Our public card is down £130. We're publishing that anyway — here's the maths.
Six weeks ago we put a virtual £1,000 on our homepage and promised to settle every pick our v1.0 model makes, in public, at the advised price — wins, losses, the lot. Today that pot reads roughly £870. The card is down 13 points over 129 settled picks.
Most tipster sites have a standard procedure for this moment: the widget disappears, the record restarts, and a new screenshot era begins. We built this site specifically to be the opposite of that. So instead, here is the drawdown, in full, with the maths that says what it does — and doesn't — mean.
The numbers, exactly
Since the ledger started on 20 May 2026:
- 129 picks settled at an advised price, one point level stakes each
- 20 winners — a 15.5% strike rate at an average advised price of 6.60
- P&L: −13.0 points (−£130 at £10 a point from the £1,000 virtual start)
- Average closing line value: +17.9%, and 69.8% of advised prices beat the closing price
- Every one of those picks, with its price and result, is on the performance ledger and the CLV audit
What −13 points actually means
Here is the uncomfortable statistical truth most tipping content skips: over 129 bets at an average price of 6.60, luck alone moves the result by about ±27 points one standard deviation in either direction. That's not a hedge — it's arithmetic. A strategy with exactly zero edge lands anywhere inside that band about two-thirds of the time, and outside it a third of the time.
Our −13 sits well inside it — about half a standard deviation from zero. In plain English: this result is statistically indistinguishable from breakeven. It is also indistinguishable from a small edge, and from a small negative edge. Six weeks of racing simply cannot tell those apart, no matter what any tipster's marketing implies. Anyone who claims certainty either way from a sample like this is selling something.
Two details in the data are worth an honest look:
The strike rate is almost exactly breakeven. At an average price of 6.60, the breakeven strike rate is 15.2%. The card is hitting 15.5%. The deficit isn't missing winners — it's that the winners so far have come in at slightly shorter prices than the card's average. That's a which-ones-won effect, and over small samples it's noise of the purest kind.
The market keeps agreeing with the picks after we publish them. Nearly seven in ten advised prices beat the close. If the model were picking genuinely bad horses, you'd expect the market to drift them, not back them. (Honest caveat, as always: our advised price is the best available bookmaker price and the close is nearer an industry average, so part of that CLV is price-shopping rather than prophecy — we've said so before and the trend matters more than the absolute number.)
None of that makes the card profitable. It isn't, right now, and we won't dress that up. It makes the card exactly what our own audit predicted when we corrected our backtest claims a month ago: a strategy that is approximately breakeven against the closing market, wobbling inside its noise band, in public.
Why we don't delete it — and the rule that binds us
The moment a record gets deleted because it's losing, every record that site ever publishes becomes worthless. Survivorship is the original sin of tipping: run ten cards, delete seven, advertise the best three. We wrote a whole essay about catching ourselves in subtler versions of that trap. Deleting the pot now would make it all a lie.
So here is the rule, stated publicly so we can't wriggle out of it later:
The v1.0 ledger runs for a full twelve months — to 19 May 2027 — and then we publish a complete autopsy, whatever it says. Down 100 points or up 100, the write-up happens. Before that date the public card changes for only two reasons: a successor model earns promotion by beating v1.0's live record over a meaningful sample (our ML shadow currently hasn't — it's running worse, at −21.3 points, and that's on the head-to-head page too), or we find and disclose a settlement bug. "It's losing" is not on the list.
What we'd actually tell you to do with this
Nothing about a breakeven-ish public experiment says "bet more". If you use Racing Alpha the way we think punters should, the daily card is the least of it: the draw bias maps, angle screens, fair-price checks on every card and price-timing base rates are tools for making your own betting sharper. The ledger's job is different — it's the proof that when we publish a number, we're still standing behind it when it turns red.
That's the whole brand. It would have been cheaper to delete the widget.
18+. Please gamble responsibly — BeGambleAware.org. Past performance — good or bad — does not predict future results, and this post is information, not betting advice.