Draw Bias, Course by Course

Every flat race starts with a lottery. Stall positions are drawn at random, and at some courses that random number matters more than a fitness doubt or a first-time hood. At Chester, a sprinter in stall one begins with a measurable head start. At other courses, the "draw bias" you hear about in the paddock stopped being true twenty years ago, if it ever was.

This is worth getting right, because draw bias is one of the most repeated and least checked ideas in racing. Commentators inherit it from other commentators. Old course guides get quoted long after the course has moved its rails, changed its watering, or — as happened across Britain in 2011 — renumbered its stalls entirely. The received wisdom is sometimes accurate, sometimes stale, sometimes flatly backwards.

So: why a stall number can decide a race at all, which biases the published data actually supports course by course, and how to check a claimed bias yourself — the only knowledge that survives a rail move.

Why the stall matters at all

A draw bias needs a physical cause. Five are worth knowing.

Distance to the first bend. On a turning track, the inside horse runs the shortest trip. If the first bend arrives quickly, wide-drawn horses must burn energy to cross over or surrender ground on the turn. The tighter the track, the bigger the effect; Chester, a circuit of barely a mile turning almost continuously, is the extreme case.

Camber and drainage. Tracks are not flat across their width; water drains off a camber, so one side of a straight course can ride consistently quicker. At The Races' Curragh guide notes a gradual camber from the stands' side down to the far side on the sprint track, leaving the stands' side that bit drier.

Watering and ground variation. Watering is never perfectly even — wind sees to that — so strips of faster and slower ground develop across the width of a course. This is the least stable cause of bias: present at one meeting, gone the next. Older watering systems created artificial biases; modern kit has removed some of them.

Field size. A bias trivial in an eight-runner race can be decisive in a 20-runner one, because a big field stretches across the whole track and leaves the worst-drawn no room to fix their position. At The Races' Royal Ascot analysis puts it plainly: in a ten-horse race, stall 8 is no great problem; in a 24-runner straight-course handicap, it is a serious one.

Pace. The draw never acts alone. A good stall is worth most to a horse with the early speed to use it — more below, and in our piece on pace maps.

The courses where the evidence is strong

Chester is the canonical case. The Roodee's tight left-handed turns mean low numbers hug the inside rail from the gate. Published figures from FlatStats and others have around 63% of 5f winners coming from stalls 1–3, and over 7f one five-season sample had the lowest three stalls supplying 33 of 70 winners — 47% of the wins from 28% of the runners. A genuine, structural bias — and the best-known in Britain, which matters when we get to prices.

Beverley over 5f had, thirty years ago, a reputation as one of the most biased trips in the country. The modern picture, from Geegeez's five-season study of 99 races (2021–2025), is softer but real: low draws won 36% against 31% for high, the two lowest stalls took 28 of the 99, and the very highest stalls were close to hopeless. An edge for low, a genuine penalty for very high — not the monster of legend.

Musselburgh over 5f is arguably the strongest current turf bias in Britain, and it runs the opposite way. Across 102 races in the same Geegeez study, high draws won 51% against 18% for low — nearly three times as often — with the percentage-of-rivals-beaten figures corroborating it rather than it resting on a few big-priced winners.

Wolverhampton, the all-weather case, shows a structural low-draw edge in sprints: the tight left-handed Tapeta oval throws 5f and 6f fields into a bend within seconds of the start. Geegeez's all-weather review has the low third winning 38.3% of 5f races with a PRB of 0.53 against 0.47 for high — modest per race, persistent over hundreds. York's 5f belongs here too: the low, far-side third won half the turf sample, with the highest stalls in big fields almost never involved.

Straight tracks: Ascot, Epsom and the Curragh

Straight-course biases are different in kind. With no bend to save ground on, the question is which strip of turf is riding fastest — and that changes with the weather.

The Ascot straight track is the clearest conditional bias. At The Races' Royal Ascot analysis describes a "golden highway" up the stands' rail — favouring high numbers — on decent or quick ground, but notes that on soft ground moisture gathers towards the stands' side and the far side becomes the place to be. Ascot has less a draw bias than a ground-reading problem, refreshed daily.

Epsom's 5f drops downhill on a pronounced camber, and the traditional claim is that high numbers on the stands' side dominate the big-field Dash on Derby day. Honestly, this is disputed: GG.co.uk's myth-busting review found the recent roll of winners far more mixed across the draw than folklore suggests. Treat the Epsom high-draw rule as a tendency in big fields, not a law.

The Curragh sprint course is wide, and At The Races' draw guide reports no consistent bias over 5f and 6f — but when one appears, usually towards the stands' side and high numbers, it can be extreme. In one 6f handicap the guide cites, the winner was drawn 30, the next ten home all 16 or higher, the best single-figure stall a remote 15th. That is straight-track bias: absent for months, then race-deciding.

The table: documented effects at a glance

Only entries supported by published figures or named analysis are included. "Strength" characterises the documented modern effect, not the next race.

Where the bias is folklore, or was

Every course on that table has a neighbour whose famous bias no longer survives the data.

Sandown's 5f, run on its own spur, was in the mid-1990s among the most biased trips in Britain — high draws dominated then. Rail movements changed that, and the modern, milder pattern actually favours low numbers near the far rail, most clearly on soft ground; a 1996 memory would have you at the wrong end of the stalls. Thirsk's celebrated high-draw sprint bias of the 1980s and 90s survives only as a modest edge. Even Chester has had seasons where a false rail blunted the low-draw advantage enough that middle draws did just as well.

There is a subtler trap too. Before 30 March 2011, British courses numbered their stalls inconsistently — stall 1 against the inside rail at some tracks, the high number at others. The British Horseracing Authority standardised numbering from that date so that stall 1 is always nearest the inside rail — so pre-2011 draw statistics may describe the opposite end of the gate. A surprising amount of pub wisdom is a true fact about a numbering system that no longer exists.

Draw and pace compound

A draw bias is rarely the stall alone. It is the stall plus what the horse does in the first hundred yards.

At Chester or Wolverhampton, the prize for a low draw is the rail position into the first bend — but only a horse quick enough from the gate collects it. A low-drawn hold-up horse can be shuffled back on the rail with a wall in front of it and finish worse than a wide-drawn rival with a clear run. On a straight track where the bias is fast ground on one side, early speed matters less than holding a position on the favoured strip.

So read the draw through the pace map, not instead of it. And remember the field-size multiplier: the stall that is a footnote in a nine-runner race can be close to disqualifying in a 22-runner handicap, where there is no room to correct a bad start.

The market already knows

Here is the uncomfortable part: everything above is public. The Racing Post prints draw analysis daily; Timeform and At The Races publish course-by-course figures. Smarkets' guide to draw bias notes that well-known biases are absorbed into the odds, and analysts have long observed that low numbers at Chester are routinely overbet for exactly that reason.

A genuine draw bias tells you who is likely to be helped by the track. It does not tell you who is a good bet, because a known bias is usually already in the price — sometimes with interest.

The practical consequence: a horse whose form reads like a 7/2 chance can be sent off at 5/2 purely because it has stall one on the Roodee, and at that price the stall has been paid for twice. Knowing the bias exists is table stakes; the interesting questions are whether the market has priced it correctly today, and whether today's rail position, watering and ground mean it even applies. It is the same discipline as reading going changes against market moves, or asking whether a well-backed horse is favourite for a reason. None of it removes risk, and it needs saying plainly: most punters lose over time, bias or no bias, and no amount of draw study turns betting into a way to make money.

How to check a bias yourself

The lasting skill is not memorising the table — it is being able to rebuild it, because it goes stale. A working checklist:

Get a real sample. Ten races prove nothing; one big-field handicap can swing the winner count. Look for 50-plus races at a course and distance, and prefer percentage of rivals beaten (PRB) to raw winner counts, because PRB uses every runner's finishing position, not just the winner's.

Compare thirds, not stalls. Splitting the field into low, middle and high thirds normalises across field sizes; an unbiased trip sits near 33/33/33 in wins and near 0.50 in PRB. Musselburgh's 51/31/18 is a bias; 36/33/31 at Beverley is a lean.

Weight recency, and mind the 2011 line. Rail moves, drainage work and watering regimes change tracks within a few seasons, and anything before March 2011 may carry inverted stall numbering. Five recent seasons beat twenty old ones.

Check today's rail position. Courses publish rail movements in the going report, and a false rail can shift the effective bias overnight — it is precisely how Sandown, and at times Chester, blunted theirs. A bias measured against one rail configuration does not transfer to another.

Watch the early races on the card. For straight-track and watering biases, the day's first results are worth more than any archive. If every winner is coming up one rail, you have learned something no historical table could tell you — though so has everyone else watching.

Then look at the price. A bias only matters to a bettor if the market has missed it, and with the well-known ones it almost never has. Most of the time the honest conclusion from your own draw study is quiet confirmation that the market has it about right — itself worth knowing, and the difference between understanding how races are run and repeating what the man at the bar heard from another man at the bar.

The stall decides more races at Chester and Musselburgh than form students like to admit, and fewer at Epsom than folklore claims. Hold both thoughts at once and you are reading the draw properly.