Horses going off at 5/2 odds - Large Sample

[JimBayle]

A customer of mine told me about the Barry Meadows thread concerning the “stat” that 5% of players are winners. I stayed up late reading the thread and only made it about halfway through.


I was impressed that there are some serious, well informed opinions expressed on the thread. I will definitely read more when I have time.


I did read one thing that was incorrect. Someone posted that 5/2 shots win more often in small fields and less often in large fields.



I thought some of you might enjoy seeing the actual numbers from a fairly large sample.



Looking at over 300,000 races from 2001 through July of 2006 at tracks in North America there were about 209,000 races without any 5/2 odds horses. This left us with 98,544 races with 116,058 horses going off at 5/2 odds.



Note: I defined 5/2 odds as >2.49 odds and <3.0 odds. Horses going off at 2.5, 2.55, 2.6, 2.65, 2.7, 2.75, 2.8, 2.85, 2.9, and 2.95. (Some tracks report odds to the nickel)



To give us reasonably big samples I broke the 116,058 horse with 5/2 odds into three groups.



22,450 horses were in fields of less than 7 horses (group A)

46,291 horses were in fields of 7 or 8 horses (group B)

47,317 horses were in fields of more than 8 horses (group C)



5/2 ODDS (>2.49,<3.0)



Group A horses won 22.23% of their races, average odds 2.701

Group B horses won 22.29% of their races, average odds 2.702

Group C horses won 22.17% of their races, average odds 2.704



Considering that the Group C horses had slightly higher average odds I consider it basically a wash. More proof that horses run “to their odds”. You would find this true in 5/2 odds horses in Maiden races as well as Stakes races, sprints as well as routes etc.(assuming your sample is large enough)



A more accurate way of comparison might be to look at horse’s win percentage by exact odds. In the above sample the largest group (by a small margin) was those with final odds of exactly 2.8-1.



The danger here is that we are looking as a smaller sample, something I am loath to do, but here are the numbers.



4,018 horses were in fields of less than 7 horses (group A)

8,694 horses were in fields of 7 or 8 horses (group B)

9,017 horses were in fields of more than 8 horses (group C)



EXACTLY 2.8-1 ODDS



Group A horses won 20.83% of their races

Group B horses won 22.13% of their races

Group C horses won 21.42% of their races



For those of you curious about the ROIs:



Betting the 116,058 horses going off at 5/2 odds: win lost 17%, place lost 14% and show lost 11%. Remember a $20 show bet at Mountaineer (into the pool) might influence the show price. Calculating how much your bet will influence the final price is an almost impossible task as you don’t know how much your bet will influence other player’s wagers.

Jim Bayle

[ryesteve]

Good stuff... thanks

[andicap]

Jim,
Thanks for posting -- hope you stay around and contribute more....

[Valuist]

Makes some sense, at least in the bigger field races. A 5/2 shot may be the favorite in a big sized field, albeit a very shaky favorite. Seen that we almost never see a morning line favorite higher than 3-1 one may conclude that these horses are "ice" (dead on the board) and ice horses don't tend to run especially well.

[Celent]

This goes a long way...I mean a LONG way....to show just how efficient the win pools really are. No wonder so many of us are desperate to find relief in the exotic pools....

Mike

[kenwoodallpromos]

I had to digest the stats for a day.
I tink it means that the top contenders in any race without a strong favorite are competitive regardless of the field size or odds on the 2nd or 3rd choice, otherwise maybe the % loss on place and show would be much less.
Beat the favorite, find a 2nd or 3rd favorite who is an overlay!

[slotterhaus]

I'd be interested in an ROI breakdown of 5/2 types according to race condition/distance. For example, what are the relative merits of older males going 6f under starter allowance conditions v 20K 3yo fillies at 7f? Obviously, we're talking about much smaller samples but I'd be interested in the limited data.

Slightly off topic, I still think win odds, although quite efficient in their own right, are overbet when patrons reach into the 3 & 4 horse single race exotics. The concentration of equal weighted boxes in these pools filled with the top win contenders is, in my view, one of the few enduring gifts in the game.

[Vegas711]

This goes a long way...I mean a LONG way....to show just how efficient the win pools really are. No wonder so many of us are desperate to find relief in the exotic pools....

Mike


Taking nothing away from Jim , but I have to strongly disagree that the win pools are efficient, if this were true we would all lose 17 % and the game would be called on account of empty wallets. The exotic pools if anything result in the ruination of many players. Would you rather try to overcome a 17% or a 24 % disadvantage?.


The KEY to the game is in the methods you use to attack the game. You have to be different from the crowd with 1 requirement your method needs to be very effiecient in predicting race outcomes. If you are going to play the game from the view point of what you read in some handicapping book or you listen to all the false ideas that are preached at your local track, you will lose, period, end of story. Example.

Here is one bit of useless info, look at who is riding your horse, if he is not one of the better jockeys pass on betting this horse.............. this is pure garbage. Like a 110 pound rider is going to control a horse.... sometimes it take 4 large gate crew men to get a horse in the gate. I can only tell you that in the last 5 plus years i cannot recall the name of one rider on any horse that I bet, guesse what it has NOT cost me one dime in losses.Same goes for trainers and all the other crap that people say is important.


All these beliefs are useless becouse either they never were effiecient in predicting race outcomes or too many players are using them.

[Vegas711]

The Problem with statistics is that they look at data from a single point of view. Example in this case how do 5/2 odds horses do in 7 horse fields? What they do not or can not take into account are all the numerous factors which are occuring simultanously. How would these stats change if we added one more requirement like how do 5/2 odds horses fare when they are the only speed, then next lets add another element how do 5/2 odds horses that are the only speed fare in races that are predictable as compared to races which contain different degrees of uncertainity?


Horse races are not single factor events but rather a collage of hundreds or thousands of interacting events. The crowd if anything has only shown that they do not understand the Game and that they have stopped improving on their ability to select the outcomes of the races. 40 years ago they predict the winner 33 % of the time , today they STILL predict the winner the SAME 33 %....... does this sound like they know any more today than they did back then?

[RXB]

40 years ago they predict the winner 33 % of the time , today they STILL predict the winner the SAME 33 %....... does this sound like they know any more today than they did back then?

Actually, it's a little higher than 33% now, but that can be attributed to reduced field sizes.

Handicappers now have much more sophisticated information than their predecessors 40 years ago. From that standpoint, we certainly know more now than then. But the reliability of horses' form has declined during the last four decades-- so from that angle, we know less. Kind of evens out, really.

[Overlay]

Even if 5-2 shots (or horses in any other odds range, for that matter) win as often in the aggregate as their odds say they should, there will still be individual cases within that large group where the odds at which the horse is being held represent an overlay or an underlay, based on the horse's actual chance of winning. The handicapper's challenge still lies in being able to tell when the posted odds offer value and when they do not.

[classhandicapper]

Even if 5-2 shots (or horses in any other odds range, for that matter) win as often in the aggregate as their odds say they should, there will still be individual cases within that large group where the odds at which the horse is being held represent an overlay or an underlay, based on the horse's actual chance of winning. The handicapper's challenge still lies in being able to tell when the posted odds offer value and when they do not.

I agree with this 100%, but I'm willing to bet that personal odds lines will NOT be totally accurate in situations where they disagree with the pubic odds either.

In other words....

If you make a series of horses 7-5 on you own odds line and they are all 5-2 on the board, if you are a very sharp handicapper/oddsmaker they will win somewhere in between what those two lines suggest and be less profitble than your line suggests.

If you make a series of horses 4-1 on you own odds line and they are 5-2 on the board, if you are very sharp handicapper/oddsmaker they will probably win somewhere in between what those two lines suggest and perform better than your line suggests.

The reason is that as individuals we are almost always working with incomplete and occasionally inaccurate information, not to mention imperfect understanding of many situations. That's why a margin of safety is very important.

[losealot]

You sure brought back some memories, Jim. I still go over your work from time to time.

I'm pleased that you're alive and hopefully, well. And still as incisive as ever.

Losealot is Tony from Springfield, MA. After studying your work, "Loseless" and sometimes "Winmore", would be more accurate user id's. Your work improved every facet of my handicapping. Thanks again, Jim!

[Hosshead]

I agree with this 100%, but I'm willing to bet that personal odds lines will NOT be totally accurate in situations where they disagree with the pubic odds either.

In other words....

If you make a series of horses 7-5 on you own odds line and they are all 5-2 on the board, if you are a very sharp handicapper/oddsmaker they will win somewhere in between what those two lines suggest and be less profitble than your line suggests.

If you make a series of horses 4-1 on you own odds line and they are 5-2 on the board, if you are very sharp handicapper/oddsmaker they will probably win somewhere in between what those two lines suggest and perform better than your line suggests.

The reason is that as individuals we are almost always working with incomplete and occasionally inaccurate information, not to mention imperfect understanding of many situations. That's why a margin of safety is very important.
I agree.
Your first example are overlays (compared to your own personal oddsline).
Your second example are underlays (compared tro your own personal oddsline).

Question is, which one shows a profit (+ROI) ?
Or is there another (successful) factor(s) that would sometimes point to the overlay and sometimes point to the underlay?
In which case, that factor would need to be incorporated into the persons original oddsline, thus starting the cycle all over again, with "new" odds .

The never ending "Tweaking" of oddslines. :bang:

[Overlay]

I agree.
Your first example are overlays (compared to your own personal oddsline).
Your second example are underlays (compared tro your own personal oddsline).

Question is, which one shows a profit (+ROI) ?
Or is there another (successful) factor(s) that would sometimes point to the overlay and sometimes point to the underlay?
In which case, that factor would need to be incorporated into the persons original oddsline, thus starting the cycle all over again, with "new" odds .

The never ending "Tweaking" of oddslines. :bang:

Since underlays by definition are horses with odds that are lower than their true chances of winning indicate, then even if an underlaid horse had such a factor in its composite profile, how could it show a profit on a long-term basis, and why would you bet it? Wouldn't betting such horses consistently have to result in a loss (again, by definition)? Or, if a single factor would point to overlaid horses (even on an intermittent basis), wouldn't it soon lose that edge through recognition of that by the public, and subsequent overbetting? I'm not saying that such factors don't have a place in the handicapping model, but I think they have to be linked to the winning chances of the horses in question, to permit laying off in situations where the odds are too low.

[BeatTheChalk]

Not nearly enough work was done by the researcher...We need lots more
races LOL LOL :lol: :D :jump: :bang:

[Vegas711]

Not nearly enough work was done by the researcher...We need lots more
races LOL LOL :lol: :D :jump: :bang:

Apparently you don't get it. Statistics no matter how large the sample may be decieving.

[kenwoodallpromos]

Apparently you don't get it. Statistics no matter how large the sample may be decieving.
I would say stats can be used in a decieving manner. Numbers are just numbers until they are interpreted.

[JimBayle]

Thanks a lot Losealot/Tony – good to hear my depressing stats helped someone along the way.



Some good posts here – some serious thinking going on.



Seems no one has a problem that horses going off at 5/2 odds win about 22.2% of their races, regardless of the field size.



LET’S DISCUSS TWO IDEAS …



IDEA #1

What if I made the following statements…



Horses going off at 5/2 odds with GREAT trainers win about 22.2%

Horses going off at 5/2 odds with GREAT jockeys win about 22.2%

Horses going off at 5/2 odds that are the lone speed win about 22.2%

Horses going off at 5/2 odds that are the favorite win about 22.2%

Horses going off at 5/2 odds that have the best Beyer Speed Number in their last race win 22.2%



IDEA #2

Let’s assume a person’s database shows a sample of 100,000 horses going off at 5/2 odds win 22.2% of the time.

Now this person restricts his database query to horses going off at 5/2 odds, with a low percentage trainer, a low percentage jockey, no speed in last three races and no good Beyer’s in the last three races.

Let’s assume this query shows a much smaller sample and that this group wins only 19% of the time.



Question is – how much can we trust the smaller sample? What if the two samples were based on long-shots instead of 5/2 shots?



Comment on these two ideas and I will make another post in a couple days telling you what I THINK.

[losealot]

Ok, Jim, I'll bite. Here are my guesses...The 19% group would have average odds somewhat higher in your $2.50 spectrum. I'd say around 2.90...the "better" group around 2.60...

Next I'll guess that the ROI's for the two groups would be very close to equal. This is not to say that one can't go into your large overall sample and ferret out a small sample that would, say, break even. But to find such a smaller group is virtually impossible by adding just one factor, especially if that factor is easily known by the general public.

It would take a combination of factors, each reducing the sample size, and at least one of these factors should be proprietary...that is, virtually unknown to the general public.

Tony Losealot

[losealot]

A post script to above reply...You taught us a long time ago, Jim, that the average odds of a given group are ALMOST NEVER the same as the average odds of the winners for that given group. So my 2.60 group probably would have winnings odds very close to 2.50 while the 2.90 group's winners would average around 2.85, making disparity in payoffs even greater and thus evening out the win % difference...bring the ROI's very close together.

Hey, if I flunk this test does this mean I have to take your course all over again?

Regards, Tony Losealot

[chickenhead]

IDEA #2
[font=Arial][size=3]Let’s assume a person’s database shows a sample of 100,000 horses going off at 5/2 odds win 22.2% of the time.

Now this person restricts his database query to horses going off at 5/2 odds, with a low percentage trainer, a low percentage jockey, no speed in last three races and no good Beyer’s in the last three races.

Let’s assume this query shows a much smaller sample and that this group wins only 19% of the time.



Question is – how much can we trust the smaller sample? What if the two samples were based on long-shots instead of 5/2 shots?



Seems to me you get a lot of information by seeing how this angle does on horses in all of the odds ranges around the 5/2 shots.

For instance, here is an angle for dirt sprints, from 1 years worth of data:

#R E A ODDS A/E
88 50.5 30 0-0.99 0.59
307 120 78 1-1.99 0.65
306 87 61 2-2.99 0.70
184 42 29 3-3.99 0.70
102 18.5 12 4-4.99 0.65
52 8 6 5-5.99 0.76
40 5.3 3 6-6.99 0.57
29 3.4 2 7-7.99 0.58
45 4 2 >8.00 0.50

I might not trust an angle based on just a single odds range, and this is a small sample to boot, but using the whole range of possible odds as a "force multiplier" gives me much more confidence that these are good horses to bet against.

[Overlay]

Let’s assume a person’s database shows a sample of 100,000 horses going off at 5/2 odds win 22.2% of the time.

Now this person restricts his database query to horses going off at 5/2 odds, with a low percentage trainer, a low percentage jockey, no speed in last three races and no good Beyer’s in the last three races.

Let’s assume this query shows a much smaller sample and that this group wins only 19% of the time.



Question is – how much can we trust the smaller sample? What if the two samples were based on long-shots instead of 5/2 shots?



Comment on these two ideas and I will make another post in a couple days telling you what I THINK.

My comment/opinion would again be that, although the public shows a remarkable record of aggregate accuracy in setting odds, I don't view those odds themselves as determinative of a horse's chances. Rather, the odds are set as a result of the public's evaluation of factors that are directly related to winning probability (like the ones you cite, such as trainer, jockey, early speed, and overall speed (among others)). It is those factors that (again, in my view) determine the horse's true winning chances. Despite the small sample you're using, I think that the hypothetical lower winning percentage of the group you describe is consistent with that view. The public has given an excessive weight in that instance to factors that don't warrant the degree of confidence that the public has placed in them. Such a lower winning percentage would result whenever the public would overrate factors (regardless of the odds range involved), although the proportional effect might be greater at higher odds ranges, due to the smaller expected winning percentages associated with higher odds.

[Vegas711]

Jim, your original test which began this discussion is fine since you are only comparing how 5/2 odds horses fair as they pertain to field size. My objection was to another poster who i felt made a wrong conclusion from your stats.

The most important part of stats is not doing them but making accurate conclusions.

It is my opinion that horse racing has too many variables that are interacting simultanously that stats are for the most part not very useful.

Lone speed is a subjective factor your study may say lets look at horses going off at 5/2 who had the lead, thus Lone speed, what if this horse is going very fast compared to what he normally runs. What if the horse is not normally a front runner( these horses are running away from their confort zone, they fare worse than their counter parts).


People forget or they just do not want to accept that horse racing is just a bunch of animals running around circles. If they feel like running, they will.

[punteray]

I love the manipulation of numbers!! Russia and the U. S. played a sporting event. The U. S. won. In the Russian Newspaper, the following day, it read "Russia came in second, The U.S. came in next to last." Truly, it is how you perceive it! Enron learned that!

Ray

[losealot]

Wondering if I flunked the test on horses going off at 5/2?

[StartedAt18]

When the Chalk is 5/2 in a field of 9 or less, you have to bet against him, he's not going to win! Now if there are 10+ horses, you can't leave that 5/2 out. It costs a lot of money to make that horse 5/2 with 10+ horse field. It's just a pattern that I've noticed. Example would be Woodbine! Man.. when they have those huge fields, they run those 5/2 all the damn time! Pissess me off! But yes what you say is true, 5/2 odds win more often in big fields. In a small field.. if the favorite is at 5/2 and he wins, that's gift man! That never happens unless there are like 3 other horse 5/2-7/2 in the race.