Favourite–longshot bias at Betfair BSP: does backing short prices still lose?

It is the oldest documented pattern in betting markets: punters overpay for longshots and underpay for favourites. Decades of bookmaker data confirm it. The interesting question for exchange bettors is whether it survives once the built-in margin is gone and you are settling at Betfair BSP, net of commission.

What the favourite–longshot bias is

Across horse racing, greyhounds and most head-to-head sports, if you sort every runner by price and measure how backers actually did:

• Short prices have historically lost the least — sometimes broken even or profited — relative to what the odds implied.
• Long prices have lost heavily. The bigger the price, the worse the return per pound staked.

The usual explanations: bettors will pay a premium for the small chance of a big win; bookmakers shade their biggest prices hardest because that's where the liability and the mug money are; and a lot of people simply don't want to back odds-on.

Why the exchange sharpens the question

Bookmaker studies are muddied by the overround — the margin baked into every book. At Betfair BSP that margin is close to zero: the price is the crowd's collective opinion with almost no house edge on top. So if a favourite–longshot tilt still shows up at BSP, it is a property of how the crowd prices, not of a bookmaker's margin. And you have to clear commission (typically 2–5%) before any of it is yours.

The shape you tend to see at BSP

On the figures: the numbers below are illustrative — they show the shape of the pattern. Run the exact query for your sport, years and market in the System Builder before acting on it.

The lay side mirrors it: blindly laying big prices at BSP tends to make money before commission and staking costs, then gives much of it back after. So the bias is real and directional, but it is a tilt, not a licence to print.

Why it isn't a system on its own

• Backing every favourite still loses. "Least bad" is not "profitable" — commission and variance eat the small edge.
• It varies by context. Field size, race type, sport, win vs place market, and country all change the gradient. Greyhounds, Flat handicaps and Group races don't behave identically.
• Laying longshots has a fat tail. One 40.0 winner wipes out a long run of small profits. Small samples lie badly here.

Where it's actually useful

1. As a prior. If your system's qualifiers are concentrated at 25.0+, be sceptical — the base rate is against you and you need a much bigger sample to trust it.
2. As a filter. Capping the price range of a back system (say, exclude everything above 12.0) often steadies the P&L for very little lost profit.
3. As a lay context. Laying specific, well-defined longshots — not all of them — in races where the favourite is strong can be a coherent angle, if the sample holds up out of time.

How to check it for yourself

In the System Builder, filter by BSP rank or a BSP range and read ROI by band, then split by sport, field size and year. Do it for the win market and the place market separately. Look at the longest losing run on any lay-the-longshot version before you believe the headline number.

The short version

• The favourite–longshot bias survives at BSP: long prices return worse per pound than short ones.
• It is a tilt in the crowd's pricing, not a bookmaker artefact — and commission still has to be cleared.
• Backing all favourites loses; laying all longshots is a coin-flip with a fat tail.
• Use it as a prior and a price-range filter, not as a standalone system.