Kempton Betting: Is there a draw bias at the track?

Betting Strategy / Simon Rowlands / 29 October 2008 / 5 Comments

Simon Rowlands has been crunching the numbers in a bid to uncover what effect the draw has at Kempton.

Another crack at Timeform Radio the other week got me thinking about a further piece of received wisdom that came up on air, namely that highly drawn horses are favoured at Kempton.

You would expect it to be so, to a degree at least, as the numbering of horses from left to right on a right-handed track means that the highest-drawn runners are nearest to the rail and to the bends, while the lowest-drawn runners are farthest away.

That said, it can be something of a mixed blessing to be drawn near the rail in such circumstances, as horses on the outside tend to head across, causing congestion there. Even if highly drawn horses are favoured, how much are they favoured? It is good to know.

Draw analysis is a complicated matter if done properly, so I thought I would look at races at up to a mile at Kempton in the last six months and use it to illustrate not only what bias does seem to exist there but what are some of the issues involved in analysing draw data.

Firstly, it is best to analyse handicaps only, as these are races in which the chances of the runners are theoretically equalised. A horse can overcome a "poor" draw, often without much trouble, if it has plenty in hand of most of its rivals. Large samples will overcome this, but you then run the risk of missing more recent developments in what can be a dynamic phenomenon.

Then there is the matter of the numbering of stalls. It is not a simple case of regarding a draw in, say, stall 10 as being the same under all circumstances. In a five-runner race, stall 5 is nearest to the rail on a right-handed track. In a 16-runner race it is nearer the outer than the inner.

To overcome this, you need to recode the draw into a "back-to-front" draw. The stall nearest the rail is recoded as "1", the next one as "2", and so on. This is achieved on a right-handed track by (n+1-d), where n is the number of runners and d is the draw. For example, in a 10-runner race, stall 10 becomes 1 (10+1-10) and stall 3 becomes 8 (10+1-3).

You also have the issue of non-runners. In most cases, stalls are not left empty but the runners are shuffled across. In the absence of more precise information it is best to treat the highest-numbered draw as coming from the stall nearest the rail and to eliminate gaps thereafter.

It also needs to be remembered that a horse running in a 12-runner race is facing a different challenge to one running in five-runner race. Whatever your stall number, you are facing 11 rivals in the former and just four in the latter. An adjustment for field size needs to be made, in other words.

As with many statistical phenomena in horseracing, it is important not to fall for the common fault of considering just winners. If there is an effect, then it should be seen on not just the winners but the losers, and the degree to which they win and lose as well. It should also be noted that while the draw itself is a discrete entity (a horse can be drawn in one stall and one stall only) the effect of the draw is likely not to be. If stalls 1, 2, 4 and 5 are favoured then you would expect stall 3 to be also. I consider data for the stall in question and the stalls on either side as a result.

The above was all taken into account before coming up with the following figures by stall number at the various distances at Kempton. The stall numbers have been returned from the back-to-front draw to their original.

It is fair to say that there is indeed a high-draw advantage at Kempton, but that its effect seems to be far more pronounced at 5f than at longer trips. Indeed, it is arguable that too much significance tends to be accorded to the effect of the draw at beyond the minimum trip at the course.

Comments (5)

1. Wayne Bailey | 29 October 2008

   Simon, just wondering if the information on stall numbering is readily available?

   Some courses 'attach' the stalls to the inside rail in some races and the outside rail in others. In other words, Stall one may be on the inside rail in one race, while in another race, it could be beside the outside rail. (not sure if I'm explaining that too well).

   Is this data freely available?

2. Simon Rowlands | 29 October 2008

   Hi Wayne.

   I am not sure if the information is freely available anywhere, but it appears in Timeform Perspective and, I believe, in the Form Book, both of which you have to pay for of course.

   Fortunately, stall movements are not an issue at the all-weather tracks.

   Simon

3. brian | 03 November 2008

   Simon my question isnt about a draw bias at kempton but about an article you previously wrote on creating your own tissue price. Jack Houghton refers to it in his 23rd July article on the same subject, the link doesnt work. I know its been a while sice you wrote it but i've not long discovered the betfair betting strategy articles (which i think are excellent, long may they continue) and i want to digest as much information as possible on the subject. Any help or pointers in the right direction would be greatly appreciated.

   Brian

4. Tom | 03 November 2008

   Great insight once again Simon.Quite an offbeat question but since your the guy to ask re Timeform and since I'm keen on looking at unexposed sorts with little form in novice races I was wondering if you could point me in the direction, if there is one, of purchasing last seasons Timeform NHF ratings. I have recently purchased Timeform Stats Review, on your recommendation in a previous article, which has been a huge success so thank you.
   Thanks in advance
   Tom

5. Simon Rowlands | 14 November 2008

   Tom.

   Big apologies for not having got back to you sooner, but your post slipped past me unnoticed. I would have recommended Timeform's Handicap Ledger, which had a breakdown of every horse's performance and master ratings, but I can't find it on Timeform's site, so it's possible they no longer publish it. If you bought a weekly Black Book around this time of year it would have a considerable number (but not all) of the NHF ratings from last season. Failing that, Timeform might be prepared to run a report with NHF ratings and send that/e-mail that to you, though presumably at a cost. Sorry I cannot be of more help.

   Simon