What Betfair Starting Price (BSP) actually is — and why every backtest should use it

If you want to know whether a betting angle really makes money, you need one honest price to settle it at. For exchange bettors that price is the Betfair Starting Price — and understanding how it is built is the difference between a backtest you can trust and one that is quietly lying to you.

The one-sentence version

The Betfair Starting Price (BSP) is a single price per runner, in both the win and place markets, calculated automatically by Betfair at the off by matching every remaining "at SP" order against the unmatched money sitting in the exchange. Everyone who asked for SP on that runner gets the same number. Nobody had to be watching a screen to get it.

How the number is formed

In the seconds after a market goes in-play, Betfair takes two pools of money:

• SP backers and layers — people who placed an order "at Starting Price" instead of naming a price.
• Unmatched exchange bets near the top of the book that are willing to be swept up in the reconciliation.

It then finds the single price at which the SP backers' stakes and the SP layers' liability balance, using the unmatched exchange money to fill the gap. The output is the BSP. Because it is computed from the whole pool at once, it is not something you can "beat" by clicking faster — it is the market's settled opinion at the moment of the off.

Two practical points that follow from this:

• There is no minimum stake and no price you have to accept in advance. You ask for SP; you are filled at whatever the reconciliation produces.
• Historically, Betfair has published that its SP beats or matches the industry (bookmaker) SP on the large majority of runners, because the exchange has no built-in overround to feed a margin.

BSP vs bookmaker SP vs early prices

Three prices get confused all the time. They are not the same thing:

This is exactly why BSP is the right yardstick for research. An early-price backtest silently assumes you were online at 9am, took the top of every market before it moved, and were never knocked back. A BSP backtest assumes something you can actually reproduce every day: you left an SP order in and took the number.

Why a backtest that doesn't settle at BSP can't be trusted

Four failure modes disappear when you settle at BSP:

1. Price fantasy. "Best odds" datasets record the biggest price any bookmaker offered. Almost nobody gets that price at a meaningful stake. BSP is a price you'd have actually been matched at.
2. Look-ahead in the price. If your data uses a price recorded some unknown number of minutes before the off, you can accidentally backtest on information that wouldn't have been available when you had to bet.
3. Survivorship in the market. Restricted and closed accounts never show up in an early-price backtest. The exchange doesn't shut winners out.
4. Ignoring costs. Exchange profit is charged commission. A gross-return backtest can turn a real 2% loss into an imaginary 3% edge.

A rule of thumb: if you can't state the exact price your backtest assumed and explain how you would have been matched at it, the backtest isn't measuring your edge — it's measuring your optimism.

The caveats worth knowing

• Liquidity. In thin markets — minor greyhound meetings, obscure early races — BSP is still calculated, but a large stake at SP can move it. For research at sensible stakes this rarely matters; for staking plans it can.
• Well-backed runners. BSP on a heavily supported favourite is often shorter than its early price. Back systems on short prices look worse at BSP than on early prices; lay systems often look better. That's not the data being unkind — it's the early price having been unrealistic.
• Reduction factors. When a runner is withdrawn late, remaining BSPs are adjusted. Good historical datasets bake this in; be wary of ones that don't.
• Place markets. Place BSP exists and is useful, but the number of places paid and the market's maturity vary — treat win and place as separate studies.

How SportsEdgePro uses BSP

Every profit-and-loss figure in the System Builder is settled at BSP — back and lay, net of a commission rate you set per country. The greyhound archive is 7.6 million runner records since 2020; horse racing is 9 million+ across UK, Ireland, Australia and the USA. When a filter shows you a 4% return, that's 4% at a price you could have taken, after costs — not 4% at the best board price someone screenshotted once.