Betting Strategy: Learning lessons in advanced punting from a man called "Steve"...

Comments (9)

1. Stu | 15 October 2008

   Love this. Using data to de-bunk hunches is so much fun.

   I'm sure you've already read it but if not, check out Moneyball by Michael Lewis. I'm guessing it'll be reet up your street.

2. Simon Rowlands | 17 October 2008

   Hi Stu.

   Thanks very much for the kind words. I will look to do some other stuff along these lines.

   As it happens, I had not read Moneyball, despite my friend Ian Dean singing the book's praises for some time now, but have a copy in my hands and intend getting stuck in to it over the weekend.

   Simon

3. R Hills is God | 20 October 2008

   So, if Braveheart's odds on poke finishes 5th of 9 in the first and then one of his creatures, a 50/1 poke with a string of duck eggs, finishes mid-div in the next, does that make him a model of consistency? It would with the measures you've used.

   IF MJ really is so consistent why do his horses always seem to trade at a premium in the place market and why am I being premium charged for laying them there? Why not come and take some value? I'm running out of backers and fresh MJ place backers are very expensive to recruit.

4. Simon Rowlands | 20 October 2008

   I have just been informed of a post by R Hills is God but it is not showing on here. The caveats are clearly stated in the above article. The point made on the Betfair Forum has been that Mark Johnston's horses "either win or finish nowhere". There is no evidence that that has been the case recently any more than for other trainers of a similar profile. I have stated that I would "expect" a trainer whose horses are "wildly inconsistent" to show up with a large standard deviation. In order to be persuaded that my expectation is incorrect I would hope for more evidence than one example that serves a specific purpose.

5. Zee Zoo | 21 October 2008

   Simon,

   I've several horse racing queries (unrelated to this article) that I was hoping you might be able to answer.

   1. Why are races run differently on different surfaces? Specifically, why are dirt races run faster in the early stages than turf ones? It can't be just a question of avoiding kickback, because horses can and do come from the rear in dirt races (except in Nad Al Sheba!).

   2. Why is it so hard for front runners to get home at Lingfield? Surely if it was a case of front runners going for home too soon off the turn, the little people would catch onto this and hold onto their steeds for a few strides longer? I know you're no fan of pace bias', so how do you explain this one?

   3. I've noticed the Racing Post gave the Time rated going for Tipperary on August 28th as Good to Firm, despite the official going being Heavy. Are their standard times completely wrong, or do you put the quickish times down to rail movement and/or inaccurate timing mechanisms?

   4. How did Yes Sir bounce out at a breakneck pace at Cheltenham on Saturday and still manage to hang on for 2nd, just a few days after finishing well held at Wetherby? In a Flat handicap, if a front runner kicked on like that he'd be swallowed up by the field. I place laid Yes Sir, and want to avoid such pain again (I've placed a lot of the 'blame' on Peter Bowen's training skills this time!)

   5. How many cards do you study a day, and how long do you spend preparing for each one?

   If you can't answer all/any of these I'll understand as I'm fully aware of how this game we love can stretch our time budgets to the limit. If you can, then many thanks in advance.

6. Simon Rowlands | 21 October 2008

   Hi there Zee Zoo.

   Excellent questions, some of which I need longer to answer fully. The following is my best shot at this stage.

   1) The nature of the surface helps to determine the optimum way in which to expend energy so as to record the fastest possible time. I am not familiar with dirt racing, but energy distributions differ on turf and all-weather according to how sapping the surface is. Essentially, quickening becomes increasingly difficult the more demanding a surface becomes and easier (in relative terms) the less demanding it becomes. A horse gets closer to its "plodding rate" at a track like Southwell (the most testing of the all-weather tracks) than at other all-weather tracks, and it gets closer still to that plodding rate the more testing the surface at Southwell specifically is. It seems to me that the lower your average speed following initial acceleration, the more significant that initial acceleration will be. In other words, you will go faster - in percentage terms, though not in actual terms - early on the more testing the surface. This is borne out by the evidence and is one reason why I am happiest sticking with surfaces which vary the least when dealing with sectionals: all-weather rather than turf and polytrack more than fibresand. Over and above that, I do think that kickback is a significant factor in all-weather racing and probably even more so in dirt racing. Horses that avoid kickback are favoured, that is those that get out in front, or sometimes those that are detached in rear. It can be an advantage WITHIN A RACE to go too fast, if by so doing you disadvantage your rivals even more than you are disadvantaged yourself. But it should always show up AGAINST THE CLOCK that you have indeed gone too fast.

   2) My first question is: is this (front-runners struggling at Lingfield) true? Unfortunately, I cannot give a definitive answer here and now, though I do hope to in due course. I did a sample of 5f races at Lingfield in 2005 and 2006 and found that 25.9% of winners were leading 2f out (only 3.7% were last at this stage), that winners were ahead of 70.1% of their rivals at this stage on average, and that winners were only about 1.5 lengths behind the leader on average at this juncture. I hope to do something more comprehensive for all distances at Lingfield and at other all-weather tracks (using TurfTrax data) but it will take me time. Intuitively, I suspect that even if the "front-runners do badly at Lingfield" appears true it may well be largely a function of the competitiveness of the races. In an "average" 10-runner race at Southwell there may only be 5 with a realistic chance of winning 2f out, maybe 6 at Great Leighs, 7 at Wolverhampton, 8 at Kempton and 9 at Lingfield. If this were true then a horse leading 2f out at Lingfield would have twice as many serious rivals to hold off than at Southwell and so on. I believe this (if it is true) is down to a combination of surface and track conformation. The polytrack surface and the slightly downhill run (at which stage not many horses are asked to make race-winning moves) to 2f out at Lingfield tend to mean that more horses are in contention late on than at Southwell, with its more testing surface, longer home straight and sweeping bend. Great Leighs is a slightly more testing polytrack surface than Lingfield with a sweeping home bend, meaning that the wheat tends to be sorted from the chaff before the straight. The same remarks apply about the bend at Wolverhampton. The outer course at Kempton (the one over which most races are run) has a longer straight, and therefore the positions 2f out are more likely to reflect what the horses are capable of having been asked for a race-winning move rather than be a snapshot of the moment immediately before things develop. As mentioned before, I will try to put figures on these things.

   3) The times at Tipperary on August 28th were more or less correct: they were validated by the Timeform Irish team (to which I am a freelance contributor) at the time. It is possible that the 5f races were shortened slightly on account of track conditions, but there is no way of knowing for sure. My standard times are quite different from the Racing Post's and are arrived at in a different manner. You should never assume a normal distribution (and use simple averages) where times are concerned, most of all if you are using anything approaching small samples. My figures suggested that the surface was somewhere between good to soft and soft.

   4) I haven't watched the whole of this race, but the closing sectional (from 3 out) is actually quite quick compared to others at the course since 1998, which is where my (less than comprehensive) records go back to. You would usually expect a horse to run from there in about 18.6% of its overall time if it had gone a true pace. I make Yes Sir's individual sectional almost exactly that. It could just be that he ran the race efficiently, for all that it appeared otherwise. Timeform has Yes Sir running to the same level at Wetherby as at Cheltenham, incidentally.

   5) None today! I largely concentrate on Irish racing at present, and look at all the cards there extensively, with the fact that I am reporting and handicapping on the racing in any case feeding into my knowledge of what is going on. When I was primarily punting for a living a few years back I would usually concentrate on two meetings and evaluate them/price them up in about three hours then bet on them, but again I was keeping my eye in with two-year-old handicapping for about 20 hours a week, so I was not coming to all of the races "cold" by any means.

   Hope this helps.

   Simon

7. Zee Zoo | 21 October 2008

   A response above and beyond the call of duty. The first two answers required a couple of 'reads' for them to sink in, but I've grasped them now. Superb stuff Mr. Rowlands, and a thousand thanks are in order. Understanding WHY things happen in a race is something that can never be understated imo.

   Incidentally, the Timeform coverage of Irish racing is invaluable. Which leads me onto another question!

   Does just ONE guy take paddock notes (if in attendance), write the Perspective comments AND handicap ALL the races at each meeting? It must be hugely time consuming doing all that, especially for Curragh days with 30 runner Maidens and handicaps all the way?

8. Simon Rowlands | 22 October 2008

   I must admit, I thought the answers were less than adequate! I will try to back them up with something more substantial at some stage.

   The tasks on the Irish team are split up. Our guy at the course (Kevin Blake) will usually do paddock and report for up to 75 runners, with the remainder (especially large-field handicaps) reported on from the office/home.

   Reports from away from the racecourse often take in more runners by necessity (average meeting size in Ireland is 104). It should be easily possible to do 150 comments in a day, though comment writers usually have other responsibilities, such as compiling breeding details on unraced horses (thanks go to Stuart Jones in particular at this stage) and providing tissues. Handicapping is done by me and Martin Rigg, and we both get stuck into reporting and comment writing as required.

   It will often be the case that someone may be asked to do the tissue on a meeting beforehand, report on it at the time and write comments on it after the event. This process means that valuable knowledge is fully utilised rather than lost in the handover between individuals. It was a concept that I was very keen on adopting at The Sportsman, fwiw.

9. Zee Zoo | 22 October 2008

   Interesting stuff. The output certainly reflects well on its authors, and in turn puts the Racing Post's half baked efforts to shame (Justin O'Hanlon and Alan Sweetman excepted). Thanks again, and keep up the excellent work!