Most tipping pages show you today's picks and quietly forget yesterday's. This one keeps them. Every selection is written down before kick-off with the price that was actually available, then settled against the real result — wins, losses and voids alike — so the profit, ROI and strike rate at the top of this page are a record rather than a claim. Selections are chosen by the Statz model, priced against the bookmaker's margin-free probability, and staked flat at one point so the return measures the picks and nothing else. There is a hard cap of five bets a day, one per fixture, and no more than 40% of a day behind a single thesis — and on days when nothing clears the bar, the page publishes nothing at all. Covers match result, both teams to score and over/under goals across the Premier League, Champions League, La Liga, Serie A, Bundesliga and more.
It is a record of what was published, not a simulation. Every bet is written down before kick-off with the bookmaker price that was showing at that moment, and it is never re-priced afterwards even if the odds drift in our favour. Settlement follows the bookmaker's own rules, so a player prop where the tipped player did not start is a void rather than a loss.
The model produces a probability, that probability is corrected for the bias the market has actually shown, and it is then shrunk toward the bookmaker's margin-free price — because a large disagreement with the market is usually model error rather than value. What survives is ranked by expected return per unit staked at the price offered.
One point, flat, on every selection. Flat staking means the ROI on this page measures the quality of the selections alone — none of the return comes from sizing some bets bigger than others after the fact.
Because nothing cleared the bar. The card is capped at five bets a day, one per fixture, with no more than 40% of a day behind one underlying thesis — and a day with nothing worth backing publishes nothing. A page that always has a tip is a page that is padding.
Yes. Every settled selection stays in the bet log and in the monthly table, and losing months are shown exactly like winning ones. A record that only appears when it flatters us would not be a record.
Each bet is stamped with the model version that produced it, so performance can be read before and after a change rather than blurred into one average. Markets are judged only on bets the current model produced.
No. These are research outputs, not instructions. Even a genuinely profitable set of selections loses regularly, and any edge is only meaningful alongside your own staking plan and bankroll discipline.