Favourites Hit 32% Year In Year Out. Why?

[flatstats]

I only have GB stats. I assume US stats (and any other country) are the same so excuse me if I am totally in the wrong ball park here but why are favs so consistent and why do they win 1 in 3 races year in, year out?

Favs are consistent. For the past 10 years on the flat turf the fav record is:

2005 30% strike rate, 0.94 A/E
2006 32%, 0.86
2007 33%, 0.97
2008 33%, 0.94
2009 32%, 0.94
2010 32%, 0.92
2011 31%, 0.92
2012 31%, 0.93
2013 33%, 0.94
2014 32%, 0.92 (to 25-Oct-14)

The A/E does vary a bit but it is probably within expectation. But why is the strike rate so consistent? Why is it that 1 in 3 favs win?

Note that fav strike rate varies with field size (the bigger the field size, the lower the strike rate) so here is the clincher:

Field sizes have shrunk over the years so why has the strike rate for favourites remained near enough consistent?

2005 11.3 runners
2006 10.7
2007 10.8
2008 10.6
2009 10.9
2010 9.8
2011 9.8
2012 9.8
2013 9.2
2014 9.0 (to 25-Oct-14)

Average field size has dropped from 11.3 in 2005 to 9.0 in 2014. This should mean favs hit a higher strike rate and yet the strike rate today is near enough the same as it was ~10 years ago. OK 2005 looks an exceptional year but there is a downward trend with field size but a level trend with fav strike rate! Why?

[raybo]

The hit rate over here is even higher and getting higher, it appears, as time goes by. The reason that hit rates on favorites are as consistent as they are? Well, the combined public determines the favorite, so if the public remains just as good at determining the favorite, then the hit rate on those will remain consistent. If the public gets better, then the hit rate will increase. It's fairly well agreed that the public is not getting worse, and probably is getting better, due to the proliferation of decent computer analysis apps, the wider availability of quality data, and losses in the number of bettors (at least in the US, and I assume many of those lost bettors are on the lower end of the spectrum, regarding analysis ability), among other things. The "A/E", if that means "average earnings", will tend to decrease as field size decreases, and as the public gets better at their win selections, and as takeout increases.

Fewer below average players, smaller fields, better analysis tools, etc., add up to more winning selections, and lower average payouts on the public's favorites.

[Stillriledup]

I think it hits around 33 because 20 is too low and 50 is too high, so it lands in the middle somewhere. The public is always going to be right more than 4 out of 5, there are too many races with "sticks" that skew that rate and 50 pct is too high, there are just too many big fields and competitive races to pick half winners...so, no matter how good the bettors get, the numbers will seem to land in the low 30s.

[flatstats]

Ray. The A/E is the actual wins / expected wins stat. This compensates for different prices and is a good indicator of valueness.

The A/E is not affected by field size because usually the bigger the field size the bigger the odds etc.

But the question here is why are favs still hitting 32% when field sizes are getting smaller?

Assume field sizes shrink even further. What if there were only 5 runners per race. Would it not be strange if favs still hit 32%?

Your point about the public getting smarter. Surely this would be reflected in the prices? So back to the A/E. If the public got smarter then they would find more winners. The prices of those winners should thus be lower and we should see a reduction A/E. But in the case of favourites it has remained a consistent 0.93 over the past 10 years.

[thaskalos]

Originally Posted by flatstats

Ray. The A/E is the actual wins / expected wins stat. This compensates for different prices and is a good indicator of valueness.

The A/E is not affected by field size because usually the bigger the field size the bigger the odds etc.

But the question here is why are favs still hitting 32% when field sizes are getting smaller?

Assume field sizes shrink even further. What if there were only 5 runners per race. Would it not be strange if favs still hit 32%?

Your point about the public getting smarter. Surely this would be reflected in the prices? So back to the A/E. If the public got smarter then they would find more winners. The prices of those winners should thus be lower and we should see a reduction A/E. But in the case of favourites it has remained a consistent 0.93 over the past 10 years.

You said previously that you are using only GB turf starts to form your conclusions...because you have no access to U.S stats. The racing scene in this country is noticeably different. Here, not only has the winning percentage of the favorites increased in recent years...but the favorites' collective mutuel payoffs have declined as well. About a decade or so ago, you could bet blindly on the favorite in each race, and expect to lose at a rate of only HALF the takeout that the tracks charge us for playing. You actually beat half the takeout without doing any handicapping at all! Alas, that is no longer the case. Betting on the favorite now costs you the entire takeout +.

So there is no doubt that the betting public here is getting "smarter".

[Augenj]

My statistics for major U.S. tracks vary wildly between tracks as shown in the following link. The numbers are pretty dynamic year to year even for the same track.

Top Horse Analytics: 10/21/2014 Weekly Statistics

[raybo]

Originally Posted by Augenj

My statistics for major U.S. tracks vary wildly between tracks as shown in the following link. The numbers are pretty dynamic year to year even for the same track.

Top Horse Analytics: 10/21/2014 Weekly Statistics

Sure, things will vary from track to track, that is expected. Nationwide, for the whole year, the fav hit rate is probably well above the 33% that was normal a few years ago. Fields are getting smaller and the public is getting better, so hit rates will increase and average payouts will decrease.

This "A/E" stuff is beyond me, how do you arrive at expected wins unless you are using the final odds? That seems to me to just cloud the fact that the public is getting it easier in recent years because of the progressively smaller fields, and they are getting better at the same time.

Using individual meets for analysis only tells you what happened at that track, and the time period is smaller, so less consistency is to be expected.

Anybody that does not expect hit rates on favorites to remain, at least as good as previous years, if not better, is not doing much thinking, IMO.

[Overlay]

Maybe one factor contributing to the percentage of favorites staying the same as average field size gets smaller would be that more bettors are purposely steering away from the horse that is most likely to win due to the low payouts involved, and are backing other horses in the field that are perceived to be offering value, and that also have an increased probability of winning (just from random chance, if nothing else) as field sizes decrease. Thus, there would be more competition for the favorite position, which would have the effect of counterbalancing the increased percentage that one might expect to see as field sizes decrease if people were focusing solely on finding and betting the most likely winner.

[MJC922]

Not sure what others are seeing coming out of their own DBs but I'll share what I have. Unless I've made a mistake here's what things look like for North American Thoroughbred races over approx. 5 years between 2008 and 2012:

FieldSize WpctFav AvgOdds
3 59% 0.62
4 50% 0.81
5 46% 1.01
6 41% 1.20
7 38% 1.35
8 37% 1.49
9 34% 1.61
10 33% 1.74
11 31% 1.85
12 30% 1.95
13 31% 2.07
14 26% 2.18

Avg field size was 8.1 by the way.

And there's this from a more official source which seems to be close enough to mine, I believe overall they came up with 36.7 pct:

http://www.drf.com/blogs/more-favorites?page=4

[lansdale]

Discussed this issue in the Benford's Law thread with J. Dineen recently. The European frequency of winning favorites is lower and the distribution of all favorites more spread out than in U.S. racing. As MJC922 points out (and Dave Schwartz has pointed out for the past few years) the mean of U.S. winning favorites in recent years is ca. 37% - about 4%-5% higher than the European (and global) rate, as cited above. It's also higher than the mean historical rate of ca. 33%.

Trifecta Mike has compared the current distribution of favorites to that of Pareto's Law, while the European distribution, as J. Dineen has pointed out, closely matches Benford's. The question I raised is, what the cause might be, and conjectured, like some here, that smaller fields might be a possibility, but if the difference in mean field size is only one horse, I think it's difficult to make that case. I also wondered about the increasing presence of illegal drugs in the U.S.. I believe TM cited the influence of rebates, which is seems very plausible.
Regardless, the difference is striking, and certainly makes U.S. racing seem like a relatively bad bet.

[lansdale]

Originally Posted by MJC922

Not sure what others are seeing coming out of their own DBs but I'll share what I have. Unless I've made a mistake here's what things look like for North American Thoroughbred races over approx. 5 years between 2008 and 2012:

FieldSize WpctFav AvgOdds
3 59% 0.62
4 50% 0.81
5 46% 1.01
6 41% 1.20
7 38% 1.35
8 37% 1.49
9 34% 1.61
10 33% 1.74
11 31% 1.85
12 30% 1.95
13 31% 2.07
14 26% 2.18

Avg field size was 8.1 by the way.

And there's this from a more official source which seems to be close enough to mine, I believe overall they came up with 36.7 pct:

http://www.drf.com/blogs/more-favorites?page=4

Hi MJ,

Thanks very much for posting these results, which accord with my instincts about how the game has changed, and reveals something which may be more significant than I thought. Possibly, I spoke too soon when dismissing field size as a possible cause of the rise in the pct. of winnings favorites. The gap of 3% between 8 and 9 horse fields is substantial, and the 9-horse mean pct. doesn't differ much from Flatstats European number for that field-size above. If you or anyone else has the results for the same period in U.S. racing, I would assume it would reflect a comparable drop in mean field size and rise in pct. of winning favorites.

Cheers,

lansdale

[MJC922]

Originally Posted by lansdale

Hi MJ,

Thanks very much for posting these results, which accord with my instincts about how the game has changed, and reveals something which may be more significant than I thought. Possibly, I spoke too soon when dismissing field size as a possible cause of the rise in the pct. of winnings favorites. The gap of 3% between 8 and 9 horse fields is substantial, and the 9-horse mean pct. doesn't differ much from Flatstats European number for that field-size above. If you or anyone else has the results for the same period in U.S. racing, I would assume it would reflect a comparable drop in mean field size and rise in pct. of winning favorites.

Cheers,

lansdale

No problem. Hope it helps. It would be substantial effort to pull 2013-2014 out of the production db at the moment so maybe someone else can provide a look at that metric. If not I will try to do it over the holidays if you're still interested.

[MJC922]

Not to veer off topic but an interesting stat I thought I'd share was the average amount of profit per winning horse dropped 6.5% from the period 2008 to 2012. That is, if you picked every winner of every race your bottom line profit was that much less. Substantial decrease IMO.

[raybo]

Originally Posted by MJC922

Not to veer off topic but an interesting stat I thought I'd share was the average amount of profit per winning horse dropped 6.5% from the period 2008 to 2012. That is, if you picked every winner of every race your bottom line profit was that much less. Substantial decrease IMO.

I would agree. That percentage decline would turn the vast majority of break-even or winning players into losing players, IMO. That's why I don't play tracks that my methods test less than 1.20-1.25 profitability. That gives me a pretty good margin to work with.

[ultracapper]

The info here seems to confirm, almost without looking any further, that the percentage of favorites winning is in direct relation to field size. 8.1 is average field size in USA, and favorites are winning at 37%. Years ago, 30-33% was the recognized norm, which correlates perfectly with the larger fields of yesteryears. An argument can be made that the handicappers aren't really getting any better at all. The game is just getting a little easier.