The greyhound box draw allocates each dog's starting box before a race, from box 1 against the inside rail to box 8 on the outside. It matters because the box sets how much ground a dog covers and which dogs it has to cross before the first turn.
Whether box 1 wins more than its share depends on the track, the distance and the dogs that draw it. To find out, count each box's wins against its starters in official results, compare them with what field sizes alone would give, and trust only a gap a large sample supports. Outside preferential box draw races the draw is random, as how greyhound racing works explains, so a wide runner is as likely as any other dog to draw box 1.
Why inside boxes suit railers and outside boxes suit wide runners
Two things make a box good or bad for a dog: the ground it has to cover, and the traffic between it and the line it wants.
Ground first. The glossary of Greyhound Racing Victoria (GRV) says a trip is usually measured a metre out from the rail, so a dog racing wider runs further than the advertised distance, and the extra builds up on every bend.
Example: Illustrative geometry, not a measured track. Treat a bend as a half circle. Running 1 metre wider all the way round adds pi x 1 = 3.14 metres, because a half circle is pi times its radius long. At an illustrative 17.5 metres a second that takes 3.14 / 17.5 = 0.18 seconds, about 2.6 lengths at the 0.07 seconds a length set by the Greyhounds Australasia Rules (GAR rule 113).
Traffic second. A railer, a dog that hugs the inside rail, has nobody to cross from box 1; from box 7 it has six dogs between it and the rail, so it must cross them or race wide. A wide runner, a dog that races out from the field, has open track outside it in box 8; from box 1 it has to angle out across the field, hampering itself and the dogs it crosses.
That clash is why Victoria, Queensland and South Australia run preferential box draw races, which sort dogs by early running style before the boxes are drawn (GRV local rule 54A; Racing Queensland grading procedures, April 2025). Greyhound Racing SA's policy bases early racing traits on IsoLynx tracking data where it can, for races graded since 5 June 2026 (checked October 2026). In Victoria's and Queensland's, box 1 goes to a dog classed as a railer whenever the field has one, so its record measures railers more than the box. Greyhound run styles shows how each style meets each box.
Does box 1 win more? Why track and distance change the answer
Box 1 can win more than its share at one track and distance and not at another, for reasons you can check before you trust a number:
- The run to the first turn. Each trip starts from its own point on the track, so the stretch between the boxes and the first bend changes with the distance, and with it the value of the inside line.
- The shape. A straight track has no bend to cut, while a two-turn trip gives an inside dog two bends to save ground on.
- The dogs. In a random draw any style can land in box 1, while a Victorian or Queensland preferential draw race gives it to a railer whenever there is one.
- The field size. Fewer starters raise every dog's even share, and a gap in the line-up alters the jump for the dogs next to it.
- Changes to the track. A new surface or a moved start alters the race, so a sample from before the change describes a different track.
No box figures from other sites appear here, because a figure is only as good as its sample, period and track. The numbers below are hypothetical, and any figure you use should carry its track, distance and dates. The greyhound racing betting hub covers the other form factors a box figure sits beside.
Vacant boxes and how they change the draw
Rule 85(3) of the GAR sets the empty boxes for every short field at the box draw:
| Dogs at the box draw | Boxes left vacant |
|---|---|
| 7 | 5 |
| 6 | 3 and 6 |
| 5 | 3, 5 and 7 |
| 4 | 2, 4, 6 and 8 |
| 3 | 2, 4, 6, 7 and 8 |
| 2 | 2, 4, 5, 6, 7 and 8 |
GRV can set the vacant boxes itself instead (local rule 41). After the draw the pattern changes. A late scratching empties only the scratched dog's box, and nobody moves (rule 85(6)). A reserve brought in at the appointed scratching time moves into the box the scratched dog left (rule 85(5)). When more than one dog is scratched, the reserve's box is drawn by ballot, unless the controlling body decides another way (rule 85(12)).
A vacant box gives its neighbours room. With box 5 empty, the dog in box 4 has space on its outside and the dog in box 6 on its inside, which suits a wide runner in 4 and a railer in 6. Short fields also lift the even share: 1 / 7 = 14.3% per dog with seven runners and 1 / 6 = 16.7% with six, against 1 / 8 = 12.5% in a full field.
How to measure greyhound box bias yourself
- Choose one track and one distance, and write down the first and last date of your sample.
- From official results, such as GRV's FastTrack results, record each runner's box and finishing place and each race's field size (checked October 2026). Its winning-boxes statistic gives a quick first look, but check what it counts before you treat it as bias.
- Leave out preferential box draw races, or count them separately.
- For each box, count starters and wins, and add up par wins: 1 / field size for every start, so a box 1 start in a seven-dog race adds 1 / 7.
- Work out strike rate = wins / starters, and par rate = par wins / starters.
- Work out standard error = square root of (par rate x (1 - par rate) / starters). The box's distance from par is (strike rate - par rate) / standard error.
- Treat anything inside about two standard errors as noise, and start a fresh sample when the track changes.
Worked: wins against starters for every box
Example: Hypothetical sample, not a real track: 500 races at Track A over 520m. Of these, 400 had eight dogs, 70 had seven (box 5 empty) and 30 had six (boxes 3 and 6 empty), with no late scratchings to keep the counting simple.
| Box | Starters | Wins | Strike rate | Par wins | Par rate | Standard error | Standard errors from par |
|---|---|---|---|---|---|---|---|
| 1 | 500 | 88 | 17.60% | 65 | 13.00% | 1.50 points | +3.1 |
| 2 | 500 | 70 | 14.00% | 65 | 13.00% | 1.50 points | +0.7 |
| 3 | 470 | 52 | 11.06% | 60 | 12.77% | 1.54 points | -1.1 |
| 4 | 500 | 58 | 11.60% | 65 | 13.00% | 1.50 points | -0.9 |
| 5 | 430 | 47 | 10.93% | 55 | 12.79% | 1.61 points | -1.2 |
| 6 | 470 | 51 | 10.85% | 60 | 12.77% | 1.54 points | -1.2 |
| 7 | 500 | 59 | 11.80% | 65 | 13.00% | 1.50 points | -0.8 |
| 8 | 500 | 75 | 15.00% | 65 | 13.00% | 1.50 points | +1.3 |
Working for box 1: par wins = 400 / 8 + 70 / 7 + 30 / 6 = 50 + 10 + 5 = 65, so par rate = 65 / 500 = 13.00%. Standard error = square root of (0.13 x 0.87 / 500) = 0.0150, or 1.50 points, and (17.60 - 13.00) / 1.50 = 3.1 standard errors.
Box 1 stands out, with 88 wins against a par of 65, an index of 88 / 65 = 1.354. Box 8's 75 wins look strong, but 1.3 standard errors is within luck, as are the middle boxes' shortfalls. Boxes 3, 5 and 6 also have lower par rates because they miss some short fields, where the even share is higher.
Now cut the sample to its first 40 races, all full fields, and say box 1 won 9 of them: 9 / 40 = 22.5% against a 12.5% par. Standard error = square root of (0.125 x 0.875 / 40) = 0.0523, so the gap is (22.5 - 12.5) / 5.23 = 1.9 standard errors. That looks like a bias and probably is not.
A box with no edge at all wins 9 or more of 40 races about 5.5% of the time (the binomial chance at 1 in 8). But you are watching eight boxes. If each race's winner is equally likely to come from any of them, at least one box reaches 9 wins about 40.7% of the time (the multinomial chance over all 8 to the power of 40 equally likely outcomes).
The starts a box needs before a gap from a 12.5% par reaches two standard errors are 4 x 0.125 x 0.875 / gap squared, with the gap as a fraction:
| True strike rate | Gap from 12.5% | Starts needed |
|---|---|---|
| 17.5% | 5 points | 175 |
| 15.0% | 2.5 points | 700 |
| 13.75% | 1.25 points | 2,800 |
Sample size in betting covers the same problem for betting records.
Using a box figure without overrating it
Use a box figure only when all four of these hold:
- The sample covers several hundred races at that track and distance since the track last changed.
- The box sits more than about two standard errors from par.
- The direction makes sense for the track, such as an inside edge on a trip with a short run to a sharp first bend.
- The figure holds with preferential draw races left out.
Even then, pull it back toward par, because the biggest gap in any sample is partly luck. One illustrative rule halves the gap, so box 1's index becomes 1 + (1.354 - 1) / 2 = 1.18.
Apply it to the dog, not the box. An index averages every style that drew the box, so a railer in box 1 can expect more from it than a wide runner. A rating that already credits a dog for leading from box 1 should not get the box added again.
Risk: Betting involves risk. A box figure is an estimate from a sample that luck can bend, past results are no guarantee of future results, and a dog in a good box still loses most of its races. Backing it too short costs money however strong the bias looks. See responsible gambling for limits and support.
Mistakes with greyhound box statistics
- Trusting another site's figures. Without the track, distance, dates and field sizes you cannot tell a real bias from a lucky run.
- Counting the box a dog drew rather than the box it started from, when reserves and late scratchings move dogs and gaps around.
- Comparing raw strike rates instead of each box's strike rate against its own par.
- Mixing in preferential box draw races, where the draw follows the dogs' styles, or mixing distances in one sample.
- Paying for a bias that is not there. With illustrative numbers, say an undated box figure lifts a dog in box 1 to 20% on your sheet, a fair $5.00, so $5.50 looks like a 5.50 / 5.00 - 1 = 10% edge. If the bias is not real and the dog's chance is 14%, each $1 returns 0.14 x 5.50 = $0.77 on average, a 23% loss, or about $115 over 50 bets of $10.
Where B337 fits
B337 does not publish box statistics or rate dogs; it compares prices, so the box read stays yours.
On the Terminal's racing board each bookmaker that prices a greyhound race sits in its own column beside exchange back and lay prices, with price history (flucs) showing whether a dog has firmed or drifted. Not every bookmaker prices every race, so the columns follow the market. Prices come from 40+ bookmakers across racing and sports and are read on a repeating cycle, so a price on screen can trail the bookmaker's own: check it in your account before betting. A free account opens a limited view of the Terminal with live odds.
Risk: Betting involves risk. Seeing the prices on a dog after the draw does not make its box an edge, and a bet at the best price available can still lose. See responsible gambling for limits and support.