Methodology

Tipzy publishes measured records for sports tipsters. This page describes exactly how those records are produced, what each number means, and where the method has limits. If a figure on this site is wrong, this page should be enough to work out why — and to tell us.

1. What we record

We read picks that tipsters have already published publicly on their own sites, and we grade them against the final result of the match.

We do not receive picks from the tipsters we track, we are not paid by them, and they cannot edit or remove what we have recorded. Nothing on this site is a recommendation to bet.

2. Nothing is published before it is settled

A tip we collect stays private until the match has finished and the tip has been graded. We never republish a tipster's live, unsettled selection.

This is a hard rule in the code, not an editorial preference: the query behind every public page filters to settled outcomes only, so an un-graded tip has no route to a public page. It exists so this site is a record of what happened, never a way to obtain someone else's tip without visiting them.

3. Where the data comes from

Every record names the site the pick was published on. We keep the tipster's own name and the site they published on as two separate facts, and we never merge them: if the same name appears on two sites, that is two records, because we cannot prove they are the same person.

Sources currently in the pipeline include Oddspedia, Forebet, PredictZ, ZuluBet, Vitibet, Adibet, ProSoccer, MightyTips, SoloPredict and Blogabet. Some of these are single forecasting algorithms run by the site itself; others host many individually named tipsters. Where a source gives each tipster a stable internal ID, we key the record on that ID rather than the display name, so a tipster who renames themselves keeps one continuous record instead of splitting into two.

Some sources feed our statistics only and are never republished as listings on Tipzy.

4. How a pick is settled

1. Once the match's scheduled finish has passed, the pick becomes eligible for settlement.
2. We fetch the final score from a third-party results provider (TheSportsDB).
3. The selection is graded against that score and assigned one of: Won, Lost, Half win (a split-stake outcome such as an Asian handicap quarter-line), or Void (no action — postponement, abandonment, or a push).
4. Grading is proposed automatically and can be corrected by a human reviewer. Where a market is ambiguous or the score cannot be resolved, the pick stays unsettled rather than being guessed — and therefore stays unpublished.

A pick we cannot match to a fixture, or whose result we cannot obtain, is never counted in any statistic. It is absent, not zero.

5. What the numbers mean

Win rate — settled picks won ÷ settled picks graded. Void picks are excluded from both sides, because no stake was at risk.

Profit — expressed in units, at level stakes of 1 unit unless the source publishes its own stake. A winning pick returns odds − 1; a loser costs 1; a void returns 0.

Yield — profit ÷ units staked, over the same settled set.

Sample size is published next to every figure, and it matters more than the figure. A 70% win rate over 10 picks is not evidence of anything. Betting records need hundreds of settled picks before they separate skill from variance, and most records on this site are not yet near that. Treat small samples as unproven, not as good or bad.

6. Odds: advertised versus implied

This distinction is the most important caveat on the site.

• Advertised — the tipster published an actual price they claim to have taken. Profit calculated from these is a price someone could have bet at.
• Implied — the source published only a probability, and we derived a price from it (1 ÷ probability). Several forecasting sites, including Forebet and Vitibet, work this way. An implied price is a model output. Nobody could have bet at it, it carries no bookmaker margin, and profit computed from it will look better than any achievable result.

We store which kind each price is and never average the two together. Where a record is built on implied prices, treat the profit and yield figures as a measure of forecast accuracy, not of betting returns.

7. Stakes

Most sources publish no stake at all, so we treat those picks as level 1-unit stakes. A minority (such as Blogabet) publish a real stake in their own units, which we store as given. Stake-weighted figures are computed only over picks that actually carry a stake — we never default a missing stake to 1 and then present the result as stake-weighted.

8. Known limits

• Matching. Picks are matched to fixtures by team names and kickoff time. Name variants ("Man Utd" / "Manchester United") are normalised, but a wrong match is possible. A pick matched to the wrong fixture is graded against the wrong result.
• Collection timing. We capture what was published at the time we read the page. If a tipster edits or deletes a pick afterwards, our record reflects the earlier version and may differ from their site.
• Coverage. We do not necessarily capture every pick a tipster publishes. A record here is a record of what we observed, not a guaranteed complete history, and a tipster's own site may show a different total.
• Third-party results. Final scores come from an external provider. Where it is wrong, we are wrong.
• Markets we cannot yet grade. We store each source's own wording verbatim and normalise it into market families. Markets we have not mapped are left ungraded and excluded rather than approximated.

9. Corrections

If you are a tipster and a record here is inaccurate — a pick you did not publish, a wrong grade, a wrong price, a mis-matched fixture, or a record merged with someone else's — contact us and tell us which entry and what is wrong. We will check it against what we captured, correct anything we got wrong, and say so.

10. What we do not do

• We do not sell picks, and we do not take payment from tipsters for a listing, a ranking, or a better one.
• We do not republish anyone's unsettled selections, and we do not reproduce another site's written analysis as our own.
• We do not present estimated, illustrative, or placeholder figures as measured ones. Every number attached to a named tipster on this site is computed from settled picks we recorded, or it is not shown.

18+. Gamble responsibly. Past results do not predict future results.