How do the Europeans Handicap?

[Capper Al]

Originally Posted by TonyMLake

Don't get me wrong.... there's some possible fluctuation... power laws like Benford's Law and Zipf's law are all power laws that range ABOUT

1 = 30%
2 = 18%
3 = 12%
4 = 9%

but are adjusted by a small multiplier... for example, in horse racing, the favorite is closer to 33%... so there's some "fudge factor" even over gigantic sample sizes.

Basically, it happes this way because there is a natural law which enforces
"rank versus frequency" ... these laws are unalterable and MUST be adhered to ESPECIALLY if the betting public has become more informed.

The only way they have been "improved" is in the case of special circumstances... for example, large statistical fluctuations or large scale cheating (far, far more likely).

I encourage any of you to study up on Benford's Law or Zipf's law. It really has nothing to do with what you are all saying but is much more related to the simple fact that a hysterical system tends to naturally find the "favorite" about 1/3 of the time and (using fractal regression) 1/3 of the time on down the line from that. Any DEVIANCE from that doesn't indicate an educated public but more likely someone with "inside knowlege" - ie, cooking the books in accounting, or, doping or nose spongeing a horse in horse racing.

I'm so sorry folks, the recent uptick in favorites being in the money (if it even exists) is NOT a mathematical likelihood and is FAR more likely to indicate wide spread cheating.

It's because of the way numbers distribute on a logarithmic scale. It's really pretty basic statistics.

Please don't take my word for it, I've give you the keys. Unlock the door by examination. I'm not going to offer some kind of passionate defense, but I used to teach college algebra, so if any of you actually really want to know I will try to help you understand. Not gonna get into a debate over 200 year old math, though... hehe...

This is amazing. I would of never believed it.

[thaskalos]

Originally Posted by TonyMLake

I understand EXACTLY why it would seem that way, but actually, the smaller field sizes make the predictive power laws MORE on target (ie, closer to 33% for the highest ranked horse). Please read up on Benford's law.

If you have an accurate data set that shows smaller field sizes with precision higher than 33%, this is a heavy indicator of CHEATING.

It's exactly how the feds catches cheaters in bookmaking, I promise. Read up on Benford's law, Zipf's Law, and Power Laws in General.

I'll try to help.

I don't understand. Are you suggesting that the favorite is as likely to win a 4-horse race as it is to win a 12-horse race? Wouldn't the increased competition diminish the winning chances of each individual horse?

And what if they suddenly started running match races? Wouldn't this cause the percentage of the winning favorites to increase further still?

[classhandicapper]

Originally Posted by Light

Why do they race like that? And why during the course of the race did none of the horses drift into the middle where there was a lot more room to run unimpeded by traffic? Never seen that before.

I think it's likely the riders believe some sections of the course are superior to others, but they can't all fit in one area, so they split.

[classhandicapper]

Originally Posted by thaskalos

I don't understand. Are you suggesting that the favorite is as likely to win a 4-horse race as it is to win a 12-horse race? Wouldn't the increased competition diminish the winning chances of each individual horse?

And what if they suddenly started running match races? Wouldn't this cause the percentage of the winning favorites to increase further still?

We know with 100% certainty that in 2 horse fields favorites have to win greater than 33%.

We know that with any skill at all in 3 horse fields favorites have to win greater than 33%.

I feel comfortable about that in 4 horse fields too.

The counter is that large fields (call it 12) sometimes contain some really terrible horses with almost no chance of winning. So adding 4-5 horses with a cumulative expected win% of 5%-10% doesn't have much impact on the win% of the favorite. The field may be bigger, but it's no biggie. Obviously though, if you add 4-5 horses and they are all really good, that would be different.

[Tom]

Please, I beseech you, learn these very basic statistical skills. I'm willing to help, but I'm not gonna listen to a lot of flames.

Then provide real world data to prove your point. Math is no substitute for reality.

[castaway01]

Originally Posted by TonyMLake

IT MAKES NO DIFFERENCE IF THERE ARE 5 or 18 HORSES.

Please, I beseech you, learn these very basic statistical skills. I'm willing to help, but I'm not gonna listen to a lot of flames.

I'm stealing this data from a post by Ron Tiller on 10/28/14.

I'm sure you'll find some way to deny the numbers, but 41% winning favorites in 6-horse fields and 32% winning favorites in 10-horse fields flies directly in the face of what you are adamantly claiming.

In the real world, you're 100% wrong.

Favorites in 6 Horse Fields
Year St W W%
2000 9248 3461 37.42
2001 8709 2960 38.28
2004 7792 2949 37.85
2005 8254 3210 38.89
2006 8342 3285 39.38
2007 8020 3125 38.97
2008 8104 3203 39.52
2009 7600 3029 39.86
2010 7274 2972 40.86
2011 7752 3151 40.65
2012 8606 3588 41.69
2013 8367 3386 40.47
2014 7820 3216 41.13



Favorites in 8 Horse Fields
Year St W W%
2000 11873 3908 32.92
2001 11991 3926 32.74
2002 11655 3843 32.97
2003 11629 3941 33.89
2004 11323 3830 33.82
2005 11411 3823 33.50
2006 11262 3791 33.66
2007 11016 3813 34.61
2008 10611 3692 34.79
2009 10463 3748 35.82
2010 9772 3547 36.30
2011 9627 3432 35.65
2012 9386 3327 35.45
2013 8966 3169 35.34
2014 7163 2540 35.46


Favorites in 10 Horse Fields
Year St W W%
2000 9192 2711 29.49
2001 9285 2869 30.90
2002 10275 3156 30.72
2003 10300 3232 31.38
2004 10031 3125 31.15
2005 8818 2705 30.68
2006 8341 2594 31.10
2007 8412 2632 31.29
2008 8441 2684 31.80
2009 8288 2711 32.71
2010 7624 2546 33.39
2011 6647 2147 32.30
2012 6087 1956 32.13
2013 5747 1861 32.38
2014 3709 1187 32.00

[castaway01]

Originally Posted by Tom

Then provide real world data to prove your point. Math is no substitute for reality.

I posted some of that data here, thanks to Ron Tiller, from the "Favourites Hit 32% Year In Year Out. Why?" thread from last October.

[biggestal99]

Originally Posted by TonyMLake

IT MAKES NO DIFFERENCE IF THERE ARE 5 or 18 HORSES.

Please, I beseech you, learn these very basic statistical skills. I'm willing to help, but I'm not gonna listen to a lot of flames.

Math is great but reality is better. 4 horses field at aqueduct on inner are NOT equal to 20+ horse heritage handicaps in the UK. You know math but horse racing well you need a primer.

If 33% of favorites won heritage handicaps its a sure trip to easy street. :-)

Allan

[traynor]

Originally Posted by Tom

Then provide real world data to prove your point. Math is no substitute for reality.

One of the reasons I am so underwhelmed by the pointification of holders of "advanced degrees" in statistics or mathematics. By my third year in graduate school I had learned enough to realize that the majority of it was worthless (or nearly worthless) for handicapping horse races.

Some of the absolute worst "race analysts" I have encountered held the most impressive sounding degrees and "academic credentials." They suffer as much as anyone else from the "to a carpenter with a hammer, everything looks like a nail" syndrome. And are guilty of ignoring "compelling disconfirming evidence to the contrary" a bit more than most.

[DeltaLover]

I am not sure 100% of what Tony is trying to say here..

I cannot believe that he claims that the favourite wins with the same frequency regardless of the field size. Probably he means something different..

Can you please clarify?

[Dave Schwartz]

Please don't take my word for it, I've give you the keys. Unlock the door by examination. I'm not going to offer some kind of passionate defense, but I used to teach college algebra, so if any of you actually really want to know I will try to help you understand. Not gonna get into a debate over 200 year old math, though... hehe...

Excuse me, Mr. Lake. But you are in tougher company here than I think you know. Around here there are guys with PhDs in math and science. Even a couple with multiple PhDs. I doubt that anyone is deeply be impressed because you taught college algebra.

I find that your theories are, at this point, purely anecdotal.


Respectfully,
Dave Schwartz

[lansdale]

Originally Posted by TonyMLake

It's always approximately 33%. There are people with limited mathematics training who will get excited and point out they believe it is higher than that now in the United States or over the last ten years or within some other (relatively) small subset of data, but even if it were true its only a matter of time before there is a regression to the mean.

It has been and will always be about 33% (and similar predictions for the second and third favorite) as an artifact of a mathematical principle known as Zipf's Law.

Hi TL,

We had a discussion of Benford's Law on the site some months back which touched on some of these issues. I think you're making an important, little-recognized point on the value of playing rank vs. frequency here, but you must realize that Zipf (which is identical with Benford) doesn't come close to reflecting the distribution of winners in U.S. racing, as most of the responses to your post suggest.

However, you mention that your play is primarily in Europe where, according to John Dineen, a very knowledgeable and math-oriented player who sometimes posts here, the Zipf (or Benford) distribution, does in fact, obtain for fields of 8-15 horses. He used research in this area to profit on European horse shipping to the U.S. for the last Breeder's Cup.

Trifecta Mike, a math professor who formerly posted on the site, suggested in the Benford thread, as well as in the picks he posted, that playing rank vs. frequency is part of his own handicapping process. Although he's unlikely to appear, it would be interesting to have his comments on this.

I am far from being a mathematician, but looking at the frequencies of U.S. racing, they appear to more closely resemble the Pareto distribution than Zipf. When we discussed this before, we mentioned exactly the same issue that you raise - that the difference reflects the element of cheating, specifically drug use by trainers. But of course this is speculation.

http://vendire-ludorum.blogspot.co.u.../label/Benford

Cheers,

lansdale

[Luckycreed]

Field size DOES matter Here are some stats on Australian racing for city tracks since 1991.

All races 74490 favourites 23151 winners 31.07% strike rate average field size 11

Field size eight or less favourites 15866 winners 5962 37.57% strike rate

Field size 16 or more 6210 favourites ran 1601 won 25.76% won.

the more horses in the race, the bigger the chance the fav gets beat that ain't rocket science.Favourites win over 60% in Aistralian football betting as you would expect in a two horse race.

[Luckycreed]

Also I don't think algebra has anything to do with it, club handicappers (race secretaries) program for competitive racing and around one in three fabs winning is what they aim for because it maximises wagering.

Too few favs and people think the game is to hard too many and all those short priced winners turn them away.If the fav starts winning or losing too much they simply start fiddling with the class system and or start using weight more aggressively which is what Australia did when we moved to ratings based handicapping a few years back

Falling field sizes might be a bit harder to deal with however,that is something the US industry really needs to try and gets it's ahead around or you guys could be heading to 40% favourites like oz harness racing.

[romankoz]

Hi Luckycreed,
I seem to remember an Oz pacer of that name in my youth. Is that correct? If so you must be of my vintage.

Anyway you are spot on. The 33% figure used by many are the sum total of all the odds brackets. All favourites paying 1/2 will win many more times than 7/2 fvaourites and all those in between will range accordingly. In the finish it averages out to 33% give or take variances in some years when you add them all together and divide by the number of odds.

It is the same in smaller fields (shorter priced favourites) compared to larger fields (longer priced favourites) usually.

I kind of find it difficult to see why realising smaller fields favs have a higher strike rate is so hard a concept to grasp.