Wisdom of the Crowds Polls

[Dave Schwartz]

This thread is the beginning of an experiement into the "Wisdom of the Crowds," based upon the book of the same name by James Suroweicki.
http://www.horsestreet.com/BBSImages/WisdomOfCrowds.jpg

My interest is to know which factors have what importance among the major handicapping factors. I assume that most of you would have similar interest in the answer.

The idea is that with the potential for some good answers from some very good handicappers we might get even better answers collectively.

I have chosen the following factors:

Pace Analysis (i.e. modeling)
Early Pace
Late Pace
Recent Speed
Past Speed
Earnings Box Class
Level-Based Class
Form Cycle (i.e. bounce and other patterns)
Form (recency, works, etc.)
Jockey
Trainer


Imagine that you are looking at a BRIS report or some other source (perhaps your own) for a non-maiden, dirt race. Further imagine that the fields above are columns in that report. The question is:

"To you, which column is most important, 2nd most important, etc.?"


There will be 5 polls - one for each of your choices. Thus, if you believe that "Pace Analysis" is most important and "Past Speed" 2nd most, you will select "Pace Analysis" in "Factors: 1st Choice" and "Past Speed" in "Factors: 2nd Choice."

I will allow an hour or so for questions and comments before posting the polls.

[jasperson]

Why did you leave out distance and Class? Speed or pace you need to know at what class it is achieved.

[CBedo]

Great book. The story of the guy and the gold mine is pretty enlightening of how insider group think can give you bad solutions. I wonder if having polls here at PA will be similar?

[Dave Schwartz]

Jas,

Can your provide a numerical score for distance?

I did list two kinds of class. What else did you have in mind?

[classhandicapper]

I think measuring ability is the most important thing. For me it's a process where it I think is best to use multiple sources of information (like speed figures, pace figures, some kind of classing methodology, charts, watching races etc..) because there are strengths and weaknesses to all the tools for measuring ability.

After that I think the trainer is the second most important thing.

After that it really depends. Some of the other factors can either be not important or very important depending on the specific circumstances.

For example, a really strong speed and rail bias can make a lot of things that are normally important somewhat irrelevant.

[jasperson]

When I look at speed I want to know at what distance and at what was the class of the race. I discount speed at 6f when the race is going to be a route. Also the same for jumping up in class. I want the speed at plus or minus 1/2 of furlong of todays races also this race should not be more than 1 notch up in class (c5000 to c6250).

[dansan]

form cycle or patterns

[Dave Schwartz]

You guys are adding words... where, in the BRIS or other reports is there a number which describes this? What is the column called?

Anecdotal handicapping does not come with weights.

[traveler]

If you try and apply the same factors with the same weights to a 6f cheap dirt claimer at Fort Erie as a 1 1/8 dirt route at Belmont you will be disappointed - your result will be "The Wisdom of the Crowds".

[Dave Schwartz]

The big boys use a global model and made $150m with it last year.

[traveler]

Won't disagree with that at all. Which would them imply to me you need to be able to play all the races they do to achieve their results(roi, not gross). Anyway, just not my cup of tea so I'll bow out. Good luck

[kenwoodallpromos]

OK I will await the polls!!

[LottaKash]

Nifty Poll Dave, I believe that I may have an inkling of where you are going with this....

Such obvious things, and yet, it already has given me something to think about.... The order and chaos of handicapping....New places & niches to explore, when trying to find the value of some things....

Watching with much interest and enthusasm....

best,

[Dave Schwartz]

Glad to see that a couple of you are catching on.

We're going to need upwards of 100 responses or so to make sense of it. Probably take about a month.

I can tell you this... Last night I took the preliminary rankings and used them in that order to produce handicapping output and, the one minor change I made produced a significant improvement.

[Donnie]

Having not read the book, will this matter in your research:

Factor 1 = 96 responses
Factor 2 = 92 "
Factor 3 = 69 "
Factor 4 = 64 "
Factor 5 = 62 '

Seems to be kind of a drop off after those first 2 factors....are "we" not looking any further than the first 2 "important" factors?

[098poi]

Dave are you trying to see what factors the public looks at most and the public being wrong most of the time, ultimately look where the public doesn't?

[Dave Schwartz]

No. In fact, precisely the opposite: I want to use the weightings that the public thinks are correct.

See, the public is actually quite smart.

If you haven't read the book, read about this:
http://www.biz.uiowa.edu/iem/index.cfm

[Dave Schwartz]

The entire matrix will play a part.

I've got an idea for weighting them together; at least enough to give us a starting point.

Perhaps tomorrow we'll have enough(??) - 100 was an arbitrary figure anyway.

Dave

[markgoldie]

Hi Dave;

Love that your mind is always working overtime. However, I have a question here. I haven't read the book but I am assuming from what you have written (and the title) that it extols the virtues of "mass' thinking and that the "wisdom of the crowd" is a potent means at getting at a truth.

I think we can all agree with that, but this notion would seem to be a subset of the well-known strength of market opinion to be very powerful, particularly when by "market" we mean that people are actively investing money.

If you remember, this notion of the power of markets, in fact, was what got Admiral Poindexter in trouble when he quite correctly instituted a market to predict where terrorist attacks would occur and by what means, etc. However, his intellectual understanding of the power of markets to deliver correct information was a bit insensitive in this case, since he was soliciting bets on who would get blown up next.

At any rate, wagering pools are very active and accurate markets, as you know. Therefore, if one wanted to understand what factors were the most important in leading to a handicapping decision, the most efficient way would seem to be reverse engineer the activity of the market itself. This would seemingly be far more accurate than taking a poll, albeit amongst nominal experts in the field. In fact, since wagering is a long-term Darwinian "survival of the fittest" contest, those who exercise the greatest influence over the market (ie. the biggest wagerers) will tend to hold the best handicapping knowledge. Therefore, if one wanted the most accurate poll, it would be best to ask the biggest bettors what they base handicapping decisions on.

This, of course, would lead us to the doorstep of the whales and I know that you're constantly looking for ways to get an edge on their methodologies. So put it another way, I'm not sure how you expect to find more accurate handicapping methodologies than those used by the whales, or as indicated by reverse-engineering the market into which the money goes, by opening a poll even amongst such luminous thinkers as we find here on the PA forum.

[Dave Schwartz]

Mark,

Not sure either.

It is, as described, an experiment. If it goes well, perhaps another, more sophisticated experiment.


Dave

[jonnielu]

I think measuring ability is the most important thing. For me it's a process where it I think is best to use multiple sources of information (like speed figures, pace figures, some kind of classing methodology, charts, watching races etc..) because there are strengths and weaknesses to all the tools for measuring ability.

After that I think the trainer is the second most important thing.

After that it really depends. Some of the other factors can either be not important or very important depending on the specific circumstances.

For example, a really strong speed and rail bias can make a lot of things that are normally important somewhat irrelevant.

You are right on here. A sound measure of ability can make all typical handicapping factors un-necessary. And, point out that very few other factors are meaningful at all.

A sound measure along with an understanding of ability can take you to the only other factor that weighs heavily on the outcome. That is whether or not the horsemen - trainer/jockey - know the ability, and are on the same page on its use.

The only other factor to understand is that it is a horserace. As a horserace, it is greatly subject to what is called the human factor. The effect of the human factor is:

The only thing that "figures" in a horse race, is that a horse race will seldom "figure" in a way that you would figure it.

jdl

[bobphilo]

Mark,

Excellent post. You and Dave are to be commended for your intelligent insights into this important topic. While I am becoming increasingly aware of the importance of the wisdom of the crowd, thanks to this thread, there are also problems with one of the arguments put forth in supporting it.

You make a very good point in pointing out the possible problems associated with the comparisons between public choices and those of professional handicappers in the media. The pari-mutual pool is not really a good model representing public opinion because it does not reflect public opinion in the sense of the democratic principle of “one man, one vote”. The whales, which may bet thousands of dollars on a horse compared to those who are $2 bettors, are represented disproportionately higher than the smaller bettors. In addition the professional media handicappers are at a disadvantage due to the fact that they have to make their selections in advance of the other bettors, without benefit of late scratches or track conditions, and sometimes, surface changes at post time. I made this point in my last post in the thread on Surowiecki’s book.

Of course, as I also stated in my first post in that thread, the fact that one argument in favor of a position has flaws, in no way means that the argument is disproved. In fact, the more I follow this thread, and its accompanying links, the more convinced I become that there really is something important here.

And Dave, thanks for beginning this poll. I am eagerly awaiting your results.

Bob

[bobphilo]

No. In fact, precisely the opposite: I want to use the weightings that the public thinks are correct.

See, the public is actually quite smart.

If you haven't read the book, read about this:
http://www.biz.uiowa.edu/iem/index.cfm

Dave, I agree 100%. While its true that favorites "only" win 33% of the time, many also forget that they only make up about 12% of the field. That gives them an impressive impact value of about 2.75, better than any other horse in the field based on ranking by betting preference. Thanks for exploding one more myth.

Bob

[Triopstor]

The big boys use a global model and made $150m with it last year.

Where are all the places one go to wager with a $150 million dollar bankroll?

Can one do this in Hong Kong?

[Dave Schwartz]

This is in the US.

[bobphilo]

I think measuring ability is the most important thing. For me it's a process where it I think is best to use multiple sources of information (like speed figures, pace figures, some kind of classing methodology, charts, watching races etc..) because there are strengths and weaknesses to all the tools for measuring ability.

After that I think the trainer is the second most important thing.

After that it really depends. Some of the other factors can either be not important or very important depending on the specific circumstances.

For example, a really strong speed and rail bias can make a lot of things that are normally important somewhat irrelevant.

Class, I agree and use a very similar methodology in my own handicapping. Handicapping really boils down to 1) determining a horse's ability and 2) determining how close to that ability level will it demonstrate under today's expected race dynamics (pace, trip, etc) and the horse's form cycle.

Ability can either be measured by 1) speed ratings, which reflect how fast a horse has run relative to how fast other horses have run in relation to their ability levels (final time adjusted by variant), or by 2) class ratings, which are based on how well a horse has performed against other horses relative to their class levels (Brisnet Class ratings use this method effectively).

Both speed and pace ratings can be improved by adjusting for pace and trip. As was pointed out on the Pace Analysis thread, these ratings are best used to adjust speed and/or class ratings rather than stand alone factors.
Like Classhandicapper, I prefer to use a combination of speed and class ratings (both adjusted for pace and trip) which gives me a good measure of ability when looked at from 2 different viewpoints.

Finally I look at todays likely pace and trip dynamics as well as where the horse is in its form cycle to predict how it will do today.

As far as trainers are concerned, while I agree that they often are major factors in a horse's performance, whatever role he/she has had on a horse's performance has already been accounted for in the horse's PPs. Unless there has been a trainer change, or a new situation in which that particular trainer either excells or does poorly in, to again go back and re-add the trainer as a factor is to double up his real contribution. Perhaps a novel approach but one with a sound logical basis.

Bob

[Capper Al]

No. In fact, precisely the opposite: I want to use the weightings that the public thinks are correct.

See, the public is actually quite smart.

If you haven't read the book, read about this:
http://www.biz.uiowa.edu/iem/index.cfm

I agree that the public is quite smart. You're trying to get a handle on the public. My guess is that the public goes by speed and jockey as long as the horse has had at least one good recent comparable race. The problem here is separating the public from the tote-board. When a horse is bet down, is it because the public bet it or a few insiders dumped a lot of money on it? It's probably a mixture of both. Good luck on your study. Keep us informed.

[Overlay]

Dave, I agree 100%. While its true that favorites "only" win 33% of the time, many also forget that they only make up about 12% of the field. That gives them an impressive impact value of about 2.75, better than any other horse in the field based on ranking by betting preference. Thanks for exploding one more myth.

Bob

Your statistic is correct, of course, and distinguishing between true and false favorites can be one approach to the game, but (as has been pointed out), the short odds involved don't leave much room for error. I agree with Dave that the crowd takes value into account to a certain degree, but not to the extent of correctly valuing every horse or exotic wager in every race. (There are too many "pick-the-winner" methodologies out there that are based on a process or elimination or isolated angles, with no regard for probabilities or odds, to allow that to happen.) I think that looking at the winning chances of all the horses in a race in comparison to their odds offers the best combination of both favorable probabilities and likelihood of profit.

[Warren Henry]

Your statistic is correct, of course, and distinguishing between true and false favorites can be one approach to the game, but (as has been pointed out), the short odds involved don't leave much room for error.

If one is good at separating the true from the false favorites, the short odds of the false favorite provide some of the best betting opportunities in this game. Nothing I like better than beating a false favorite with a nice priced contender.

[Overlay]

If one is good at separating the true from the false favorites, the short odds of the false favorite provide some of the best betting opportunities in this game. Nothing I like better than beating a false favorite with a nice priced contender.

I should have been clearer that I was trying to distinguish between confining play to favorites (and attempting to separate the good bets among them from the bad bets), versus looking for overlays among horses at all odds ranges. (It appears that you understood my meaning, but I just wanted to clarify.)

[bobphilo]

Your statistic is correct, of course, and distinguishing between true and false favorites can be one approach to the game, but (as has been pointed out), the short odds involved don't leave much room for error. I agree with Dave that the crowd takes value into account to a certain degree, but not to the extent of correctly valuing every horse or exotic wager in every race. (There are too many "pick-the-winner" methodologies out there that are based on a process or elimination or isolated angles, with no regard for probabilities or odds, to allow that to happen.) I think that looking at the winning chances of all the horses in a race in comparison to their odds offers the best combination of both favorable probabilities and likelihood of profit.

I understand what you're saying that if one can find a horse going off at a longer price than a more accurate odds-line indicates, it is a good bet whatever its odds. What I was saying was that when one rejects the favorite based on the fact that it doesn't win a majority of races one is missing the point that every other horse in the field wins less often. Whether one gets fair odds relative to a horse's chance of winning is another question.
We would both agree that passing on overlaid favorites because they don't win a majority of races is bad betting strategy.

Bob

[Jackal]

Having not read the book, will this matter in your research:

Factor 1 = 96 responses
Factor 2 = 92 "
Factor 3 = 69 "
Factor 4 = 64 "
Factor 5 = 62 '

Seems to be kind of a drop off after those first 2 factors....are "we" not looking any further than the first 2 "important" factors?

I found out the results of the poll and didn't participate any further. Moreover we don't have enough active participants for the poll to be accurate.

Personally, I treat each race as an individual. I don't have a "method" of handicapping any given race.

[Dave Schwartz]

I have completed the first phase of the process: the table of factors.

Perhaps someone would like to offer a weighting strategy for computing the table. Lacking one, I used a Fibonaci.

http://www.horsestreet.com/BBSImages/WOC-01.png


In my vision, "Pace," "Early Pace" and "Late Pace" are all "Pace." Therefore, for my purposes I see the table like this:

http://www.horsestreet.com/BBSImages/WOC-02.png

[Sly7449]

Greetings,

Any idea when Phase 2 will begin or is the Project continuing after the release of the Seminar which addresses Rank 3 and 4 of the Survey's Top 6?

In order to complete the Series, would it be necessary to await the Package of all six Seminars or is each Topic a Stand Alone?

Thanks

[Dave Schwartz]

Seems this entire thread ran out of interest. WOTC is still alive another project I am working on. You should be able to see the beginnings of it in December.

[Bullet Plane]

I read this book a few years ago. You have ten butchers try to give an estimate on how much a cow weighs...then an audience tries to guess. The collective audience answer is more accurate than any one expert.

They have had crowds beat experts in every field ..including math, science, stockmarket...whatever.

This is the phenomenon then goes on in horseracing...and why the game is so hard to beat. You are facing a crowd that has 1000 times more brain power than you do...If you are a great expert...So what? It does not help.

[thaskalos]

I read this book a few years ago. You have ten butchers try to give an estimate on how much a cow weighs...then an audience tries to guess. The collective audience answer is more accurate than any one expert.

They have had crowds beat experts in every field ..including math, science, stockmarket...whatever.

This is the phenomenon then goes on in horseracing...and why the game is so hard to beat. You are facing a crowd that has 1000 times more brain power than you do...If you are a great expert...So what? It does not help.If what you are saying is true, wouldn't the game be IMPOSSIBLE to beat?

[Bullet Plane]

If what you are saying is true, wouldn't the game be IMPOSSIBLE to beat?
I'm not saying it. It is what the author of the book is saying. But that and the takeout should make this game damn hard to beat. Yes.

Not impossible. You have luck as a factor. When I have a good bet, it is a "felt sense."

But to answer your head on without quibble....I don't know enough about the phenomenon. Perhaps Dave will enlighten us.

[ezpace]

thanks for the thread and all your work.