1,000+ tipsters tracked, 20,000+ picks settled across 10+ sources. What those published picks actually did, graded against the final result — a record appears here once it reaches 5 settled picks, because below that the numbers are noise. How this is measured · What the whole corpus returned
The bar next to each name is that record’s plausible range of true yield, not the number itself. Recent results count for the most, so the bar narrows while a record stays active and widens again once it goes quiet — a long history alone does not keep it tight. A bar that crosses break-even means the record is not yet proven, however large its estimate looks. A record with little evidence sits close to what we assume about an unproven stranger — a small loss to the bookmaker’s margin — until real results move it.
| # | Tipster | Source | Settled | Win rate | Estimated yield, as a range from -10% to +10% against break-even |
|---|---|---|---|---|---|
| 1 | Adibet | Adibet | 213 | 79.3% | Estimated yield +0.6%, likely range +0.0% to +1.2%. This clears break-even even at its most pessimistic reading. |
Ordered by sample size unless you rank by something else — a record with few settled picks tells you nothing, whichever direction it points. Ranking by a metric puts records without it last, never at zero. The interval blends realised returns with the market’s own closing-line verdict and narrows as evidence accumulates, so a new or thin record gets a wide, honest range rather than a confident-looking single number. Open a record for its full settled history.