Every global odds API skips Australian racing. Here is what a complete racing data API actually needs — fields, fixed odds, exchange, tote, sectionals, connections — and who carries it.
Backtest your betting model against 232K+ historical odds records. Step-by-step Python tutorial with real data from the Krok Odds archive.
Racing arbitrage is more available in Australia than most punters realise, and less available than arbitrage marketing suggests. Both statements are true, and the gap between them is where all the practical detail sits.
The structural case is strong: Australian racing markets carry 14-25% margin against 4-6% on main sports markets, the Betfair Exchange offers deep racing liquidity and lets you cover a single runner, and thoroughbred racing is the only Australian betting market with legislated minimum bet obligations. That combination produces genuine, regular opportunities.
The constraint is execution — prices move, legs fail to fill, stakes get limited, and accounts get restricted. This guide covers the mechanics, the arithmetic, and the failure modes.
The dominant form in racing. Back a runner at a fixed-odds bookmaker, lay the same runner on Betfair at a shorter price. You have covered exactly one runner and you profit whichever way it goes.
This works in racing where it is impractical in most sports because you do not need to cover every outcome — laying one runner in a 14-horse field is a single transaction.
Backing every runner in a field at different bookmakers such that the implied probabilities sum below 100%. In practice this is rare in racing, because covering 14 runners means 14 bets across many accounts with prices moving underneath you. It occurs mostly in small fields — four and five-horse races, and match-race style markets.
For back/lay, three numbers determine everything: the back price, the lay price, and the commission rate.
Lay stake = (back stake × back price) / (lay price − commission adjustment)
More usefully, in the form you actually use:
Lay stake = (back odds × back stake) / (lay odds − commission × (lay odds − 1))
Say a runner is $6.00 at a corporate bookmaker and $5.40 to lay on Betfair, with 6.5% commission.
| Leg | Detail | Amount |
|---|---|---|
| Back | $100 at $6.00 with the bookmaker | Returns $600 if it wins |
| Lay | $113.90 at $5.40 on Betfair | Liability $500.90 |
| If the runner wins | +$500 bookmaker, −$500.90 exchange | Roughly −$0.90 |
| If the runner loses | −$100 bookmaker, +$113.90 less 6.5% commission | +$6.50 |
That configuration is not a true arbitrage — it is unbalanced. Rebalancing to equal profit either way gives roughly +$2.30 on $100 staked, about 2.3% on the back stake. That is a typical racing arbitrage: small, real, and requiring both legs to fill.
The arbitrage calculator handles the rebalancing including commission — doing it in your head under time pressure is how people end up with unintended positions.
Understanding the source tells you where to look.
| Source | Frequency | Typical size | Where |
|---|---|---|---|
| Bookmaker slow to reprice after a plunge | Common | 2–6% | All meetings, minutes before jump |
| Exchange drifting on a runner books like | Common | 1–4% | Metro meetings |
| Post-scratching repricing lag | Regular | 3–10% | Any code |
| Barrier draw reprice lag | Regular (majors) | 3–15% | Feature races |
| Promotional price boosts | Regular | 5–20% | Metro Saturdays, capped |
| White-label feed lag | Common | 2–8% | Smaller brands |
| Genuine model disagreement | Occasional | 1–3% | Harness, greyhounds |
Two of these are worth highlighting. Promotional boosts produce the largest arbs available to a retail punter, because the bookmaker is deliberately pricing above fair and funding it from marketing budget rather than from the book. They are capped, usually to $50 or $100, but the percentage is excellent.
White-label feed lag is the most consistent and least crowded source. The dozens of brands running on shared platforms update on the platform's cycle, which is often slower than the corporate books. When a price moves at Sportsbet, the white-label brands frequently take minutes to follow.
Arbitrage frequency scales with market inefficiency, which in racing scales with neglect.
| Segment | Fixed-odds margin | Exchange liquidity | Arb frequency | Practical? |
|---|---|---|---|---|
| Metro thoroughbred Saturday | 14–18% | Deep | Moderate | Yes — best combination |
| Midweek metro thoroughbred | 15–19% | Good | Moderate | Yes |
| Country thoroughbred | 16–22% | Thin | High | Limited by liquidity |
| Metro harness | 16–20% | Moderate | High | Yes — underused |
| Country harness | 18–24% | Very thin | Very high | Often not executable |
| Metro greyhound | 18–22% | Moderate | High | Yes |
| Country greyhound | 20–25% | Very thin | Very high | Rarely executable |
| Feature race days | 14–18% | Deepest of the year | High around draws | Yes — best days |
The pattern is a trap: the segments with the most arbs are the ones where you cannot get set. Country greyhound markets throw off apparent opportunities constantly and the exchange side frequently has $40 available. Metro harness is the sweet spot most people miss — genuinely high frequency with enough liquidity to matter.
The dominant risk. Prices in the final ten minutes before a jump move constantly, and an exchange lay placed at a price that has already gone will sit unmatched. You are then holding a naked back bet you did not intend.
Mitigations: take the thinner side first; do not chase an arb that has already moved; and size to what is actually available rather than what is quoted.
Specific to racing and frequently overlooked. If a runner is scratched after your fixed-odds bet is placed, the bookmaker applies a rule-4 deduction to your winnings. Your Betfair lay is not adjusted the same way — exchange markets are re-formed. The result is that a scratching can turn a locked position into a small loss.
Mitigation: prefer arbitrage closer to the jump when the field is more settled, and account for it in your expected return rather than pretending it does not exist.
Inevitable, not possible. Bookmakers identify arbitrage patterns quickly — round-number stakes, betting into every price boost, always taking the top of the market, activity concentrated in the minutes before jump.
Australian thoroughbred racing has a partial defence that sport does not: minimum bet rules in NSW and Victoria require bookmakers to lay a bet to lose a set amount on thoroughbred racing within specified windows. That is a genuine legal floor, but it applies only to thoroughbreds, only in those jurisdictions, and only within the specified windows — harness and greyhound racing have no equivalent.
A back/back "arb" between Ladbrokes and Neds is not an arb — they are the same book and the position is visible to Entain as one customer. The same applies across the white-label platforms, where dozens of brands share one price. Any scanner not suppressing these is producing opportunities that will be voided.
Betfair is the deliberate exception: Sportsbet and Betfair share a corporate parent, but the exchange is an independent price and those pairs are legitimate.
| Metric | Realistic figure |
|---|---|
| Return per opportunity | 1–4% after commission |
| Opportunities per day (executable) | A handful, not dozens |
| Average stake before restriction | $50–$500 depending on book and code |
| Time before first restriction | Weeks to a few months |
| Capital required | Substantial — exchange liability is large relative to back stake |
| Effective hourly rate | Modest, and it declines as accounts close |
Anyone presenting racing arbitrage as a reliable income should be asked how many accounts they still have open. The activity is real and the returns are real; the constraint is that the counterparties get to stop trading with you, and eventually they do.
Worth knowing about, because it is where most of the practical value sits.
A low hold is a market where the best available prices across bookmakers sum to slightly above 100% — say 100.5% to 102% — rather than below. It is not a guaranteed profit, but you are betting into a 1-2% margin instead of an 18% one.
Low holds are dramatically more common than true arbs, involve no timing pressure, and do not look like arbitrage to a bookmaker's risk system. For most punters they are a better use of attention than chasing the rarer, faster-moving true arb.
See the low-hold screen for the computed version across racing and sport.
Racing arbitrage suits you if you have meaningful capital, several funded accounts, the discipline to record every transaction, and the tolerance for a business model whose counterparties will progressively refuse to trade with you.
It does not suit you if you are looking for a passive income, if your capital is limited (exchange liability requirements are unforgiving), or if you want to keep unrestricted accounts for other purposes.
For most punters, the sensible version of the same insight is simpler: use the exchange as your default for racing, take the best available price on every bet, and treat low holds as the everyday version of arbitrage. That captures most of the structural advantage with none of the execution risk.
Krok Odds scans 140+ Australian bookmakers against the Betfair Exchange across every thoroughbred, harness and greyhound meeting, with same-corporate-group pairings suppressed. Free tier available.
Open the arbitrage scanner →
David has been running advantage betting strategies across Australian bookmakers since 2023 and contributes long-form retrospectives, case studies, and operational pieces drawn from years of running real bets in AU markets. His writing focuses on the realities of running a sustainable AU advantage operation — what works, what fails, and the operational details most blogs gloss over.
Stop guessing on greyhounds. Use box draw data, speed ratings and track patterns to build a systematic greyhound betting strategy.
