Track bias in horse racing is when one running position, such as the lead, the rail or wide out the back, produces more winners on a race day than the runners' prices predicted. The ground, the rail position and the way races are run can all produce one, and two results that were pure chance can look exactly like it.
So measure a bias on the day against what the market expected, and use it in small adjustments. Most of what you need to measure one is on Racing Australia's meeting pages (checked 7 October 2026), and every price in the worked examples is illustrative. A bias is one input among many in how to read a form guide.
Leaders, inside runners and off-pace runners
A bias is named after the runners it helps:
| Bias | Runners winning more than expected | Who gains | Who loses |
|---|---|---|---|
| Leader bias | Horses leading or close to the lead at the 800m | Front runners, and the runner in the box seat, on the rail one back from the leader | Backmarkers |
| Inside bias | Runners close to the rail, above all in the straight | Horses that can hold the fence | Runners caught three or four wide |
| Off-pace or wide bias | Runners from midfield or behind, finishing wider out | Swoopers, which Racing Victoria's glossary describes as racing near the back before a fast finish, often down the outside | Leaders and runners stuck on the inside |
More than expected is the whole test. Leaders win plenty of races on a fair track, because they can save ground and avoid traffic, and a short-priced leader is expected to win often. A bias shows only when results beat the chances the prices gave those runners.
Rail position and its usual effect
Racing Australia's meeting pages state where the rail is set. Notations from its Victorian track conditions page, checked 7 October 2026:
| Rail notation | What it means |
|---|---|
| True Entire Circuit | The rail is at its innermost line all the way round |
| Out 3m Entire Circuit | The rail is 3 metres out from that line all the way round |
| Out 2m 900m - 400m, True Remainder | Out 2 metres from the 900m mark to the 400m mark, and true elsewhere |
| Out 9m 1700m - 800m, Out 11m Remainder | Out 9 metres for part of the circuit and 11 metres for the rest |
Moving the rail shifts the racing line onto a different strip of turf. When the rail goes out, it covers the strip earlier meetings raced over, and runners near the new rail race on ground with less wear; the usual reasoning is that this fresh strip helps horses that can hold the fence.
When the rail stays in one position for several meetings, the strip next to it takes the wear, so late in that run the inside can be the most worn part of the track. When the rail comes back in, the strip it had covered has been rested. None of this is a rule, so treat the rail as a reason to watch the early races closely rather than as a bias in itself.
The ground also wears as the card goes on. A Soft 7, in Racing Australia's wording, is a track that will chop out, so on a wet day the strip nearest the rail can deteriorate race by race. Club notes on Racing Australia's track conditions pages sometimes record work on part of the course, such as an outside strip aerated from the 350m mark to the winning post, which can leave one lane racing differently from the next. How the rating itself moves is covered in track ratings explained.
How to read a track bias today
The bias that matters is today's, built race by race:
- Before the first race, check the meeting page for the rail, the rating and any notes on the track.
- After each race, note where the winner and placegetters were at the 800m and 400m, from the replay, and roughly where they finished in the straight: near the rail, in the middle or out wide. Racing Australia's form prints the same 800m and 400m positions in each runner's line for its next start.
- Write down the starting price of the horse that led and of the winner. Those prices give the chances the market expected.
- Note the pace. A leader that won a slowly run race was helped by the tempo, which a speed map should have flagged, not necessarily by the track.
- Start a fresh count after a rating change or a rail move between races.
The stewards' report for each race adds what a replay can miss, such as runners that raced wide without cover or were held up for room.
Small samples: why two races are not a pattern
Take a meeting where the leaders win the first two races at illustrative prices of $4.00 and $5.00. Their implied chances are 1 / 4.00 = 25.0% and 1 / 5.00 = 20.0%, so both winning has a 0.25 x 0.20 = 5.0% chance on a fair track. That looks rare at one meeting. Across 10 meetings on a Saturday, each with the same 5.0% chance, at least one opens that way 1 - 0.599 = 40.1% of the time, because 0.95 multiplied by itself 10 times is 0.599.
Six races give you more to go on, but less than they seem. An illustrative meeting:
| Race | Leader's starting price | Implied chance (1 / price) | Leader won |
|---|---|---|---|
| 1 | $4.50 | 22.2% | Yes |
| 2 | $6.00 | 16.7% | No |
| 3 | $3.30 | 30.3% | Yes |
| 4 | $8.00 | 12.5% | No |
| 5 | $5.00 | 20.0% | Yes |
| 6 | $11.00 | 9.1% | No |
| Total | 110.8%, or 1.108 expected wins | 3 wins |
Worked race by race, the chance that leaders at those prices win 3 or more of the 6 is 6.60% for exactly 3, 1.03% for 4 and 0.08% for 5, with all 6 under 0.01%: about 7.7% in all. Each figure covers every way that many leaders can win: for each way, multiply the winners' chances and 1 minus each loser's chance together, then add the products.
The implied chances include the bookmaker's margin, which pushes the test towards caution. But the prices were set before anyone knew which horse would lead. A horse that does lead saves ground and avoids traffic, so it can beat its pre-race chance on a fair track too, which pushes the other way. Read 7.7% as a rough guide, or run the count on the runner your speed map picked to lead before each race.
On those numbers, a fair track hands out a run like this to about one meeting in 13 (1 / 0.077 = 13). Every extra pattern you look for (leaders, rail runners, wide runners, backmarkers) is another chance for luck to supply one. Small samples are the reason.
A working rule, which is a judgement rather than a statistic: two races are a hint. Four to six that agree, with a cause you can point to such as a rail move or rain, justify a small adjustment. No run of results on one day justifies doubling a chance.
Worked: adjusting a price for a suspected bias
Race 7 at the same meeting. Your speed map says Runner 5 should lead from barrier 2 with no rival for the front, and before any bias you frame it at 15%, a fair price of 1 / 0.15 = $6.67. One way to size the adjustment is to blend the day's evidence with your starting belief that the track is fair:
| Step | Working | Result |
|---|---|---|
| Leaders' wins against expected | 3 / 1.108 | 2.71 times expected |
| Weight on today's races | 6 / (6 + 54), treating your fair-track belief as worth 54 races | 0.1 |
| Multiplier | 1 + (2.71 - 1) x 0.1 | 1.171 |
| Runner 5's chance | 15% x 1.171 | 17.565%, about 17.6% |
| Fair price | 1 / 0.17565 | $5.69 |
| Rest of the field | each chance x (100 - 17.565) / (100 - 15) = x 0.970 | a 12% backmarker becomes 12 x 0.970 = 11.64%, or 1 / 0.1164 = $8.59 |
The 54 is an assumption you choose, not a measured number: a larger one makes you slower to believe a bias, a smaller one quicker. Now suppose Runner 5 is offered at $6.50:
| Runner 5's real chance | What $10 at $6.50 is worth on average |
|---|---|
| 15%, the track is fair | 10 x (0.15 x 6.50 - 1) = -$0.25 |
| 17.6%, the bias is real | 10 x (0.17565 x 6.50 - 1) = +$1.42 |
The whole edge rests on the bias being real, which six races cannot prove. One result moves it a lot: had leaders won 2 of the 6, the ratio would be 2 / 1.108 = 1.81, the multiplier 1 + 0.81 x 0.1 = 1.081, and Runner 5's chance 15% x 1.081 = 16.2%, a fair price of 1 / 0.162 = $6.17 that $6.50 only just beats.
Risk: Betting involves risk. A bias adjustment is a guess about the ground stacked on a guess about the race, and a bet worth having only if the bias is real can lose even when it is. See responsible gambling for limits and support.
Bias and speed maps together
A speed map says where each runner should be early; a bias says which of those positions is paying today. Use them in that order, and count each advantage once:
- A leader bias with one natural leader on the map is the strongest case for a lift, as with Runner 5.
- A leader bias with three runners who all want to lead is weaker, because the pace they set can cancel it. The box seat runner may gain more than any of them.
- An inside bias makes the barrier draw matter more: an inside draw that holds the rail gains, while a wide runner that races without cover loses ground and the better strip as well.
- An off-pace bias with a fast pace on the map points the same way twice. A hot early pace is part of why backmarkers win, so adjust for it once, not once for the pace and again for the bias.
Mistakes that turn a bias into a bad bet
- Calling it from two races. At 10 meetings, an opening where leaders at $4.00 and $5.00 win the first two races happens somewhere 40.1% of the time.
- Ignoring the prices. Three short-priced leaders winning is what the market expected, not a bias.
- Ignoring the pace. A slowly run race suits the leader on any track.
- Carrying a bias to the next meeting. The rail and the rating can both change in between, so the count starts again.
- Applying the raw ratio. Multiplying Runner 5's 15% by the raw ratio (3 / 1.108 = 2.708) gives 40.6%, a fair price of 1 / 0.406 = $2.46. If the true chance is 15% and you take $3.00 on that read, each $10 is worth 10 x (0.15 x 3.00 - 1) = -$5.50.
Where B337 fits on a biased day
B337 does not read races or detect a bias. There is no form guide or ratings engine in it, and it makes no selections, so the race-by-race count and the adjustment stay yours. What it does is put the market in front of you. On its odds comparison board the Terminal lines up prices from 40+ bookmakers across racing and sports, with Betfair back and lay and price history (flucs). You can compare your adjusted price with the market and see how prices moved as the day's results came in.
Each price is shown as at the time displayed and read on a repeating cycle, so confirm it with your bookmaker before you bet; not every race draws a price from every bookmaker. A free account opens a limited view of the Terminal with live odds. Betfair is a trade mark of its owner, and B337 is not affiliated with it.
Risk: Betting involves risk. A price that beats your bias-adjusted frame can still lose, often several times in a row, and there is no guarantee of profit. See responsible gambling for limits and support.