Forecasting the results of horse races is an intriguing problem involving a large number of variables with complex relationships. As a result many solutions exist which vary widely in complexity and accuracy. However, in broad terms, these approaches can be sub-divided into two main types: form analysis and systems.
Most race enthusiasts start by adopting the form analysis approach which requires a high level of engagement with the sport and a knowledge of form and its importance. Then, after a period of time, many adopt an approach either based purely on systems, or incorporate systems to a greater or lesser degree within the form analysis approach.
One advantage of the purely systematic method is that, once developed and tested, a system simply requires the user to press the start button on his/her computer and the selection process is completed automatically. Consequently there is far less involvement in the sport in general, and certainly no requirement to follow it closely, which would be viewed as a benefit by some punters but removes much of the interest for keen followers of racing.
Like the majority of racing fans my interest in the sport was developed through form analysis in an attempt to solve this forecasting problem in a structured and logical fashion. And when I used to provide cover for the ATR Speed Page I would return to this approach which, for betting purposes, I have not used for many years preferring the more profitable systematic methods.
The reason for this change of approach was simply because it is difficult to provide a coherent argument for a decision which is based purely on a system. Had I written “Horse X is the selection in this all age handicap because it ran within four days” then I suspect I would not have been invited to provide Speed Six write-ups for very long.
Over the years, whilst I have modified my other methods significantly, one element of my form analysis approach has remained consistent: to try to find improving horses. When used alongside ratings the justification for this is obvious: if there’s a top-rated runner which is likely to improve then its chance of success is greatly improved.
But nowadays I place even greater importance of improvers even in the ever-present handicap races. Consequently, in a handicap I would prefer a potential improver that is rated maybe as much as 7lbs inferior to an exposed top-rated runner that is very unlikely to better his/her current rating.
Identifying improvers is not an exact science. For some horses it is obvious, but for others it is not as clear. One key factor is the number of times the horse has raced which is most applicable in juvenile contests.
A two-year-old on his/her first run has an expected win rate of about 7%, without taking into account other factors such as pedigree or trainer. However, juveniles on their second start win at a rate of about 13% in non-handicaps, and 20% in nursery races, and in terms of ratings the average amount of improvement from first to second run is around 6lbs.
Other potential sources of improvement are a change of stable, change of ground, change of race type (non-handicap to handicap) specifically the runner’s first handicap run, and a change of race distance, to name but a few. The latter is often overlooked and therefore is examined in more detail in the following sections.
Before analysing the effect of the attribute, it is necessary to stipulate what a change in distance actually is. For the purposes of this analysis I have decided on the following: Horses which are running at least one furlong more than their previous maximum race distance. Naturally other definitions can be used, such as using percentage change in distance, or relaxing the maximum distance requirement, but that is the rule I have set for this article.
One further note is that in the following sections all profit/loss figures are given to off time exchange prices before commission, and the analysis is restricted to turf flat races.
Handicaps would seem to provide the best race grade for testing this approach since there are so many in British racing nowadays, and trainers have been known to use all the tricks available to them in order to eek out a couple of wins in this competitive race classification.
During the period of analysis there were 22,000 qualifiers under the distance change rule given earlier and these made a level stake loss of 6p/£. The poorest performances, in terms of profit and loss data, were by 2yo and 3yo runners which is possibly due to fact that a change in trip is more likely to be flagged up with these runners which are then underpriced by the more cautious layers.
Removing these from the sample leaves approximately 4,700 performances and a level stake profit of 4p/£. If trainers were trying to use a change of race distance to get ahead of the Handicapper then it is likely that the horse would have been running in handicaps previously.
Of these 4700 performances, 3921 were attributed to horses which had raced in a handicap most recently, and these made a profit of 11p/£ which is reasonable. In fact, previous race grade is a better discriminator than last race finishing position, with horses running after a success returning a significant loss.
Returning to the sample of 4700, there’s an interesting distinction when broken down by the new race distance. Those stepping up to a distance of 13 furlongs or more, i.e. they had previously only raced at a trip of a maximum of 1m4f, the profit jumps to 16p/£ from just over 2,500 runs with the handicap to handicap runners improving this to 22p/£ from over 2,100 performances.
Out of the initial sample of 22,000 runners, the Mark Johnston stable featured 512 times, over 100 more than any other trainer. Amazingly these 512 runners made a profit of 26p/£ with a win rate of 17%. Richard Hannon, Mick Channon and William Haggas also had a profitable profile from a decent number of bets and are worth noting.
As expected the non-handicaps are not such a profitable race grade; the 14,700 runners upped in distance made a loss of 14p/£ during the period of analysis. Focussing on those at the front of the market, horses with a starting price of less than 10/1, reduces the loss to 4p, and horses racing after competing in a handicap race made a profit of 15p/£ from 446 runs which is the stand-out figure for non-handicaps and is possibly worth monitoring next season.
Richard Hannon had 600 runners in the initial sample of 14,700 and these made an exceptional profit of 16p/£; Richard Fahey’s 300 runners made 13p/£ and William Haggas horses returned a profit of 45p/£ from 250 runs.
When conducting the traditional form analysis, improvers are certainly horses to note carefully, whether they simply have an improving ratings profile, are unexposed, or are likely to do better due to the racing conditions, and this analysis has shown that those stepping up in trip should be analysed with care, or even, for some trainers, used as the basis for a betting system.