Profiling The Oaks Winner 2012
Systems. Analysis. Insight. Experience. The Difference. Blogging at bettingwarbler.wordpress.com
Who are the contenders for the 2012 Oaks? What should we take into account when looking for the winner? Today on the blog, Phil Pemberton takes us through his process.
You know the old adage about lies and statistics? You don’t? Well good, read on. If you do, then just suspend your cynicism for the next five minutes - you never know you just might just enjoy yourself and learn something too.
The Police Criminal Investigation Department, Anti-Terrorism Unit, Homeland Security, and the covert units we’re not supposed to believe exist, all use profiling techniques to identify their targets. Long before they were doing it, the business-to-consumer companies and high street retailers were gathering and analysing data about customers’ purchasing behaviour, product preferences, correlation between product sales, promotion, positioning, price and a whole range of other spurious macro data such as the weather conditions, store location, day of the week. Not satisfied with that little lot, they started to data mine every fragment of data they could find out about the individuals that make the purchasing decision (usually the significant other), influence the purchasing decision (usually the significant other), and do the actually purchasing (usually the significant other) – interestingly, they don’t care anywhere near so much about the person that lifts and carries the purchased items over the threshold or assembles said purchases. They want to know your salary, profession, inside leg measurement, shoe size, religion, whether you prefer reading fictional drama to watching live sports, and even whether you prefer Davina McCall’s shiny hair to the floppy locks of Roger Federer. If you could only see what data is held about you across hundreds and thousands of databases, you wouldn’t speak to a living soul ever again let alone use your debit card for any financial transaction.
That’s a very long prologue to introduce this point: if it works (and apparently it does) on human beings, can we use profiling based on historical trend data and trainer/horse/jockey/owner preferences to identify winners of races?
Let’s look at the forthcoming Investec Epsom Oaks as a test case – which, without spoiling the ending, turns out to be quite informative. I have used summary data from the last ten years – there may be some value in going further back but a decade seems an appropriate time span, particularly as training methods themselves have changed so much in that period.
First a glance at the winners of the last ten Epsom Oaks and their starting price.
|Last 10 Epsom Oaks Winners|
|2011 - Dancing Rain (20/1)||2006 - Alexandrova (9/4 fav)|
|2010 - Snow Fairy (9/1)||2005 - Eswarah (11/4 jfav)|
|2009 - Sariska (9/4 fav)||2004 - Ouija Board (7/2)|
|2008 - Look Here (33/1)||2003 - Casual Look (10/1)|
|2007 - Light Shift (13/2)||2002 - Kazzia (10/3 fav)|
The first thing that may strike you is that only three favourites and one joint favourite have triumphed (40%) and, of the remainder, there have been some big priced winners. The mean price of the winner was 9.25/1, and the median price of the winner was 5/1 (using the median as the average price smoothes/lessens the skew of the lowest and highest outliers).
2. Elimination Through Funnelling
As in the criminal world, we can eliminate several candidates simply because they break one or more cardinal rules or golden strands of evidence. Remembering that this is only based on ten year trend data, we can put all the runners through a funnel that contains our masking criteria and only work with the horse that come out of the other end of the funnel – this is depicted in figure 1 below.
The following horses were eliminated at the preliminary funnelling stage:
1. Didn’t finished 1st or 2nd last time out: Maybe, Nayarra, Toptempo, Was
2. Have already run at Oaks trip of 1m4f: Colima, Toptempo, Vow
3. Not run in last month: Kissed (last run 33 days ago, which seems a bit harsh)
3. High Probability Factors
The initial funnelling process has deselected 50% of the fourteen runner fields, including a number of fancied horses. Although I must admit to feeling a little inflexible in strictly eliminating Kissed for having a run just two days before the cut-off date! By the time she arrived back in his box and enjoying his first roll in the hay, he’d have qualified so I might revisit that as I work through the process.
After the binary factors of stage one (i.e. you’re either in or out), there are a set of factors that have a high degree of incidence when selecting the winner – 80% or greater is the cut-off point.
There are two such 80% probability indicators: draw bias and previous winner over 1m2f .
a) Only 20% of winners overcame a low draw – 80% of winners were drawn in stalls 5+.
Consequently, horses drawn in stalls 1-4 are given a 20% (0.2) probability; others are given a 80% probability (0.8). There’s no sliding scale applied. At the time of writing the stalls draw has yet to be made so we need to revisit this criterion.
b) 80% of winners had won previously over 1m 2f.
After applying criteria in b), the horses are segmented fairly evenly in their two groups:
1. Previous winner at 1m 2f (0.8) = Coquet, The Fugue, Kailani
2. Not previously won at 1m 2f (0.2) = Betterbetterbetter, Devotion, Shirocco Star, Twirl
When we know the draw, we will have the following combinations:
Horses with a 0.8 x 0.8 (0.64) probability.
Horses with a 0.8 x 0.2 (0.16 ) probability.
Horses with a 0.2 x 0.2 (0.04) probability
4. Medium Probability Factors.
It's worth noting that, whilst 40% of favourites have won in the last ten years, only 20% have won in the last five years. Clearly being favourite isn’t a positive factor but it’s not a huge negative either.
The one qualifying factor in this group is that 60% of Oaks winners (in the last ten years) had won their previous race, so these horses are given a 0.6 multiplication factor and 2nd placed horses (worse than 2ndplace already eliminated at stage one) are allocated a 0.4 factor.
Winner of previous race (0.6) = Coquet, The Fugue, Kailani.
Finished 2nd in previous race (0.4) = Betterbetterbetter, Devotion, Sirocco Star, Twirl.
So, the running total is now: Coquet, The Fugue and Kailani are 0.8 x 0.6 = 0.48.
Betterbetterbetter, Devotion Shirocco Star, Twirl are at 0.2 x 0.4 = 0.16
If we hadn’t eliminated Kissed for her 2 days longer absence from the race course, she also has a 0.48 rating at this juncture having met all the other positive criteria.
5. Trainer Bias
The final evaluation criterion is based on trainer/course form in particular three year olds at Epsom. It is not a determining factor per se, so it does not have a weighting/probability factor attached. It is,if you like, akin to a tie-breaker in the event of a dead heat.
Sir Mark Prescott has the best record with three year-olds at the track with a 37.5% strike-rate, but he does not have a representative in the race. Of those that do: John Gosden has a strike-rate of 9.1% with his three year-olds at Epsom; Hughie Morrison 11.8%; and Sir Henry Cecil 17.9%. Mahmood Al Zarooni has only had 4 previous 3yo runners which is too few to be significant although none have won; and Aiden O’Brien who has saddled two previous winners of the Oaks is currently in a rut with no winners with his last 33 three-year-olds at Epsom.
6. Summary and Conclusions
Three horses – or four if you include Kissed – stand out on profile as the most likely winner of the 2012 Oaks: Coquet, The Fugue, Kailani and Kissed.
The two horses near the top of the market to be opposed are Vow (6/1) and Maybe (5/2F).
The draw may prove to be hugely significant in the final analysis. If any of the above horses were to be allocated stalls 1-4, then they should be removed from the final short-list (by virtue of being multiplied by a 0.2 factor).
In the absence of the draw data, Coquet and The Fugue have the proven trainer-course form although it would be a little foolish to discount either Al Zarooni, for his insignificant number of runners, or Aidan O’Brien whose recent record at the course is terrible but his current general form, especially in the classics, is impeccable.
You couldn’t confidently rule any of the short-listed runners on their position in the market –although a move for The Fugue could see her reclaim favouritism which would be a negative as far as the fit-to-profile goes. The nearest fit to the median price (5/1) are The Fugue and Kissed (Kailani currently priced at 9/1 is fairly close too).
A dutching bet on the three qualifiers (before draw bias applied) at generally available prices of 3/1(The Fugue), 9/1 (Kailani) and 25/1 (Coquet) would equate to 2.57/1.
Using his filly Twirl as a benchmark, Aiden O’Brien should know how his leading fillies Maybe and Kissed compare to The Fugue who beat Twirl by 4½ lengths when comfortably claiming the Musidora Stakes over 10.5F at York. Similarly, Mahmood Al Zarooni’s Mini dress was only touched off a short-head by Coquet in a listed race over 10F at Goodwood on 24 May, so he should have a decent handle on the chances of his runner, Kailani, against that rival.
There is a slight negative against The Fugue in that only two Oaks winners in the last ten years had run in the 1000 guineas at Newmarket. To be fair to her, she was making her reappearance in that race and was subsequently an impressive winner of the Group 3 Musidora Stakes thereafter. I would loathe to discount her on the Newmarket statistic – and, of course, 20% of winners DID run in the1000 guineas. The same negative also applies to Maybe although she is already eliminated.
If I had to pick one horse before the draw is made and at the prices currently on offer, it would be Kailani. An impressive winner of the Group 3 Pretty Polly Stakes at Newmarket over 10F on her seasonal reappearance and only her second time on a racecourse. All the other criteria fit and she is unexposed and open to any amount of improvement. The same comments could be applied to Kissed who rates the biggest danger although strictly speaking she just misses out on the days since last run criterion and her trainer has a poor recent record at Epsom with three year olds.
There’s also something appealing about backing the two horses with the initial ‘K’ although that’s no less significant that the position of Venus or the existence of the Fate line on the left palm of the Jockey!
Kailani is available to back at 12/1 with SportingBet.
A dutching bet on Kailani and Kissed at 12/1 and 13/2 respectively has equivalent odds of 3.75/1.
Good luck and, as with any bet, heed the following: if it’s successful remember where you heard it tipped first and praise his name to all that would listen; if not, remember every betting decision is yours and yours alone.
Footnote : Post stall draw considerations
With the stalls allocation for the Epsom Oaks now complete, we need to reflect the bias of the draw to our short-listed selections. The horses drawn 1-4 inclusive are scaled down by applying a probability factor of 0.2, the others drawn 5-14 are scaled by 0.8.
The following horses were adversely affected by the draw: The Fugue drawn in stall 2 and Twirl in stall 1 are both scaled down by a factpr of 0.2, the remaining qualifers were drawn in stalls 5 and above and are scaled by 0.8. The final rankings list is therefore:
1. Coquet, Kailani, [Kissed] (0.384)
2. Betterbetterbetter,Devotion, Shirocco Star (0.128)
3. The Fugue (0.096)
4. Twirl (0.032)
In short, of the protagonists, The Fugue is the big loser in the 'Stalls Tombola' and is relegated from joint top-ranking to seventh.
The good news is that Kailani (13th) and Kissed (14th) secured plum draws and the recommended bets on both is further validated.
Visit Phil's BettingExpert profile : LongTermProfits
You can follow Phil on Twitter @longtermprofits
And visit Phil's blog : BettingWarbler.wordpress.com
You must be logged in to post a comment! Sign up + or log in in the top right corner.
Very interesting analysis and good to read a different perspective on things. The race itself does seem wide open this year, with perhaps Ballydoyle not as confident as they might be about the favourite.