Gate Stats & Ground Loss

Where the draw helps, hurts, or lies

Where a horse starts can be worth several lengths before it has done anything. This chapter is about reading the draw — and the ground a horse loses when the draw, or the trip, goes wrong.

Theory: the draw

The starting gate number — draw, barrier, post position — shapes early positioning, traffic, and the ground a horse travels. An inside draw saves ground but can trap a horse behind a wall of speed. An outside draw avoids early traffic but forces a wider trip or an early burst to cross over. Whether inside or outside wins out depends on track geometry, distance, surface, field size and running style — never on the draw number alone.

Gate statistics on MWP

For every track-distance, MWP computes how each gate has actually performed across thousands of starts. Green = structural advantage, red = disadvantage, grey = neutral. This is data, not opinion — but (Chapter 6) check the sample size: a gate with 40 starts is a hint, a gate with 400 is a fact.

Apply five filters before you trust a gate stat:

  1. Sample size — a 25% rate from one gate across 8 races is noise. Demand 30+ races at the exact distance and surface.
  2. Slice by distance — gate 1 at 5f and gate 1 at 1¼m are completely different propositions.
  3. Slice by surface — turf rail movements change the effective course; dirt doesn't move.
  4. Weight the recent meeting — track conditions evolve weekly. Last week's "good rail" can be this week's "dead rail" after rain.
  5. Field-size context — gate 1 in a 5-horse field has a 20% baseline by random chance alone.

A worked example. Sha Tin, 1200m: across more than a thousand starts, gate 1 wins 12.0% (a +3.7-point edge on its fair share) while gate 13 wins 4.8% (−2.5). Same track, same trip — the rail gate wins more than twice as often as the wide one. That is the whole reason a fancied favourite drawn 13 here (the AKASHVANI case below) is fighting the draw before the gates open, and why an overlooked horse drawn 2 is quietly handed an edge the market hasn't priced.

Distance to the turn — the geometry that decides how much the draw matters

The single most important geometric fact about any race is the distance from the gate to the first turn — the run-up. It decides whether the draw is decisive or barely matters.

A short run-up makes the draw enormous. When the field meets the bend almost immediately, a wide-drawn horse has no time to cross over and tuck in. It is forced to either burn energy getting across early or get shuffled back and round the turn buried wide — and it never recovers. Textbook short run-ups:

A long run-up neutralises the draw. Give the field several hundred metres of straight running before the bend and every horse has time to find its position on merit — the gate barely matters. Long run-ups:

Straight courses remove the turn entirely. On a zero-turn race the draw stops being about ground saved and becomes about which part of the track is running best and where the pace is. Straight racing is far more widespread than people assume: Newmarket's Rowley Mile is straight up to 1m2f, Niigata has Japan's only straight course (1000m turf), Sha Tin has a 1000m straight chute, and Flemington's "Straight Six" runs 1000–1200m without a bend. On these, "high numbers" or "low numbers" being favoured is a daily bias question, not a permanent geometric one.

Where to find the run-up. Western racing authorities rarely publish run-to-first-turn in metres — you usually infer it from the track map and the position of the start relative to the bend. Japan's haizara.jp and HKJC's sampling-point markers are the few systematic sources. The practical habit: before you bet a turning race, ask "how far to the first bend?" — a long answer means downgrade the draw factor; a short answer means it may be the most important factor in the race.

Ground loss on turns

A horse drawn wide, or taken wide by its jockey, runs a physically longer race. The rule of thumb: a horse loses roughly one length per path off the rail, per turn (some research says nearer half a length on well-banked turns). For practical handicapping, use roughly 2 lengths for a 4-wide trip and 4–5 lengths for a 6-wide trip over two turns.

A horse beaten a length after racing three-wide throughout did not run a length worse than the winner — it ran further and lost a length doing it. That is an upgrade. The form note "raced wide" is a ground-loss flag; combined with a rating a few points below the horse's norm, it is one of the cleanest excuses in handicapping.

Why MWP doesn't publish an explicit ground-loss number. It is a deliberate choice. Jockey path is partly the rider's decision, so an automatic debit would partly punish bad rides; the effect is already partly inside the horse's actual time; accurate path measurement isn't available across every track in every country; and an outside gate already carries a gate-bias debit, so a ground-loss debit on top would double-count. Treat ground loss as a manual tiebreaker when you watch replays — "wide both turns", "no cover", "stuck inside" are handicapping gold.

Rail position on turf

Turf courses use moveable inner rails to spread the wear. The rail might sit at 0 feet or be moved out 9, 18, even 36 feet. Each setting changes the effective track size and the bias. The general rule: the further the rail is moved out, the more the course favours off-the-pace horses — fresh grass appears across the width and closers have room to rally outside. Rail set tight (0–9 ft) → expect inside-speed bias; rail set wide (24–36 ft) → expect closers to do better.

Horse for course — track familiarity

A horse that has won at today's track before carries information the bare ratings miss — and the market doesn't fully price it. We measured it across the MWP database, market-relative:

Two honest caveats. It is jurisdiction-dependent: a real edge in the smaller and mid-sized markets above, priced out entirely in the deep pools of Britain and Australia, and actually inverted in the modern US. And it is the win at the track that counts — merely having run there without winning is neutral. So the practical move is a downgrade, not an upgrade: be most suspicious of a well-fancied horse shipping to a track it has never handled, especially in the smaller jurisdictions.

Track bias — the daily version

Gate statistics capture the long-term bias. The daily bias is different — weather, rail position and ground maintenance can make a track favour inside speed one day and outside closers the next. Watch the first four or five races: if every winner led from an inside draw, you have a live bias. Apply it to the rest of the card.

Bias matters most when you're holding a horse whose chance depends on it. A speed type at a speed-favouring oval has a structural tailwind; the same horse in a fair race becomes a different bet. The market only partly prices this in — which is the opportunity, and (when you're on the wrong end) the trap.

<!-- TODO (Aksel/team): bias may eventually warrant its own chapter with deeper prose, sectional analysis, and rail-position case studies. For now lives here under gate-stats. -->

Story 1: speed wins on Dubai dirt — the false dominance trap

Meydan dirt produces some of the most reliably overrated form figures in international racing. The combination of a long uncontested run to the first turn, a forgiving surface for front-runners, and a tendency for fields to give up the early lead means wire-to-wire winners look more dominant than they really are. A 5-length romp from the front on Meydan dirt is often the figure of a comfortable 1.5-length win in a fair race — but the form line records the 5 lengths.

MAGIC ART had posted exactly such a win in his previous start. Watch it — note how the field never gets close enough to contest him:

MAGIC ART — the big speed win that set the scene

The race the form is from in the case below is the next start. He went off 3.2 in the market. MISS YEECHANCE — a known fast starter — was drawn inside him. The chain that had to hold for a repeat — get the lead → bias amplifies → win comfortably — needed all three links. The first one broke. Watch #3: he didn't get to the front, ran wide, and was beaten well:

MAGIC ART #3 — the race the form is from (didn't get the lead)

The honest question for the case isn't "should you have known?" — most of it was knowable. The question is whether 3.2 covered an honest read of the risk.

Story 2: Bro Park — the day the bias went missing

The other side of structural bias: it can be deliberately removed.

LORD MOUNTBATTEN was the banker. A known front-runner at Bro Park, which has a long, documented speed bias. The distance was slightly longer than ideal but on this track that didn't worry me. ESPEN HILL, the favourite, was back from injury, faced a surface that's acceptable but not his best, and was riding Bro Park for the first time — a profile that should have been impossible to win from off the pace on this track.

It set up perfectly. LORD MOUNTBATTEN broke alertly, took the lead, and ran his race. I would have laid £20,000 at 1.01 in-play if I could have.

Then ESPEN HILL did something I've almost never seen at Bro Park — a sustained outside run from the back, catching LORD MOUNTBATTEN on the line. In the winner's interview the trainer — who rode the horse herself and is a real crowd favourite at the track — credited the track staff who had worked all week to neutralise the speed bias, and succeeded. I found out at the same moment as everyone in the grandstand.

LORD MOUNTBATTEN at Bro Park — caught on the line

A great win, a great trainer story, a great track team. The bet was right; the world changed underneath it on the day. The case is what you do with that information next — and what you do with the version of yourself who wants to abandon the process after a loss like that.

Practical: the draw bites

Three real races where the gate did the talking — a fancied favourite undone by a wide draw, an inside-bias confirmation, and a ground-loss-explains-the-loss upgrade.

Reading our form going in (Akashvani): the 2.4 favourite is the form pick (138.1) but drew gate 13 — which wins 4.8% at Sha Tin 1200m against 12% on the rail. He's fighting the draw before the gates open; the gate-2 horse (Harold Win, a K W Lui runner) is the one the draw quietly favours.
Reading our form going in (Triumphant More): a 17/1 also-ran on figures — but drawn 2, one of the best gates here (+2.2 edge). In a level handicap the structural lift of a top gate on an overlooked horse is exactly the value the market leaves on the table.
Reading our form going in (Huge Wave): his last figure is deflated — "rapid progress on the outer," a wide, ground-losing trip. Correct for the path and he's a 138–140 horse priced (9.0) like a 134, with a barn (Jimmy Ting) that has its runners fit. The result line buried the run.

A real ground-loss case: Docklands — Diomed Stakes G3

Watch #5. Docklands dwelt at the start, was in touch with the leaders, but when she needed a run over 3 furlongs out, there was no room. Switched right to the outside over 2f out, found top gear again and closed well — but the ground she lost on Epsom's two turns was too much to overcome.

Diomed Stakes G3, 7 June 2025 — Docklands #5

The subtle version of trouble: not a dramatic bumping incident, but a horse denied a run at the moment that mattered, forced to shift wide, and asked to make up lengths that physics says she can't. These are the excuses the public underweights — because on paper, "3rd beaten 1½ lengths" looks fine.

Racing Post result card with silks

Reading our form going in: at Royal Ascot next time, that Epsom prep — slow break plus waiting for room on a turning track — had capped a figure already good enough for a flat, fair mile. Adjusted for the trip she reaches the principals; 15/1 prices the bare "beaten 3rd," not the run.

Quiz

Up next: pace — how the shape of the race decides who can win before the gates open.