May 2026: +20.7 Units From a 36.7% Win Rate
In May our model fired 120 settled signals and won 44 of them. A 36.7% win rate — and the most profitable month we've had: +20.7 units at flat stakes. If that combination sounds wrong to you, this post is for you. It's also the full audit: the losing weeks, the −5 unit day, and exactly where the profit came from.
Why a 36.7% win rate made money
Most tipsters sell win rate because win rate is easy to sell. "80% winners!" means nothing if those winners came at odds of 1.20 — you can win 80% and bleed money. Profit lives in one place only: the gap between the price you take and the player's true chance.
At average odds of 3.17, break-even is a 31.5% win rate. We hit 36.7%. That five-point gap, repeated 120 times, is +20.7 units. The model isn't trying to predict winners — it's trying to find prices the market got wrong, and in May the market kept mispricing the same thing: underdogs on clay.
Where the profit actually came from
Split the month by price band and the story tells itself:
| Price band | Bets | Winners | P&L |
|---|---|---|---|
| Odds 3.00 and longer | 67 | 27 (40.3%) | +30.7u |
| Odds below 3.00 | 53 | 17 (32.1%) | −10.0u |
The entire month's profit — and then some — came from the longshots. The market is sharpest where the money is: favourites and short prices. It's laziest on the players nobody's watching. That's where a model crunching 160+ features per match over 1.6 million matches finds its edge.
The five biggest wins of the month:
- Ekaterine Gorgodze @ 4.50 — WTA Istanbul (5 May)
- Elina Svitolina @ 4.40 — Rome (14 May)
- Henri Squire @ 4.20 — Hamburg (16 May)
- Nikoloz Basilashvili @ 4.20 — Rome (9 May)
- Thanasi Kokkinakis @ 4.10 — Roland Garros (25 May)
The part most results posts leave out
A 36.7% win rate means losing is the normal state. Nearly two of every three bets lost. Here's what that felt like inside the month:
That 17–21 May stretch is the test every subscriber eventually faces. Five days, nothing landing, the record sliding. If you'd quit on 21 May you'd have missed the +10.2u run that started 48 hours later. Variance at 3.17 average odds is brutal in the short term and irrelevant in the long term — but only if the underlying edge is real.
How we know the edge is real (not luck)
One month of profit proves nothing — a coin can flip heads ten times. The metric that separates skill from luck is closing line value: did the market move toward our price after we bet? In May our picks beat the closing line roughly three times out of four, with an average CLV of +5 to +9%. Luck doesn't do that 90 times in 120 attempts. Mispriced openings do.
Every pick is published before the match and settled publicly — wins, losses and voids — on the live results page. Nothing is deleted. Nothing is edited after the fact. Audit it yourself.
Get June's picks as they fire
Every signal, the live dashboard, player profiles and the in-play engine — one plan, £20/month.
Start a 7-day free trial →FAQ
How can a 36.7% win rate be profitable?
Profit = win rate × odds. At average odds of 3.17, break-even is 31.5% — anything above that is margin. We won 36.7%. High win rates at short odds are how tipsters look good while losing you money; value at long odds is how records compound.
Are these results verified?
Every signal is published before the match and settled publicly at fault.bet/results. The page is regenerated automatically from the same database that drives the bets — there is no separate "marketing" copy of the record.
What stakes do these numbers assume?
Flat 1-unit stakes on every pick, so the record can't be flattered by staking tricks. Subscribers also get suggested stakes (fractional Kelly) with every signal.
What was the worst moment of the month?
27 May: five picks, five losses, −5.0 units. It's on the record like everything else. The month still finished +20.7u — that's what an edge plus discipline looks like.
18+ only. Gamble responsibly — BeGambleAware.org. Past performance does not guarantee future returns. Never bet more than you can afford to lose.