Why we publish every losing bet

There is an easy trick in this game, and almost everyone uses it. Show the wins, bury the losses. Post the winning Saturday, go quiet on the four losing ones. Do that for long enough and even a losing method can look like a licence to print money.

We built the AI Lab to do the opposite. Every prediction the model makes is logged before the off and scored afterwards, win or lose. Every system a punter brings gets run over real British races, with fallers and pulled-up horses settled as the losers they are. Nothing gets quietly dropped because it had a bad month.

What that honesty buys you

Two things, mainly.

First, you can trust the numbers you see, because you are seeing all of them. When we say a system loses over time, that is the real record, not a tidied-up version of it. When the model has a rough week, it stays on the page for everyone to read.

Second, it makes the useful bits genuinely useful. The value in racing was never a magic selection. It is understanding how prices work, where the bookmaker's margin sits, and how often a given price actually wins. You can only learn that from an honest record, and an honest record has to include the losers.

So that is the deal here. No hidden slips, no cherry-picked Saturdays. Just the whole thing, week to week, and an open invitation to check our working. If you want to see where that record stands right now, our track record has every result in the open.