins, we create (ultimately) arbitrary thresholds, we order and rank things, etc. That's how we make sense of the world, not just in handicapping but in all things.

Or--according to Korzybski--how we pretend to understand the world by affixing labels to things, then responding to the label (as if it were real rather than a purely artificial construct) rather than the thing labeled. By falling into the trap of defining something as "worth" 2.73 when combined with something else with an acceptable range of 9.4 to 12.7, the tacit assumption is that enough identical (or nearly identical) situations have been thoroughly monitored and evaluated in the past to know the exact value of each--not just in isolation--but in all possible combinations with all other relevant influences. That is pretty much impossible in horse racing.

With all due respect to the bean counters, I don't think a very good job has been done quantifying the (relatively easy) quantifiable aspects of horse racing, much less the qualitative aspects. The current state of software--including that which its developers embellish with phrases like "artificial intelligence" and "machine learning"--is not much better (if at all) than the old time "systems" that assigned points for various "factors" to arrive at some "bet the high number" simplification.

The difficulty is in the confounding variables--influences overlooked, ignored, hidden, or otherwise obscured--that make quantification much less useful than it might otherwise be. "Precise" humerical evaluations look very impressive, and may give aid and comfort to those who believe in them, but (in horse racing) are usually only descriptive, not prescriptive.

[DeltaLover]

Originally Posted by traynor

Or--according to Korzybski--how we pretend to understand the world by affixing labels to things, then responding to the label (as if it were real rather than a purely artificial construct) rather than the thing labeled. By falling into the trap of defining something as "worth" 2.73 when combined with something else with an acceptable range of 9.4 to 12.7, the tacit assumption is that enough identical (or nearly identical) situations have been thoroughly monitored and evaluated in the past to know the exact value of each--not just in isolation--but in all possible combinations with all other relevant influences. That is pretty much impossible in horse racing.

With all due respect to the bean counters, I don't think a very good job has been done quantifying the (relatively easy) quantifiable aspects of horse racing, much less the qualitative aspects. The current state of software--including that which its developers embellish with phrases like "artificial intelligence" and "machine learning"--is not much better (if at all) than the old time "systems" that assigned points for various "factors" to arrive at some "bet the high number" simplification.

The difficulty is in the confounding variables--influences overlooked, ignored, hidden, or otherwise obscured--that make quantification much less useful than it might otherwise be. "Precise" humerical evaluations look very impressive, and may give aid and comfort to those who believe in them, but (in horse racing) are usually only descriptive, not prescriptive.

Nice post..

I would add to:

Originally Posted by traynor

The difficulty is in the confounding variables--influences overlooked, ignored, hidden, or otherwise obscured--that make quantification much less useful than it might otherwise be.

that this 'difficulty' is magnified by the stochastic nature of the game which is pretty accurately reflected in the odds offered by the various pools...

Either we want to accept it or not, the take out in this game is so large, that it is quite possible to evaporate the edge of even the most astute horse bettor, as his competitors improve their abilities to handicap and play.

[PhantomOnTour]

I think some of y'all are missing the boat on physicality handicapping.

The paddock is as important as the warmup, and no one is talking about that.
It's all nervous energy, ears pricked, kidney sweat etc etc etc...but what about that runner who has swelling in his left rear or seems off in his right rear or is walking short?
That is to be seen in the paddock with the horses moving in slo-mo, giving you a nice long look at his legs, equipment (bits!!!), demeanor etc...

I know only about 2% of us even get to the track on a regular basis anymore, but that's the spot to be.

[Tom]

Like it or not, they are horses - living creatures,not data points.

[overthehill]

I certainly would not mock you. I still remember making a big bet blind without knowing the price on a horse making its second start on the turf over 25 years ago. in its first start on the turf it was blocked three times during the running of the race and closed for third beaten 2 lengths for all the money.it went off 70-1 that day and was trained by Oliver Hosang a trainer i had never heard of. In its second start it was the 4-1 ml second choice from the #12 post and the jockey had hopped off it to ride the 3-1 ml favorite. I bet out with confidence hoping for 5-2 and was shocked when it paid $26 after it wired the field by 6 lengths! Unfortunately I havent seen that kind of opportunity again given all the opportunities to see replays and trip handicapping vogue those types of trip horses are now more likely to be underlays.

[thaskalos]

Originally Posted by DeltaLover

:

Either we want to accept it or not, the take out in this game is so large, that it is quite possible to evaporate the edge of even the most astute horse bettor, as his competitors improve their abilities to handicap and play.

And the process is so subtle, that most of us won't even know when that point is reached.

[traynor]

Originally Posted by PhantomOnTour

I think some of y'all are missing the boat on physicality handicapping.

The paddock is as important as the warmup, and no one is talking about that.
It's all nervous energy, ears pricked, kidney sweat etc etc etc...but what about that runner who has swelling in his left rear or seems off in his right rear or is walking short?
That is to be seen in the paddock with the horses moving in slo-mo, giving you a nice long look at his legs, equipment (bits!!!), demeanor etc...

I know only about 2% of us even get to the track on a regular basis anymore, but that's the spot to be.

Not missing it at all. I am a big fan of watching the horses as they first enter the track, on the way to the paddock, in addition to the paddock and warmups. I think the most important part is to really focus on each horse for a period of time long enough to get a good idea of its appearance, condition, and demeanor. And periodically focus back on each entry to see if the first impressions are accurate, or whether it appears differently on later inspections. With all due respect to Joe Takach and others who seem to believe a quick glance on a video is sufficient, that may be why so many find it "difficult."