The greyhound "giant-killer" effect, tested

A tennis analytics study went round recently: back a player who has just upset one of the sport's big names, and you lose money — the "giant-killer" tends to bounce, worn out or overhyped, and fades in their very next match. It's a good story. We wanted to know if the same pattern shows up in greyhound racing, where the "giant" isn't a person but a market price. So we ran the numbers on 85,000 Betfair BSP runners.

The tennis version, briefly

The original study looked at players who'd just beaten one of the sport's elite, then tracked how those "giant-killers" performed in their following match. Backing them lost money at scale; fading them showed a positive return. The proposed explanation was straightforward — physical and mental fatigue, a letdown after the biggest win of the year, extra media weight going into the next round.

Building the greyhound equivalent

Greyhound racing doesn't have "giants" by name, but the market ranks every runner by BSP before the race, and that rank is a fair stand-in: a dog priced rank 4 or worse in a race is one the market expected to lose. So a "giant-killer" here is any dog that won last time out despite being ranked 4th or lower in its own market — a result the crowd didn't see coming. The question: does backing that dog again, next time out, at whatever price the market now sets, show the same fade the tennis study found?

What we found

Across 84,869 qualifying runs (GB, 2020–2026), backing greyhound giant-killers next time out at BSP returned −0.64% ROI, at an average price of 12.4. On its own that looks like a small loser. The number that matters is what it's compared against: runners in that same 9–17 BSP band, with no upset-win condition at all, return −0.90% ROI across 628,961 runners. The giant-killers aren't losing money faster than the market around them — if anything, marginally slower.

Before trusting that, we did what we'd tell you to do with any of your own filters: split it out of time. Runners from 2020–2023 and 2023–2026 both land in the same place — nothing shifted, nothing decayed, nothing was a one-period fluke.

Does it get worse the bigger the upset?

Splitting by exactly how big a surprise the win was tells a noisier story:

Rank 6 shows a positive number, and it's tempting to read that as "the biggest upsets are where the edge is." We'd flag that as exactly the kind of single good-looking cell we've written about before: the smallest sample of the three, at the widest average price, where a handful of long-priced results can swing the total. Split it out of time the way we did with the headline number, and it stops looking clean. We're not backing it, and we wouldn't build a filter on it without a lot more scrutiny than fits in one blog post.

What this actually tells you

No bounce, no fade, no edge either way. The greyhound market prices last week's upset winner about as accurately as it prices everything else at that BSP level — which, if you've read our piece on reading ROI properly, is itself the more interesting finding. A market that doesn't overreact to a single surprise result is a market that's doing its job. That's a less exciting headline than "greyhound giant-killers are a myth waiting to be busted," but it's the honest one, and it's exactly the kind of question worth checking before you build a system around an assumption instead of a number.