Hanicapping is Difficult: Proof

[thoroughbred]

The results of the Bloodstock (BRIS) Breeders Cup contest offers proof, sort of, of how difficult successful handicapping is.

To review: The object of the contest was to pick one horse in each of the 8 Breeder's Cup races. Starting with the first race, if your pick either WON, PLACED OR SHOWED, you remained in the contest, otherwise you got eliminated.

The winner of the contest would be the person who remained in the contest the longest.

Now think about this:

We can assume that a few thousand people entered the contest.

Since it was through Bloodstock, we might extpect that the handicapping skills of the entrants were above the average of the "run-of-the-mill" handicapper.

You didn't have to pick the winner; you were successful even if your horse was ONLY in the money

What was the result? If I'm reading the contest results correctly, THERE WAS ONLY ONE PERSON STILL IN THE CONTEST AT THE END OF THE FOURTH RACE!

I guess the obvious conclusion is: "No one ever said that handicapping is easy."

[JimG]

...especially when we are talking about the Breeders Cup. Not only is it a championship for horses but for horseplayers as well. Even if you lose, which I did this year, it is a rush to take a shot at some of the payouts.

JimG

[Topcat]

Let's see, if picks hit the board at the rate that favorites do the odds of picking four winners in a row would be
.63 x .63 x .63 x .63 or 15% of the time for one player.

So if only one player was left after four races then my guess is that there were not thousands of entrants.

Other opinions?

[thoroughbred]

Originally posted by Topcat
Let's see, if picks hit the board at the rate that favorites do the odds of picking four winners in a row would be
.63 x .63 x .63 x .63 or 15% of the time for one player.

So if only one player was left after four races then my guess is that there were not thousands of entrants.

Other opinions?

two comments:
I'm not sure your analysis comparing favorites is correct.
And, even if so, it's well known that favorites win only 1/3 of the time, so for 4 races we get about 3%

[ranchwest]

According to Quirin's research, horses win at approximately the rate of their odds. Since the first four BC races were won by horses whose odds were 12.3, 11.9, 5.1 and 9.6, accepting Quirin's research tells us that to pick those four races the percent for the average player would be .00167%. This would be for winners. I'm not sure how to get back to horses who come in the money, but since favorites win 1/3 of the time and come in the money 2/3 of the time, let's assume that what we need to do is multiply by 2. That gives us .00334%. I'd say that that implies there were probably thousands of people in the contest, although we would hope that this group was a better bunch of handicappers than the average of all players. If all of the above is accurate, then, on average, it would take 29,941 participants in the contest to produce one winner. I'd guess that there were fewer participants, indicating that the winner was a better than average handicapper.

Of course, in actuality the math is much more complex because any _one_ of three horses qualifies and the odds of each of those is different, so there are a lot of possible winning combinations, but each participant can have only one.

My point is merely that it was difficult.

The difficulty of these particular races is further evidenced by the pick 3 payoffs. For the first three races, $1,671. The next rolling pick 3 was $2,396. The following, the one beyond which anyone in the contest qualified was $1,641.

One of the many factors here is that the European horses don't carry the same data as U.S. horses and it is mostly Americans in the contest and the European horses fared well.

[Tuffmug]

Most handicappers come into a knife fight armed with a pair of nail clippers!

It takes the right tools to beat this game (and I'm not necessarily talking about computers)

[Topcat]

Originally posted by thoroughbred


two comments:
I'm not sure your analysis comparing favorites is correct.
And, even if so, it's well known that favorites win only 1/3 of the time, so for 4 races we get about 3%

Tbred:
As i read the contest you just have to hit the board so a "winner is win, place or show hence the .63

[Topcat]

Originally posted by ranchwest
According to Quirin's research, horses win at approximately the rate of their odds. Since the first four BC races were won by horses whose odds were 12.3, 11.9, 5.1 and 9.6, accepting Quirin's research tells us that to pick those four races the percent for the average player would be .00167%. This would be for winners. I'm not sure how to get back to horses who come in the money, but since favorites win 1/3 of the time and come in the money 2/3 of the time, let's assume that what we need to do is multiply by 2. That gives us .00334%. I'd say that that implies there were probably thousands of people in the contest, although we would hope that this group was a better bunch of handicappers than the average of all players. If all of the above is accurate, then, on average, it would take 29,941 participants in the contest to produce one winner. I'd guess that there were fewer participants, indicating that the winner was a better than average handicapper.

Of course, in actuality the math is much more complex because any _one_ of three horses qualifies and the odds of each of those is different, so there are a lot of possible winning combinations, but each participant can have only one.

My point is merely that it was difficult.

The difficulty of these particular races is further evidenced by the pick 3 payoffs. For the first three races, $1,671. The next rolling pick 3 was $2,396. The following, the one beyond which anyone in the contest qualified was $1,641.

One of the many factors here is that the European horses don't carry the same data as U.S. horses and it is mostly Americans in the contest and the European horses fared well.

Point well taken-it is difficult. You raise an interesting speculative point. Do handicappers merely mimic the odds line in their selections? Of course we know the mass of handicappers as measured by dollars do but are the rest of us any different?

Let's see I checked out one database of mine and checked to see how often the above horse at 12.3, 11.9, 5.1 and 9.6 hit the board and multiplied out the probabilities. I come up with .0006
-multiple that times 1,500 and you get one your one winner.

I'm intrigued by the implication that even software handicappers do no better then the public at large. Maybe this is why the % of winning favorites has been so consistent over the last 60 years. Dad and Grandpa didn't need no stinkin computers!

[ranchwest]

I think you're right. Generally, we are so busy making odds lines that we don't look for advantages. Then, we look for our advantage in an overlay. Seems like a backwards approach to me.

[thoroughbred]

Originally posted by Topcat


Point well taken-it is difficult. You raise an interesting speculative point. Do handicappers merely mimic the odds line in their selections? Of course we know the mass of handicappers as measured by dollars do but are the rest of us any different?

Let's see I checked out one database of mine and checked to see how often the above horse at 12.3, 11.9, 5.1 and 9.6 hit the board and multiplied out the probabilities. I come up with .0006
-multiple that times 1,500 and you get one your one winner.

I'm intrigued by the implication that even software handicappers do no better then the public at large. Maybe this is why the % of winning favorites has been so consistent over the last 60 years. Dad and Grandpa didn't need no stinkin computers!

----------------------------------------------------------------------

TopCat:

I don't know all the answers either. But here are a few comments (questions?) about your note.

You say the mass of handicappers mimic the odds. On the other hand it's the mass of handicappers who make the odds. So how can they both make the odds and, at the same time, mimic the odds?

You also use the fact that over 60 years the rate of win for favorites hastn't changed, as a way of concluding that modern handicapping techniques, such as using computers, are no better than what Dad and Grandpa were able to do. There are other interpretations that may be more valid. For example. It may be that it is far easier to recognize the good horses that end up being the favorites. In other words this is an easier task than recognizing other horses with potential. Being easier, it can be done with or without computers. But it may be, that the more difficult task of recognizing a more occult horse as having a chance at winning, requires more sophistication. Here computers, perhaps, used properly give us an edge.

As I reminded us: "No one ever said handicapping is easy."

[tanda]

No doubt handicapping is difficult, but the main reason for the difficulty in the BRIS contest has not been raised.

Through the first six races the betting favorite (who I believe were also the ML favorites) did not finish in the money in any race. That is highly unusual, although less so in the Breeders Cup. And I do not belive that there is an explanation for it other than it was a fluke. There was no demonstrably false favorite in any of those races. Some of the favorites were only my second of third choice, but none were complete throwouts.

A player would have had to avoid using You, Flute and Officer. How many of us did that? I did not like You or Flute to win, but thought they had good chances to hit the board. Also, since odds were irrelevant (except for a tie-breaker), most of us had no problem using chalk if we thought it was legitimate.

Also, any player who used Exogenous with Flute or another out of the money horse as a backup were beat.

So, you had to avoid the big three mentioned above, not use Exogenous with Flute backup, not use any other betting favorite in the other three races AND in addition to avoiding those temptations, still use a horse that finished in the money. If you did all that, you still had to complete two more races.

No matter what, it is tough to survive all eight, but the magnitude of difficulty went up exponentially when three big favorites (and possibly legitimate favorites) ran out of the money as well as three other defensible favorites. That does not happen much. In fact, I bet a database search would reveal a very low possibility of it occurring 6 races in a row. My estimates are less than 2/10 of one percent in a normal race, maybe 8/10 of a percent in large field, contentious races such as the Breeders Cup.

[thoroughbred]

[QUOTE]Originally posted by tanda

[B]No doubt handicapping is difficult, but the main reason for the difficulty in the BRIS contest has not been raised.

------------------------------------------------------------------------

Tanda,

O.K. Let's forget about the Breeder's Cup.

What about this? I only have data for the case where we have to bet EVERY race and only choose ONE horse to WIN in each race. In this particular case, (which does represent what not all, but many, handicappers do), there is the startling result that:

ALMOST NOBODY EVER ACHIEVES AN AVERAGE WIN PERCENTAGE OF 40% OR MORE.

The key word here is "average." In other words a person's success rate over a long period of time, not a short term winning streak.

So on any given day, under the conditions described, a horse player should EXPECT to win less than 4 out of 10.

So again: "No one ever said that handicapping is easy!)

[Topcat]

[Originally posted by thoroughbred [/i]

<You say the mass of handicappers mimic the odds. On the other hand it's the mass of handicappers who make the odds. So how can they both make the odds and, at the same time, mimic the odds? >

Actually what I asked is: “Do handicappers merely mimic the odds line in their selections? Of course we know the mass of handicappers as measured by dollars do but are the rest of us any different? “

To clarify I should probably have said “ Do ALL handicappers mimic the odds line. I worked with Paul Peterson and few others years ago in a study on predictive factors on the final odds. Do you know what the most predicitve factor was? The morning line. The findings were that many handicappers bet back to the odds line either through confimation of their selection or as a starting and ending point in their selection.

I think we all confuse the so called betting public as some sort of betting democracy.I wish it were becaus i beleive my advantage would improve. The fact is though that if I bet $200 it is 100 times more important to the odds line than the $2 bet. So my question is do even bigger bettors bet back to the odds line?

You wrote:
<It may be that it is far easier to recognize the good horses that end up being the favorites. In other words this is an easier task than recognizing other horses with potential. >

Yes I think you have something here. The “good” horse all have good numbers be they a simple DRF speed rating, win record etc. or more sophisticated numbers -they all stand out.

And yes I do agree that computers can give us a n edge. I just don’think as a whole that the advantages of the information age have translated into any discernable change in the last 60 years as reflected in the % of winners.-hence my musings on bettors mimicing the odds line.

Of course it may very well be that there is something else going on here. Some sort of natural throttling of odds. I had it pointed out to me that if I look athe winning % of the top two that it is consistently around 50% and that it suggest that each race is really a match race between the two most likely favorites and the rest of the field.- An interesting thought-not sure if it holds up though.

[ranchwest]

It seems to me that Quirin's research appears to still be valid, that horses win in correlation with the frequency indicated by their odds. Does anyone disagree?

[Dave Schwartz]

RanchWest,

>>> It seems to me that Quirin's research appears to still be valid, that horses win in correlation with the frequency indicated by their odds. Does anyone disagree?<<<

In the far-less than absolute sense, I agree. That is, on a scale of 0.00 to 1.00 I am about 75% of agreement. (How's that for fuzzy logic? <G>)

What I mean is that to agree 100% would be to say that the win pool is (typically) an efficient market, which it is not.

In Quirin's day the win percentages flowed along with the odds, with the exception that low odds horses were underbet (relative to the takeout) and very high odds horses were over bet. This was illustrated by the accepted loss of 9% when betting favorites and other low-priced horses.

Now the loss on favorites has jumped to around 20%, indicating that favorites have gone from being underbet to being overbet.

If one were to describe the effect of the tote today I believe it would be best said by saying that the low odds and high odds horses are overbet and middle odds horses are underbet.

Interestingly enough, if you put the odds ranges from Quirin's book into a spreadsheet, he had the same "hump in the middle" back then as well.


Regards,
Dave Schwartz

[BMeadow]

Do horses win in proportion to their odds? Absolutely. In the January 2000 issue, we printed the results of a research project completed by Jim Cramer of Handicappers Data Warehouse. More than 200,000 nationwide races were checked.

The complete results are listed in the article, but here are the win percentages and ROI of various odds ranks:

Favorite 32.8 0.81
2nd Ch 20.8 0.82
3rd Ch 15.0 0.80
4th Ch 10.9 0.79
5th Ch 7.8 0.79
6th Ch 5.6 0.78

There was also a breakdown by odds which showed clearly what I've been writing since the dinosaur era, that 6-5 shots win more than 7-5 shots who win more than 8-5 shots, etc.

For a complete index of all our articles and research projects, go to www.trpublishing.com and click on Meadow's Racing Monthly.

[thoroughbred]

Originally posted by BMeadow
Do horses win in proportion to their odds? Absolutely. In the January 2000 issue, we printed the results of a research project completed by Jim Cramer of Handicappers Data Warehouse. More than 200,000 nationwide races were checked.

The complete results are listed in the article, but here are the win percentages and ROI of various odds ranks:

Favorite 32.8 0.81
2nd Ch 20.8 0.82
3rd Ch 15.0 0.80
4th Ch 10.9 0.79
5th Ch 7.8 0.79
6th Ch 5.6 0.78

There was also a breakdown by odds which showed clearly what I've been writing since the dinosaur era, that 6-5 shots win more than 7-5 shots who win more than 8-5 shots, etc.

For a complete index of all our articles and research projects, go to www.trpublishing.com and click on Meadow's Racing Monthly.

-----------------------------------------------------------------------
BMeadow,

Could you please give me your take on what appears startling to me, that the ROI between the favorite and the 6th choice only differs by 0.03?

Also, If I'm reading the ROI numbers correctly, (and I'm not sure that I am), it looks like it hardly makes any difference at all wheher we bet the favorite or any other horse up to the 6th choice. Is my interpretation correct?

Thanks.

[andicap]

Well, the losing streaks would be a helluva longer betting the 6th choice than the favorite. If you're betting percentage of bankroll you would finish with less money.
Even betting flat amounts, would you really still be betting the same $100, say, after losing 15 straight races that's very likely to happen with a 6% win rate.
So when you do win, you'll win less because after all those losses, you'll have taken your bet size down a few pegs.

[ranchwest]

My position is that we're supposed to be looking for winners, not horses to match an odds profile. Whether a winner pays $4 or $104, he's the winner. You can look for stability or you can look for value, but there's only one winner.

With that in mind, many of you have suggested that a better road to an improved ROI is to relax your expectations for percentage of winners. 14 to 18% is what many are suggesting.

So, here's my question, and I believe it is significant. What percentage of picks should win at 6-1 odds or above? In other words, let's say, for illustrative purposes, that I hit 17% winners and have an ROI of 20% or whatever figure I'd be satisfied in having. Should that 17% be split as 3% longshots (6-1) and 14% others or 4/13 or 8/9 or what?

It takes less money to affect the odds of long priced horses, so I'm thinking that rather than looking just at ROI, I should also be looking at percentage of longshots. It seems to me that this would be especially useful in analyzing spot plays where sample sizes may be small. If I'm hitting a good percentage of longshots, then chances are sooner or later I'll hit a 40-1 horse and my ROI will go up.

Thoughts?