In this article, I continue to look into run style and its impact on the outcome of horse races, writes Dave Renham. This piece focuses on the run style profile of individual horses and initially examines data from 2021, before comparing with results from the first part of the 2022 flat season, up to June 24th.
Before divulging my findings, for new readers I will briefly discuss what is meant by run style. Essentially, run style is the position a horse takes up very early on in the race. These are split into four categories as follows:
Led (4) – front runners; horses or horses that take an early lead; Prominent (3) – horses that track the pace close behind the leader(s); Mid Division (2) – horses that race mid pack; Held Up (1) – horses that race at, or near the back of the field early.
The number in brackets is the run style score that is assigned to each section. These numbers can be a useful tool for number crunchers like myself and they will be used at certain points in this article.
If we look at any Geegeez racecard and click on the pace ‘tab’ we get some past run style data for the race in question. Here is an example from April of this year – a 5f handicap at Windsor:
As can be seen, the run style figures from each horse's previous four races are shown (LR, 2LR, 3LR, 4LR). These figures are quite tight / close and hence it is difficult to be confident about predicting the order in which the field is likely to order itself early in the race.
The most important run style prediction is always which horse is most likely to front run and that is tricky here too. La Roca Del Fuego topped the list, just, on 14 points, so was marginally the most likely front runner, and as it turned out did lead from start to finish.
However, pre-race, one could not have been confident that La Roca Del Fuego was going to lead. In an ideal world when trying to predict the front runner, we would prefer a horse to be well ahead numerically of the rest of its field. For example, Horse A has 16 points (the maximum possible for a four-race sample), and Horses B, C, D, etc all have scores in single figures. Even then we cannot guarantee Horse A will lead but all things being considered, the chances are very likely he/she will.
Some less regular readers at this point may be asking themselves why trying to predict the front runner is a useful thing to try to do. The answer is simple: front runners are the best value at most distances on the flat; and many distances over the sticks, too. For example, in 5f handicaps in the UK from 1st Jan 2018 to 31st Dec 2020, if you had predicted who would front run pre-race and place a £1 bet on every single horse you would have won nearly 20% of all your bets for an impressive profit of £619.46 (ROI +32.7%).
Now to the article proper as it were:
My focus today is on UK handicaps of 5 furlongs to 1 mile; I am using these races as there is a strong front running bias in general at shorter distances. The bias is strongest over 5f (see example above), but it is still potent up to a mile on most courses. My initial dataset looked at all such races in 2021.
To start with I focused on all horses that had raced at least 4 times in 5f - 1 mile handicaps in 2021. From there I wanted to check a few different things.
Horse run style averages (UK turf flat handicaps, 5f-1m, 2021)
First stop was producing run style averages for each horse: this was performed in exactly the same way that I have created trainer, jockey and course run style averages in the past. I simply added up the Geegeez pace / run style points for a particular horse over the 2021 season and divided it by the number of races. The higher the average the more prominent the horse tends to race. The averages ranged from 4.00 (horses that led in every race they contested in 2021) to 1.00 (horses that were held up in every race they contested in 2021). Just 12 horses had run style averages of 4.00, which will come as no surprise as I was looking at ALL their runs in these handicaps over the year.
There was a horse that raced 37 times in 2021 – yes, 37! The horse in question was Qaaraat. Qaaraat had a run style average for the year of 3.11 thanks to leading 11 times, racing prominently 21 times, mid-division three times, and being held up just twice.
Here is a selection of horses with their run style averages for 2021. I have chosen those with some of the highest run style averages, and those with some of the lowest – the number of races they contested in also shown:
|Horse||2021 races||2021 run style average||Horse||2021 races||2021 run style average|
|How Bizarre||5||4.00||Diffident Spirit||4||1.25|
|Master Matt||4||4.00||Hope Springs||4||1.25|
|Pinnata||6||4.00||James Park Woods||4||1.25|
|Rains Of Castamere||7||3.86||Otto Oyl||4||1.25|
|La Roca Del Fuego||6||3.83||Rectory Road||12||1.25|
|Eye Of The Water||5||3.80||Eyes||13||1.23|
|King Of Stars||10||3.80||Treacherous||13||1.23|
|Twilight Madness||4||3.75||Amazing Amaya||5||1.20|
|Just Glamorous||7||3.71||Our Little Pony||5||1.20|
|Zulu Girl||7||3.71||Power On||10||1.20|
|Al Simmo||6||3.67||Air To Air||6||1.17|
|Autumn Flight||12||3.67||Fantasy Believer||6||1.17|
|Boogie Time||9||3.67||La Rav||6||1.17|
|Global Esteem||11||3.64||True Mason||12||1.17|
|Gometra Ginty||11||3.64||Duke Of Firenze||19||1.16|
|Antagonize||8||3.63||The Cola Kid||13||1.15|
|Charming Kid||8||3.63||Papas Girl||7||1.14|
|Just Frank||8||3.63||Surprise Picture||7||1.14|
|Air Raid||5||3.60||Alba Del Sole||8||1.13|
|Alba De Tormes||5||3.60||Clashaniska||8||1.13|
|Animal Instinct||5||3.60||Desert Land||16||1.13|
|Wings Of A Dove||5||3.60||Libby Ami||11||1.09|
|Thegreyvtrain||24||3.58||De Vegas Kid||12||1.08|
|Gobi Sunset||7||3.57||Golden Apollo||12||1.08|
|Healing Power||7||3.57||Van Dijk||14||1.07|
|Mountain Brave||9||3.56||Catch My Breath||14||1.00|
|Goddess Of Fire||13||3.54||Divine Messenger||6||1.00|
|Bert Kibbler||6||3.50||Fastnet Crown||6||1.00|
|Della Mare||4||3.50||Mayson Mount||5||1.00|
|Louie de Palma||6||3.50||Raatea||7||1.00|
|Modular Magic||6||3.50||Sin E Shekells||5||1.00|
|Rhubarb Bikini||6||3.50||Stone Of Destiny||6||1.00|
To be honest, I wasn’t sure how relevant looking at run style averages from a longer period of time (rather than the four most recent races) would be; but I use longer term data for trainers and jockeys so felt there was some logic to justify analysing it.
Now I had the run style averages for 2021 for each horse, I grouped them as follows:
|1.49 or below|
|1.50 to 1.99|
|2.00 to 2.29|
|2.30 to 2.59|
|2.60 to 2.99|
|3.00 to 3.49|
|3.50 to 4.00|
From there I looked at the performance of each of the groups in terms of 2021 results. Here is what I found – I looked at strike rates first:
As the graph neatly shows, horses with higher run style averages based on the 2021 season were more successful in terms of strike rate. Horses that had an average of at least 3.5 for 2021 scored nearly 20% of the time. If we now do a comparison of return on investment (ROI%) we can see a clear correlation:
I used a line graph here as it is slightly easier to see than if using a bar chart. There was a huge return on investment for horses with an average of 3.5 or more – more than 40p in the £.
Horse Led Percentages (UK turf flat handicaps, 5f-1m, 2021)
I did the same type of analysis but using 'led percentages' rather than run style averages. In order words, I worked out in what percentage of races each horse led early during 2021. For instance, if a horse ran ten times and it led early in four of these, its figure would be 40%. As with run style averages, I grouped the led percentages to ensure acceptably sized datasets:
The chart shows a very similar pattern to what we saw with run style averages: this time, horses that led the most in percentage terms were the most successful.
Here are the figures in terms of return on investment:
Again, there is excellent correlation with both graphs; in fact all four graphs correlate strongly. Horses that led in 50% or more of their races in 2021 were extremely profitable – a return of £1.28 for every £1 bet. It should also be noted that these returns are based on starting prices, so with early prices, BOG or Betfair SP one would expect to improve markedly on this baseline figure.
Let's stop using history to predict the past...
Now statisticians will tell you, quite rightly, that using past data from one particular year in this way is going to produce slightly skewed results. This is because we are looking retrospectively at horse performances; we know horses that lead early win more and so looking at horses that led the most often in 2021 should produce the kind of positive results we have seen.
However, there are two points I’d like to make. Firstly, these data prove the point once more about how important early speed is, and secondly it shows that creating horse run style averages seems to be a worthwhile project. Indeed, the run style averages actually outperformed the led percentages, at least at the business end of their respective spectrums (the highest run style averages versus the highest led %’s).
At this point in my research I decided to use the 2021 run style averages I had created and apply them to races in 2022 – up to June 24th. Of course, these run style averages are based on the previous year with no new runs in 2022 taken into account. However, I was hoping to demonstrate that the higher run style averages would still outperform the lower ones. This is what I found.
The strike rates are much more even as you might expect, but still there is a positive edge when we get to a run style average of 3 or more. Conversely, the two lowest strike rates also occur for the two lowest run style groups. The best part, naturally, is seeing the profit/loss figures – profits for those averaging 3 to 3.49 and 3.5 to 4; and the commensurate losses for horses averaging 2.59 or lower are quite steep when viewed as a group.
As we have done to this point, let us again overlay the 2021 led percentages on the 2022 results hoping for a similarly upbeat picture:
It is gratifying to see similar results here. Specifically, horses that led 20% or more in 2021 have outperformed lower 'led percentage' groups both in strike rate terms and in returns on investment. Meanwhile, a 2021 'led percentage' of 33.3% to 49% produced a small profit from 2022 runs, with considerably bigger profits generated by the 50% or more group.
The main takeaway from this research into 5f to 1mile handicaps is that horses which led more often (in percentage terms) over a recent period of time are more likely to be profitable to follow than horses which have led less frequently. The same can be said for horses with higher run style averages.
The million dollar question, however, is how many races should we use? The four currently published in the 'pace' tab on the Geegeez racecard is a great starting point. We know from earlier research that horses which led at least once very recently are more likely to lead early than horses that have not. Likewise a last-four-race run style average is useful too (also shown in the 'pace' tab under the column ‘Ave’). The higher the average, again, the more likely it is that a horse will lead.
In answer to the question, my best guess is that anything between four and a dozen races would be optimal. In this piece, for example, some horses had run style averages based on their last four runs, and some had an average based on a lot more than four runs. One could argue this is not perfect and I'd have some sympathy with that argument; but, for me, the time it takes for data collection is important.
Using this more flexible approach (a minimum of four runs) meant it took me less time to create all the data I needed to start writing the article. I shared nearly 150 individual horse run style averages earlier; in total I had to calculate nearly 6000. If I had tried to create ‘last eight runs averages’ for example for all horses I probably would still be trying to do that at Christmas, and probably Christmas 2025! Research is just that, research. It will never be perfect, but for me it is a fun way to learn more about racing and to help me share ideas with the wider geegeez.co.uk audience.
Thanks, as always, for reading.