There's a comforting theory in every betting shop: the favourite is over-bet, so the real value must be further down the card. Back a few big prices, the logic goes, and one good winner pays for all the losers.
We settled 179,202 UK and Irish runners between March 2025 and July 2026 at official starting price and sorted them by how big the price was. The pattern is one of the cleanest we've ever pulled — and it points the other way.
| Starting price | Runners | Won | Return |
|---|---|---|---|
| Odds-on | 3,420 | 58.6% | −7.1% |
| Evens – 2/1 | 8,052 | 36.7% | −10.0% |
| 2/1 – 4/1 | 22,287 | 23.2% | −11.6% |
| 4/1 – 8/1 | 38,259 | 13.1% | −16.1% |
| 8/1 – 16/1 | 39,435 | 6.8% | −21.7% |
| 16/1 – 33/1 | 32,091 | 3.3% | −28.5% |
| 33/1 and bigger | 35,658 | 1.0% | −49.9% |
Every step down the board makes the bet worse. Not slightly worse — dramatically worse. By the time you reach 33/1 and bigger, you are winning one race in a hundred and handing over half of every pound you stake.
It is one of the oldest and most reliable findings in betting markets. Outsiders are systematically over-backed relative to their real chance, so their prices are shorter than they should be. Short-priced horses are much closer to fair.
Why? Because a big price sells something a short price can't: a story. £10 on a 40/1 shot buys you five minutes of imagining £400. That daydream has a market value, and punters pay for it — roughly 50p in the pound, as it turns out. Bookmakers are not fools; they price the hope, not just the horse.
No — and this is the bit people miss. Every band on that table is negative. The least-bad bet is still a losing bet. Choosing where on the board to lose money more slowly is not a strategy.
Profit doesn't live in a price band. It lives in the gap between a price and a horse's real chance — which changes race by race and requires you to be right when the market is wrong. That's a much harder claim to make, which is why we publish every one of our model's picks before the race and settle them in public, losers included.
Every model pick logged pre-race with timestamped odds. Wins, losses, and the favourite baseline — side by side.
Do outsiders offer better value?
No. Returns get worse as prices lengthen — from −7.1% at odds-on to −49.9% at 33/1 and bigger.
How often do 33/1+ shots win?
1.0% of the time — one in a hundred, across 35,658 runners.
What's the least-bad price band?
Odds-on, at −7.1%. But every band loses; "least-bad" is not the same as good.
All 179,202 UK & IRE runners with an official starting price, 16 March 2025 – 12 July 2026, flat and jumps, non-runners excluded. Bands are by official SP (decimal): odds-on <2.0; evens–2/1 2.0–2.99; 2/1–4/1 3.0–4.99; 4/1–8/1 5.0–8.99; 8/1–16/1 9.0–16.99; 16/1–33/1 17.0–33.99; 33/1+ 34.0 and above. Returns are £1 level stakes at SP, win only, no each-way. Past performance does not predict future results.
Do favourites make money? — 18,925 races,
and no price band turns a profit.
Are odds-on favourites safe? — they win
nearly six times in ten and still lose.