Win rate is the wrong number.

Every tipping advert leads with the same two figures: how many winners, and what percentage. Thousands of winners found. One in four picks wins. Half hit the places. All of it can be true, and all of it can be worthless, because the one number that decides whether you made money — the prices those winners were backed at — is the one number missing.

The benchmark nobody quotes

Here is what doing nothing clever achieves. We measured every UK and Irish race with a starting price and a result from October 2025 to July 2026 — 10,698 races — and backed the favourite in every one, on paper:

• · The favourite won 36.3% of races — ≈3,900 “winners found” in ten months.
• · Level stakes on every one returned −5.9%.
• · Over jumps the favourite wins even more often (37.7% against 35.4% on the flat) — and loses slightly more money (−6.1%).

So a 36% strike rate is free. Anyone can have it, today, with no model at all — and it loses about 6p in the pound, forever, because the bookmaker’s margin is baked into every price. A service quoting a win rate below that number is advertising that it underperforms picking the shortest price on the card. A service quoting one above it has told you nothing yet either: win rate rises as you pick shorter horses, mechanically, no skill involved. Tip only 1/2 shots and you will “win” two times in three, all the way to the poorhouse.

The winners-reel trick

The other half of the pitch is the reel: a scrolling list of past winners, each with a lovely price attached, filterable every way except the one that matters. A list of winners is not a record — it is the record with the losers deleted. Pick a horse in every race for a year and you are mathematically guaranteed thousands of winners to put in the reel, whatever your skill, because racing runs tens of thousands of races. The reel proves activity. It cannot prove profit, and it is built so you never ask.

The only questions that sort a record from an advert

• Can I see every pick, including the losers? Not a sample, not a reel — the complete settled list, with the bad months in it.
• Is there a price on every pick, and could I have taken it? A win counted at a price that was gone by the time anyone read the tip is a win nobody had.
• Is the money added up? Strike rate, place rate, winners-per-month — if the page quotes everything except return on stakes, the return on stakes is negative. No exceptions have ever been observed.
• Is there an honest error bar? A +20% return over 40 picks is a coin flip dressed up. Anyone quoting a return without a range either doesn’t know that or hopes you don’t.

Our answers, since you’ll ask

This site exists to be the version of this that survives those questions. Every settled pick is listed, losers first-class citizens, with the level-stake return and a 99% interval on any slice you care to look at. The prices are the board prices published at 8am, before the first race. The methodology — what counts, what doesn’t, and why — is written down on how we measure, including the findings that flatter us and the ones that don’t. And where our own record at a course can’t be distinguished from losing, we suspend our own picks there, in public, automatically. That is what a record looks like when it is kept for the reader instead of the advert.

Benchmark figures measured October 2025 to July 2026 across 10,698 UK and Irish races; favourite defined as the shortest starting price in the race. Last updated 2026-08-31.