How we pick matches, price them and grade the result

Who writes the previews

The match analysis and the headline verdict on every preview are produced by an AI model (Claude, made by Anthropic), working from the market prices on our odds board, live fixture and result data, and public reporting on form and team news. A human editor reads each preview before it is published, checks the fixture, kickoff and home-away order against live data, and decides which previews go live. The editor does not change a verdict after publication.

We say this plainly because most tipster sites do not, and because it changes how you should read the page. The model is good at pricing what is in the data and at saying so honestly. It does not have sources inside a dressing room, it does not watch training, and it is not a person with a hunch. Treat every verdict as a reasoned reading of the public record, nothing more.

How matches are chosen

Our odds board pulls consensus prices from The Odds API twice a day for the World Cup, Champions League, Premier League, La Liga, Serie A, Bundesliga and Ligue 1, and ranks fixtures by a heat score: how competitive the market is, how much attention the game carries, how soon it kicks off, and the league's weight. Previews are picked from the top of that board, with a preference for games where the market price and our reading of the match disagree. During international windows, when the board has no club fixtures, we use the same approach on the Nations League with prices from public comparison snapshots, and we say so in each preview.

We publish about two previews a week in season. We do not preview every game, and we do not publish a verdict on a game we could not price honestly.

Where the odds come from

The 1X2 prices on each preview are market-consensus figures at the time of writing, recorded with the date they were checked. The headline verdict is priced at the level we observed for that market on the same day. Prices move, and we do not update a verdict's price after publication, so the number you see is the number we are graded against, even when the market has since improved.

Neither the board nor the previews are a live odds service. Check the current price before making any decision.

How verdicts are graded

Only the headline verdict is graded. Alternatives, hedges and "if you prefer" angles in the body of a preview do not count, in either direction. The market's own settlement rules decide the result: a 2-2 draw wins both teams to score, a 2-1 win loses a -1.5 handicap, a draw refunds a draw-no-bet stake and is recorded as void. Results are checked against a named source, which is linked on the track record entry.

Every graded verdict carries a note. Losses get the most honest notes. When a bet wins for the wrong reason, the note says so, because a win with a wrong thesis tells you as much about our reading as a loss does.

Previews are never deleted. A losing preview stays published with its result banner, with the original analysis intact.

What the record measures

The track record shows strike rate and level-stakes profit at one unit per verdict, at the price quoted. It has graded 51 verdicts so far. That is a sample, not a system: short runs prove nothing in either direction, and we say so on the page. Judge it in hundreds.

What we are not

We do not sell tips, run a subscription or take money from bookmakers for placement. The site earns a commission when a reader opens a Skrill or NETELLER account through our links, which is disclosed on every page that carries one. That commission does not depend on what you bet or where, and it does not influence which matches we preview or what we say about them.

Questions about any of this go to the contact page.