RaceMetrics has added four new filters to System Builder: Draw IV, Pace IV, a tactical score combining them, and a draw-by-pace cell testing both together. All four figures have appeared on racecards for some time. Until now, you could only observe them. You can now back-test them.
What the data shows
Analysis of 626,913 flat runners since 2018 reveals a clear pattern. Strike rates climb from 9.3% with a bad draw to 11.1% with a favoured draw. The draw demonstrably affects which horses win.
However, the A/E column—which measures performance against market odds—stays flat across all draw bands, ranging from 0.99 to 1.01. The market has already priced in the draw's impact. Knowing the draw matters is not the same as knowing it is worth backing.
Why a general draw bias does not exist
One thing worth adding, because it explains why these filters are built the way they are. The figures above band runners by draw IV — how favoured a stall is at that particular course and distance. Band them by the raw stall number instead and the effect vanishes.
Across every British and Irish sprint of the last two years — flat, six furlongs or shorter, eight runners or more, 38,289 runs — horses drawn in the low third won 10.1%, the middle third 9.0% and the high third 9.1%. About one percentage point separates best from worst. Aggregated across the sport, the draw is worth nothing at all.
That is not a contradiction of the numbers above; it is the reason for them. Draw bias is a property of a particular course at a particular distance, not of racing. At one track the low numbers are favoured because the bend comes quickly; at another the stalls are on the far side and the high numbers have it. Both effects are real and worth money. Add them together and they cancel, which is exactly what the flat line above shows.
So a claim about the draw only means something if it names a course and a trip. That is why the draw figures are held per course and distance, rebuilt whenever a course runs a distance it has not used before, and why "low draws are favoured" — repeated everywhere — is noise rather than an angle.
How to use it
These filters are building blocks for systems, not systems themselves. Add draw or pace filters to an existing system and test whether they improve your numbers. Alternatively, use them to exclude runners you would otherwise back.
The pace figures draw from a horse's last ten runs. Roughly 9% of flat runners have no run-style history, and 23% have only one to three runs. Set a minimum run sample before trusting a pace figure, or you risk reading noise.
All four figures are also available as they stood last time out, allowing you to identify horses that finished well despite an unfavourable draw.
