Measured 2026-08-11
We can’t tell you how to pick winners — nobody can, and we’ve the failed experiments to prove it. We can tell you how to stop giving away ten points on the horses you were going to back anyway.
We took every one of our settled funded bets that has a Betfair starting price on record — 210 of them, from 16 June to 8 August — and worked out what they returned depending only on where and how the bet was placed. Same horses, same days, same stakes. Nothing about the selections changed.
| How the bet settles | Return | vs taking the price | |
|---|---|---|---|
| Taking the price we quote | −0.6% | — | what most people do |
| Betfair SP, 5% commission | +9.0% | +9.5 pts | the standard rate |
| Betfair SP, 2% commission | +11.4% | +12.0 pts | with a discount plan |
| Betfair SP, no commission | +13.1% | +13.6 pts | not achievable — shown for scale |
210 bets is still a small sample and these are point estimates, not promises. Resampling our own record puts the gap somewhere between about 2 and 18 points nine times out of ten. The direction is solid; that much precision isn’t. Which is why the check below matters more than the table above.
Correction, 2026-08-11: an earlier version of this page put the exchange gap at 22 points. That came from the 101 bets with a bookmaker SP also on record, which turned out to be a drift-heavy, low-strike-rate slice of the book rather than a fair sample — it overstated the gap by well over double. The table above is the whole population.
A table built from 210 of our own bets is exactly the kind of thing that should make you suspicious, so we checked it against a record we had no hand in. Taking every market favourite in British and Irish racing over the nine months to June 2026 — 8,094 of them, no selection, no judgement, just whichever horse was shortest at 8am — and settling each one both ways:
| At the 8am price | −7.8% |
| At Betfair SP, 5% commission | +2.9% |
A gap of 10.7 points, on thirty-nine times the sample, from a different source, arriving at the same answer as our own record. This one is not a small-sample point estimate: it is roughly fourteen standard errors from zero, which is as close to settled as anything on this site gets.
Worth saying what it doesn’t show. Backing every favourite at Betfair SP returned +2.9% here, and that is nota system — it carries its own error bar of nearly two points and we’ve tested and failed to confirm effects far stronger-looking than that. The reliable part is the gap between the two lines, not the sign of the second one.
BOG means that if the horse’s starting price is bigger than the price you took, you get paid at the bigger one. You take 5/1, it goes off at 8/1, you’re paid at 8/1. It costs nothing and it applies automatically.
It matters because prices move, and often against you. Measured against the exchange close, 64% of our bets drifted between publication and the off; among the winners the closing price was bigger than the one we quoted 58% of the time, by an average of 11%. Without BOG, every penny of that is lost.
We’re not going to put a number on what BOG itself is worth, because we can’t do it honestly: bookmaker SP is only on record for half our bets, and it’s the drift-heavy half, so any figure we computed from it would flatter BOG. What we can say without measuring anything is that it’s a one-way option — it can only ever pay you more than the price you took, never less. There is no case for turning it down.
Not every firm offers it, and terms vary — it’s usually UK and Irish racing only, and often only from a set time in the morning. It’s worth checking before you open an account rather than after.
Betting exchanges charge commission on your net winnings instead of building a margin into every price. Lose a bet at Betfair SP and it costs you nothing extra; lose one at a bookmaker and you’ve already paid the margin on the way in. That structural difference is the whole of the gap in the table above.
Taking Betfair SP is a specific order type: open the market on the Exchange(not the Sportsbook), switch to the SP tab, enter a stake with no odds, and it matches at the off. You can place it the night before. Liquidity isn’t a constraint the way it is when taking a fixed price, because SP orders are pooled and reconciled together.
The trade-off is real and worth stating: taking SP means giving up the early price and any bookmaker concession that came with it, and it assumes you’d have been matched at the SP — fair for liquid UK and Irish win markets, less so for a large stake in a small one.
Betfair’s base rate is 5% of net market winnings. Discount plans exist that trade a monthly subscription for a lower rate, and on our record the difference is not small: moving from 5% to 2% is worth +2.5 points of return — roughly a quarter of the entire exchange advantage, for a setting you change once.
Whether it pays depends entirely on how much you stake. A plan that charges a flat monthly fee is a fixed cost against a saving that scales with turnover, so there’s a crossover: stake a few pounds a day and the plans that hand back free bets are ahead; stake meaningfully more than that and the low-commission plan wins. Below roughly £5 a day in stakes it is not close, and the answer is no. Work out your own daily turnover before subscribing to anything.
One detail that’s easy to miss: commission costs proportionally less at short prices, because short prices win small and often. On a fairly-priced bet at 5%, expect it to cost about 2.5% of your stake at even money, about 4% at 4/1, and about 4.5% at 10/1. It never reaches the headline 5%.
Staking systems that scale with your confidence only work if you have a measurable edge. Ours is statistically indistinguishable from zero — the honest formula returns a stake of nothing at all. So the goal isn’t to maximise growth; it’s to keep the bankroll alive.
Flat stakes of 1–2% of whatever you’ve set aside. The same stake after a winner as after a loser. Chasing a losing run is what turns an affordable hobby into a problem.
It doesn’t make you a winner. Our published selections score indistinguishably from the market’s own prices — we’ve tried twelve separate approaches to beat it and written up every failure on how we measure. Better execution means losing more slowly, not winning.
Anyone promising you a winning system is either mistaken or selling something. Our full record, wins and losses, is on results — nothing deleted.
