Greyhound ratings score each dog's recent runs against the clock, so a field of eight can be compared and priced. To build your own, set a dated time standard for each track and distance and adjust each run's time for the night's track speed, the box and the dog's weight. Measure the adjusted time in lengths against the standard, average a dog's recent runs into one rating, then convert the field's eight ratings into rated prices that add to 100%.
The national rules make the arithmetic simple: under the Greyhounds Australasia Rules (GAR), races are timed electronically and 0.07 seconds counts as one length (rule 113). B337 does not rate dogs or pick winners, so any rating here is one you build and test yourself; the greyhound racing betting hub covers the races and bets the ratings feed into.
Speed ratings and other greyhound rating systems
A greyhound rating system is any method that puts dogs on one scale. Three are common:
- A time rating, also called a greyhound speed rating, compares a dog's finishing time with a standard for the track and distance. Steps 1 to 5 build one.
- An early speed rating does the same with the first split, the time to the first timing point, and feeds the first-turn read.
- A grade is the class system each state runs. It decides which races a dog may enter, not how fast it is.
Ready-made ratings from a form service need the same Step 5 test before you trust their prices.
Dogs carry no weight, so a greyhound time rating needs no allowance for weight carried, unlike a horse rating. Horse racing ratings covers official handicap ratings and speed ratings for horses.
Step 1: set a time standard for each track and distance
Times compare only at the same track and distance, so each standard covers one track and one distance, built from official results. State controlling bodies publish them: Greyhound Racing Victoria's FastTrack results, for example, list each runner's box, rug, weight, first split time, finish time and margin (checked 7 October 2026). To set a standard:
- Collect the winning times at the track and distance over a fixed period, and write the period down.
- Keep the sample to one surface and set-up: if the track is resurfaced or a start is moved, begin a new standard.
- Take the median, the middle time, so one freak run cannot drag the standard.
- Date the standard and refresh it as new results come in.
Example: Illustrative numbers for a hypothetical track. At Track A over 500m, the median of 150 winning times from one period is 29.40 seconds. A run of 29.40 rates 0; a faster run rates above it and a slower one below.
A beaten dog's time comes from the margin: time = winner's time + lengths beaten x 0.07. With illustrative times, a dog beaten 3 lengths by a winner that ran 29.52 ran 29.52 + 3 x 0.07 = 29.73. Where results print each runner's finish time, as FastTrack's do, use that instead. Reading the first split and the run home beside the finish is covered in greyhound sectional times.
Step 2: adjust each time for the track, the box and the weight
A raw time carries noise the dog did not cause. Remove three pieces, each measured as the seconds it is estimated to have added.
The track on the night: a meeting's variant is the average of (winning time - standard) across its races, each against its own distance's standard. With illustrative numbers, six races ran 0.15, 0.08, 0.12, 0.20, 0.05 and 0.12 seconds slower than standard: (0.15 + 0.08 + 0.12 + 0.20 + 0.05 + 0.12) / 6 = 0.12 seconds slow.
The box: compare the same dogs' times from one box with their times from the middle boxes at the same track and distance, so class does not leak in. In a hypothetical sample at Track A over 500m, dogs ran 0.05 seconds slower from box 8 than from boxes 3 to 6: the box 8 allowance is 0.05 seconds, worth trusting only after many runs from that box.
The weight: a dog's body weight is a fitness signal. The Stewards publish each dog's weight before its race, recorded to the nearest 0.1 kilogram (GAR rule 100). A dog whose weight varies by more than 1.0 kilogram from its last race or satisfactory trial may not run without the Stewards' permission, and after 28 days without a run they must allow up to 2.0 kilograms (rule 101).
Within that band, test a simple rule against each dog's own record, such as an illustrative 0.02-second allowance for a run 0.5 kilograms or more away from the weight of its best recent runs.
| Step | Change in seconds | Running time |
|---|---|---|
| Raw time, beaten 3 lengths by a 29.52 winner | 29.73 | |
| The night's variant | -0.12 | 29.61 |
| Box 8 allowance | -0.05 | 29.56 |
| Weight allowance | -0.02 | 29.54 |
Rating in lengths = (standard - adjusted time) / 0.07 = (29.40 - 29.54) / 0.07 = -2.0, so the run rates two lengths slower than standard.
Step 3: combine a dog's recent runs into one rating
One run is a sample of one. Weight the recent runs at today's distance, for example the last four at 4, 3, 2 and 1 from the latest. With illustrative ratings of -2.0, -1.0, -2.6 and -0.8 lengths, latest first:
rating = (4 x -2.0 + 3 x -1.0 + 2 x -2.6 + 1 x -0.8) / (4 + 3 + 2 + 1) = -17.0 / 10 = -1.7 lengths
Before you average, read the race comments. A dog that was checked, crowded or slow away ran a time that measures its trouble, not its ability, so leave those runs out or weight them down, and note which you did. How to read greyhound form decodes the comments.
A dog stepping up in trip or new to the track has no time at today's track and distance, and a maiden may have only trial times: treat those ratings as rough, or leave the race alone.
Why a few runs cannot settle a rating
Even after every adjustment, a dog's times scatter from run to run, and the uncertainty in its average falls only with the square root of the number of runs. With an illustrative scatter of 0.15 seconds (one standard deviation):
| Runs at the track and distance | Standard error = 0.15 / square root of runs | In lengths = seconds / 0.07 |
|---|---|---|
| 1 | 0.1500 seconds | 2.14 |
| 3 | 0.0866 seconds | 1.24 |
| 6 | 0.0612 seconds | 0.87 |
| 12 | 0.0433 seconds | 0.62 |
After three runs a rating still carries a standard error of about 1.2 lengths, more than the gap between many dogs in a field. Two dogs rated half a length apart on a few runs each are, as far as the data can tell, level.
Step 4: turn eight ratings into rated prices
A rated price needs a chance for every dog, and the chances must add to 100%. One simple method gives each dog a weight that falls by the same factor for every length it is rated behind the top dog, then shares the field by weight:
weight = factor to the power of (lengths behind the top-rated dog)
share = weight / total of the eight weights, and rated price = 1 / share
The factor sets what a length is worth. This example uses 0.6, an illustrative figure: choose yours by testing which factor makes your past rated chances match how often dogs won.
Example: A hypothetical eight-dog race, ratings in lengths against the track standard, factor 0.6. The eight weights total 3.5474.
| Box | Rating | Lengths behind the top dog | Weight | Share | Rated price |
|---|---|---|---|---|---|
| 1 | +1.0 | 0 | 1.0000 | 28.19% | $3.55 |
| 2 | -0.6 | 1.6 | 0.4416 | 12.45% | $8.03 |
| 3 | +0.2 | 0.8 | 0.6645 | 18.73% | $5.34 |
| 4 | -1.7 | 2.7 | 0.2518 | 7.10% | $14.09 |
| 5 | -2.4 | 3.4 | 0.1761 | 4.96% | $20.15 |
| 6 | -0.2 | 1.2 | 0.5417 | 15.27% | $6.55 |
| 7 | -1.1 | 2.1 | 0.3421 | 9.64% | $10.37 |
| 8 | -3.0 | 4.0 | 0.1296 | 3.65% | $27.37 |
Box 2 works as 0.6 to the power of 1.6 = 0.4416, a share of 0.4416 / 3.5474 = 12.45%, and a rated price of 1 / 0.1245 = $8.03. The shares add to 99.99% after rounding. Rated prices are worked from the unrounded shares. A time rating says nothing about today's draw, so adjust the shares for the first turn afterwards, as greyhound betting strategy does; how to frame a betting market covers framing in general.
Then set the rated prices beside the market. With illustrative prices:
| Box | Rated price | Best bookmaker price | Exchange midpoint | Read |
|---|---|---|---|---|
| 1 | $3.55 | $3.80 | $3.70 | Above both |
| 3 | $5.34 | $4.60 | $4.80 | Below your price: no bet |
| 6 | $6.55 | $8.50 | $8.80 | Above yours, below the exchange |
Box 1: 3.80 / 3.5474 - 1 = 7.1% above your price and 3.80 / 3.70 - 1 = 2.7% above the exchange. Box 6: 8.50 / 6.55 - 1 = 29.8% above your price, but 8.50 / 8.80 - 1 = -3.4% against the exchange. Treat a gap that wide as a reason to recheck the rating, for a first-turn conflict or a stewards' comment, before you treat it as value.
Risk: Betting involves risk. Your rated price carries the error of every assumption behind it, and a bet above it can still lose, and often does. See responsible gambling for limits and support.
Step 5: test the ratings against results and the closing price
A rating earns trust only from a record of a few hundred races. Run two tests.
Calibration: group dogs by rated chance and compare the winners you expected with the winners there were. Illustrative results from 100 eight-dog races:
| Rated chance | Dogs | Expected winners (sum of rated chances) | Actual winners |
|---|---|---|---|
| 25% and over | 100 | 30.0 | 27 |
| 10% to under 25% | 300 | 45.0 | 46 |
| Under 10% | 400 | 25.0 | 27 |
For the top band, luck alone moves the count by about the square root of (100 x 0.30 x 0.70) = 4.6 winners either way. So 27 against 30 is no evidence the ratings are wrong, and no evidence they are right either.
The closing price: for every dog you bet, record the exchange's final price at the jump. Closing line value = price taken / final price - 1, so an illustrative $3.90 taken against a final $3.60 is 3.90 / 3.60 - 1 = 8.3%.
If the prices you take on your rated overlays keep beating the final price, the market is moving towards your ratings; if they keep losing to it, the market knows something your ratings miss. Closing line value explains why this test needs far fewer bets than profit does.
Note: Past results are no guarantee of future results. A well-calibrated record shows your ratings matched past races, not what the next race will do.
Mistakes that break a home-made rating
- Comparing raw times across tracks or distances: a 29.40 at one track and distance says nothing about a 29.40 at another.
- Keeping a standard after the track changes, so every rating there drifts.
- Trusting one fast run: on the 0.15-second scatter above, a single run carries a standard error of 2.14 lengths.
- Rating a checked or crowded run as if it were clean.
Setting your rated prices against the Terminal
The rating half of Steps 4 and 5 stays yours; the market half is where B337 helps. The Terminal lines up prices from 40+ bookmakers across racing and sports, greyhound races included, beside Betfair back and lay, flucs and closing lines. Each rated price can then be read against the best available price, the exchange and the last price before the jump.
Its EV overlays answer a different question from your rating: they compare each price with a fair price estimated from Betfair's market. A dog can clear your rated price and show no overlay, or the reverse. Prices are read on a repeating cycle, so confirm the figure in your bookmaker account before you bet. 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. Your rating and the market can both be wrong, so a gap between them is a question to check rather than a promise, and there is no guarantee of profit. For free and confidential support call 1800 858 858 or visit gamblinghelponline.org.au, and see responsible gambling.