What use is "trainer form" when analysing a race?

Betting Strategy / Jack Houghton / 28 November 2008 / 2 Comments

Jack Houghton wrestles with the fundamentals of "trainer form" and wonders if it can really help punters find value.

This isn't an attack article. So don't write in. Well, do write in; just don't write in saying this is an attack article; because it isn't. It's a genuine search for assistance and enlightenment. I'm asking for help.

My question is: what use is "trainer form" when analysing a race? It's something I've pondered for years; and something I've yet to reach a satisfactory conclusion on. The gut says it has value; the head is struggling to work out what that value is, or where it lies.

What's clear is that the majority of racing journalists, pundits and commentators throw "trainer form" about with gay abandon; without a concrete understanding of what they are talking about. When one of them tells us a trainer is "in form", it generally means they are aware of a recent winner from the same outfit, and not much else.

I'm guilty of it myself. I'll be on Timeform Radio. I'll analyse a race and recommend a horse to back. The presenter will say, in agreement: "Yes, and the yard's in form, isn't it?" And I will nod along in agreement. Although I'm starting to realise that nodding isn't such an effective radio broadcasting technique. And anyway, I haven't the first clue whether the yard is "in form" or not. Or, for that matter, whether it's any use to know whether a yard is "in form" in the first place.

What I'm certain of is that the most prevalent method of assessing "trainer form" - counting recent winners - is so crude as to be of no use whatsoever. What the Racing Post have demonstrated to some extent with their trainer percentage-to-form figure and what Simon Rowlands demonstrated to a fascinating extent back in May of this year, is that accurately assessing trainer form is extremely complex.

And, of all the methods I've seen, Rowlands' version - looking at the percentage of rivals beaten - stands a long way clear in terms of statistical rigour. But, even with this accurate assessment of trainer form, I'm still doubtful as to what predictive use it has. In other words, how can it help punters find value?

Let's say you're analysing a race in which Dermot Weld has a runner. You know, using Rowlands' figures, Weld is flying high: boasting a 69.7 per cent of rivals beaten figure. Does that make his 7-4 shot a 6-4 shot? Or is it something you can't really apply to individual horses? Do you need to consider the form of all trainers with representatives in a race and use this as an integral assumption in determining the tissue price of all the horses running? I don't know, but I'm guessing the latter is closer to the right answer.

Or perhaps it's not supposed to be used as a variable in assessing a horse's chance at all? Instead, should it be used as a last minute sense-check? If your selection's trainer is in the doldrums, keep your money in your pocket.

Perhaps, but I have more fundamental problems with "trainer form". First, as it's a reflection of historical events, it will necessarily lag behind reality. A trainer's "form" is a constantly shifting beast; and trying to apply an out of date measure to a future event brings masses of inaccuracy. And perhaps more importantly, this inaccuracy will vary across trainers, depending on how frequently they have runners. Doesn't this mean you'll be comparing measures which appear consistent, but which actually contain significantly varying margins of error? But then, some would argue, very similar things could be said about all ratings.

The second problem I wrestle with in regards to gauges of "trainer form" is that they attempt to apply an average measure to an individual horse.

What the hell do I mean by that? Not really sure, but I was thinking... In most yards, horses are trained in some kind of cycle. Crudely speaking, they come back in from a break, begin training and, depending on the make-up of the horse, might be ready for their first run ten or so weeks later. And, again, depending on the horse, but also depending on the training outfit, the horse might reach something like its peak form after another run or two. It is likely therefore that there will come a "wave of peak form" for a number of horses in a yard, providing enough of them are able to fulfil their training and racing timetable. It is my guess that this kind of phenomena is reflected in methods of assessing "trainer form".

So, on the surface, having some view of trainer form looks like it might be useful; it tells us when some of a trainer's stable is at its best. But what about the horse in a yard who hasn't managed to keep pace with the average progression? Won't the follower of "trainer form" overvalue its chance? Or what about the horse who requires a much shorter preparation to reach peak fitness? Won't its chance be undervalued? Perhaps these individual blips are more than accounted for by the positives gained from an overall knowledge of relative trainer form? I really don't know.

I was hoping this article would be cathartic; that by committing to paper a "trainer form" stream-of-consciousness I'd miraculously find enlightenment. But I'm at the end now and it feels like the start: my gut says one thing and my head says another. This explains the plethora of question marks above. What do you reckon?

Comments (2)

1. Simon Rowlands | 10 December 2008

   All fair comment.

   I tend to use trainer form in the context of what could be expected of that trainer and then to apply it, rather loosely, to the individual horses in question. For instance, in the example given, Jim Bolger's 69.6% was a seasonal high: he had started off the campaign on 53.3, at which time there seemed to be a lot more mileage in opposing his horses. By contrast, Aidan O'Brien only briefly dipped below 60% and was normally rock solid at 62-66. There was no stage at which his stable was markedly in or out of form, by his own standards.

   You could convert these figures into "fair odds", or as some sort of variable to feed into a fair-odds measurement, but I'm not sure that's necessary.

   Simon

2. Mark Chapman | 14 January 2009

   Trainer form however you analyse it, is flawed; but surely not as flawed as not taking it in to account at all. My first job when assessing a race is to look at trainer form.

   I click on every trainer's name, on the Racing Post website race card. This gives every runner he/she has had in the last two weeks. With the exception of that day's racing, which we know or can look up anyway. So the results are slightly out but not by far. Can also look up results during the day and alter the in form / out of form trainers.

   Being able to see recent results in detail, (not just a two week summary as in Todays Trainers); gives me the opportunity to find in form / out of form trainers sooner, when nearer the start of a run. And also change their rating sooner if things change. Sometimes a trainer can be in form one week and out of form the next. Or the Hannon's and Channon's etc. have so many runners their form is best judged on a weekly basis (sometimes days). Those with few runners are impossible to quantify, unless I believe they should have had more runners, so might have a virus.

   I rate each trainer by *** (only one or two trainers get this during the season), **, *, */, //, /, /-, -, -x and x (don't touch with a barge pole). Do 1-10 if you prefer. Looking up how many wins, seconds, thirds, fourths, etc. and how fancied they were. An 8/1 shot who is 3rd I think has probably run to form; where as an odds on shot that is 3rd probably has not. As with any true value assessment, we are always thinking in probabilities. We do not need to know if it definitely ran to form or definitely did not.

   With small trainers I might take thirds, fourths and fifths in to account more for being in form (depending on their prices). But not usually with a top trainer, unless it is a top class race.

   When there is a first time out runner some knowledge (from Timeform Statistical Review) of trainers records is beneficial. A fourth placed effort in a large field of Newmarket maidens by a 2 year old trained by Roger Charlton would be a good run.

   Say there is an out of form horse, yet the trainer has recently returned to form. I find these horses are much more likely to return to form than one with a trainer who continues out of form. This can pin point good value bets. Conversely, say the horse appears in good form yet the trainer has gone out of form since it's last run. His chance may not be as big as the form book suggests.

   Trainer form is only one part of my form study, finding out (using Timeform) what rating the horse is capable of, going, distance, temperament, course, pace etc. The two questions I ask are: What rating is the horse probably capable of given the conditions? And how likely is it to run to that form? Trainer form directly influences the latter question when working out my 100% book.

   The percentage I add to each horse for trainer form is not always the same. Have always found it easier to look at each horse as a whole character when allocating a percentage. Rather than give points for distance, going, trainer form etc.

   One who does not need much getting fit will have his chance identified and boosted by the horses record first time out or after a long break. So that stat may take preference over trainer form. There will be the odd occasion where an unfit horse gets wrongly boosted, but there are other factors that can identify this type. Horse and trainer's first time out strike rate. It is (imo) much better than not taking trainer form in to account. He is running a big risk of backing too many horses who might be good value bets on speed ratings or anything else. But stable form means the likelihood of running to form is lower, so it's true chance is less. So may not be a value after all.

   Sorry for the length of this post, you did ask how.

   Mark