What is Betfair Starting Price?
The price the exchange settles at when the race starts. On the same races, it came out accurate and the bookmakers' price did not.
The plain version
That matters more than it sounds, and this page is the evidence for it. But first, what the thing actually is.
A bookmaker sets a price. An exchange does not — it matches people who want to back a horse against people who want to lay it, the way a stock market matches buyers and sellers. Betfair Starting Price, universally shortened to BSP, is the price that process settles on at the moment the race goes off.
So BSP is not an opinion about the horse. It is the number a large number of people, many of them betting serious money, could not push any further in either direction before the stalls opened.
Every price has a margin built into it
Turn every price in a race into a percentage chance and add them up. If the prices were a clean forecast, they would total 100%. They never do. The excess is the margin — the bit that is there to make money for whoever set the prices.
On the 1,997 races we compared, the two markets are barely recognisable as the same activity:
| Market | Probabilities add up to | Margin |
|---|---|---|
| Betfair Starting Price | 100.2% | 0.2% |
| Bookmakers' Starting Price | 119.2% | 19.2% |
That is the single most useful thing on this page. Before anyone has picked a horse, before any skill or judgement enters the argument at all, one of those markets is asking you to overcome nineteen percent and the other is asking you to overcome almost nothing.
Two honest qualifications. The exchange takes a commission on winnings, which the margin figure above does not include, so the real gap is narrower than 19% to 0.2%. And an exchange only works if somebody is there to take the other side — on small races late at night, they sometimes are not.
Is the exchange price actually right?
A low margin is not the same as an accurate price. So we tested the harder question: when a market says a horse has a 25% chance, does that horse win a quarter of the time?
The statistic for this is a calibration slope. If a market is perfectly calibrated its slope is 1.000. Below 1 it is not spreading its opinions far enough apart; above 1 it is too confident — its short prices are too short.
Both prices were measured on the same 1,997 races and the same 18,529 runners. Not two studies compared afterwards. The same races, side by side.
| Market | Slope | 95% confidence interval | Verdict |
|---|---|---|---|
| Betfair Starting Price | 0.974 | 0.925 to 1.023 | Contains 1.000 — not distinguishable from perfect |
| Bookmakers' Starting Price | 1.095 | 1.041 to 1.148 | Excludes 1.000 — measurably off |
The confidence interval is the part that carries the weight. The exchange interval straddles 1.000, which is the statistical way of saying we looked hard for an error and could not find one. The bookmaker interval sits entirely above 1.000, and being above 1 has a specific meaning: the short prices are shorter than the horses deserve. The favourite, in other words.
We are quoting slopes rather than a headline percentage deliberately. To compare two markets you first have to strip the margin out of both, and there is more than one defensible way to do that. The choice barely moves the slopes, but it does move any single percentage figure — so a percentage would be partly a fact about our method. Our own reviewers flagged this, and quoting the slope is how we settled it.
Why this is the only benchmark we use
Every number published on this site is measured against BSP, for three reasons.
It is the closest thing to a price you could really have taken. Testing an idea against a bookmaker's early morning price proves nothing if that price would have vanished the moment you asked for it.
It is recorded by someone else. BSP is set by an exchange with no interest in whether our research looks good, and it cannot be quietly revised after the result is known.
It is a hard opponent. This is the real reason. Beating chance is easy — the horses a rating likes are usually the horses everyone likes, and the market has already priced them. A rule that looks brilliant against random selection and adds nothing against BSP is the most common result in this entire field, and it is the result we have had over and over.
What this does not mean
It does not mean the exchange price is unbeatable, and it does not mean we beat it. We have not. Nothing here has cleared our validation bar.
It does not mean the exchange price is right about any individual race. A calibration slope is a statement about a market being right on average across thousands of races, and a market can be well calibrated overall while being badly wrong about the race in front of you. Finding out whether that kind of error is findable in advance is the open question we are actually working on.
And this was measured on a defined sample of 1,997 races, not on all of racing. Exchange prices in our archive only go back a short way, which is a limit we record rather than paper over.
Elite Pass has no validated betting edge and does not claim one.
Measured against the real price
Gold is the daily read for races where the evidence converges, and an honest No Read everywhere else — settled against BSP, in public, whichever way it goes.
Join Gold early access
Be first in when Gold Race Intelligence opens. No daily spam, no tipster noise.