Winning systems: Running styles in All-Weather racing

In his latest column for, Dr Peter May takes a closer look at horse running styles on the all-weather, examining the predictably of success and if it can prove profitable.

  • Thursday 31 January
  • Blog
100% Bonus up to £100

In recent years the idea of using pace in a forecasting model has gained a greater level traction and rightly so.  On this web site you can find a significant amount of data relating to pace and taking this a step further there are now analyses of stride patterns to give the more dedicated punter a greater edge over his/her rivals.  This increased awareness of pace has probably resulted from the steadily increasing number of all-weather races in Great Britain.  Pace does seem to be more important in these events than races on turf hence the need to consider it more closely.

Whilst I have taken pace into account in many of the forecasting models, I have also found that running-style, a simpler but related component, can also significantly improve the accuracy of a model. 

Representing running-style in a model can be as complex or a simple as the race analyst wishes.  The more complex methods determine the running-style of the horse based on several races, but I have found that an approach based purely on the last race can also have a positive impact.  Consequently, my running-style component is determined from an analysis of a runner’s previous run only with one of five possible values assigned.  It is important to note that I am not trying to predict the pace of a race or how it will be run, my approach is to simply identify the way a horse might run using its last race as an indicator.

To denote the five possible running styles, I use the following characters: <<, <, -, > and >>.  These are assigned as follows: If the in-running comment includes the words: outpaced, always in rear, always behind, or soon behind then the running style is categorised as “<<”.  For horses that were held up off the pace then the “<” symbol is used.  Horses that raced close up to the pace, tracked the leaders, were prominent or led, are given a “>”, and those that made all, virtually all, or made most are given a “>>”.  All other runners are given the default indicator, a “-“. 

Naturally, this could be improved upon, especially with the information now available on this site, but it was useful for the models I developed in the past.

In order to test whether it remains an important component for all weather forecasting models, I ran some analyses for the years 2015 to 2018.  This running-style variable does not take any assessment of value into account, so the best metric for measuring its importance is the win rate.

At the very basic level we get the following results:

 Running-Style On Last Start
Win Percentage in AW Races<< ->> All
Non Handicaps5%11%9%15%27%13%

So, in non-handicap all-weather races, horses which were off the pace last time out had a win rate for their next race of around 5%, whilst those which made all, or attempted to make all, won at a rate of 27%.  In handicap races the figures range from 6% to 16%. Clearly, based on this information, the running-style variable will have an impact on any model designed to forecast the likely winner of a race.

Many of the runners included in the previous analysis will have had very little chance of success, and a proportion would have been simply in the race to gain a handicap mark, or get their current mark lowered.  So, it was no surprise that those which were held up well off the pace last time out also won at such a low rate on their next start.  By restricting the analysis to runners which started at 10/1 or less we remove most of the “non-triers” etc and get the following results:

 Running-Style On Last Start
Win Percentage in AW Races<< ->> All
Non Hcaps SP <= 10/114%22%22%24%33%24%
Hcaps SP<=10/114%15%14%17%18%16%

A horse which was well off the pace last time, and is now starting at 10/1 or less, has a win rate of one in seven (14%) up from the previous 5%. The full range of win rates now stretches to 33%.  In handicaps the lowest win rate is also 14%, but the upper figure only increased by two percentage points to 18%.

Returning to the full data set (horses of all prices) and analysing by race distance we get the following:

 Running-Style On Last Start
Win Percentage AW Non-Handicap Races<< ->> All
5-6 furlongs4%11%11%15%33%14%
7-9 furlongs6%10%9%15%22%13%
10-12 furlongs2%12%10%16%29%14%
13+ furlongs0%11%10%18%20%15%
Win Percentage AW Handicap Races<< ->> All
5-6 furlongs8%10%9%12%18%11%
7-9 furlongs5%9%8%11%15%10%
10-12 furlongs4%10%8%12%11%11%
13+ furlongs5%10%8%13%15%12%

It’s not too surprising that the highest win rate is for runners in non-handicaps over 5-6 furlongs that tried to make all, or made all, last time out.  As far as handicapping is concerned these horses are already exposed, having raced well previously, so there’s no need to keep their true ability hidden in order to gain a favourable handicap mark, consequently they win at a higher rate.  I would have expected a larger divergence between the figures across the race distance for handicaps, but the implication of these results is that a strong running style is almost as important over long-distance races as it is over sprints.

Field size is very much intertwined with running-style and pace, and an analysis of small fields in both handicap and non-handicaps shows a strong upward trend across the categories with far more winners in the “>” and “>>” groups than in the “<<” and “<” categories.

This pattern is also maintained when broken down by track. Lingfield has the weakest trend, but the other tracks all reflect the results presented earlier.

But, whilst running-style is clearly a useful attribute for forecasting the likely winner, does it help with profit?  For the high-level analyses, those without any conditions, the loss at Bookmaker’s starting price increases from -38p/£ in non-handicaps for horses with a previous run categorised as “<<” to -5p/£ for those given a “>>” last time out.  And for handicaps there’s a steady progression from -48p/£ to -22p/£.  This indicates that the price is not fully accounting for running-style assessed on the runner’s last race.

Consequently, I have little doubt that including a measure of running-style based on a runner’s last race will improve the forecasting capabilities of a model and may increase its profitability. However, I am equally confident that a more precise disaggregation of the in-running comments would be even more beneficial, or better still the introduction of a method that utilises the data presented on

Winning systems: Running styles in All-Weather racing
Sign up to bet365. Click to View Bonus Code Details
Up to £30 in free bets
Get £20 in free bets
£20 Risk Free First Bet
Up to £30 in free bets
Up to £30 in free bets
Up to £40 in Bonus funds
100% Bonus up to £100
£20 Free Exchange Bet
Get a £10 risk-free first bet

Existing User?

Forgot your password?

New User?

Sign up using our simple one-page form and you'll be able to access free video form, tips and exclusive content straight away.