Our methodology

Each analysis starts from research into the fixture: recent and season-long form, home and away performance, goals scored and conceded, over/under and both-teams-to-score patterns, head-to-head history where it is relevant, competition context, and information about squads, call-ups, injuries, suspensions and coaching changes. That information is gathered from identified sources, listed on the page, and from structured fixture and statistics data.

The gathered information is then processed and organised into the fixed page structure described on the About page, and the analysis, market assessment and selection are written from it. Editorial standards and quality controls are applied before publication, including checks for missing or unsupported data, duplicated text and internal wording that does not belong on a public page. We do not claim that every page has been individually reviewed by a subject-matter expert, and we do not present the selection as a specialist forecast.

Reliable xG and xGA data may be considered when available, alongside goals, shots, corners, home and away performance, market odds, implied probability and comparisons between markets. No statistic automatically determines a pick. Sample size, quality of opposition, tactical matchup, schedule and verified injuries or suspensions are weighed according to their relevance to the fixture.

Head-to-head records and historical probabilities describe past observations; they are not forecasts with certainty. Small samples and uneven opponents can make an apparently strong percentage misleading.

Separate rule-based processes organise fixture data, scores, standings and the settlement of completed results. These operational processes do not change a published analysis, selection or price.

A prediction becomes part of the public record when its pre-match analysis, selection and available published odds are released. Once kickoff has occurred, that editorial record becomes historical: the original selection, price and analysis are not rewritten in light of the result.

Later fixture information may add the final score and a derived settlement state. Those additions describe what happened; they do not alter what was published beforehand.

Published odds are the decimal price recorded during preparation or publication and remain frozen in history. A later observed price, when available, is labelled separately and never replaces the published price. Odds may vary by bookmaker, region and time. Implied probability is a market-price comparison, not a guaranteed chance of success.

WIN, LOSS and PUSH are determined from the final score for supported markets. VOID is used only when explicitly supported by the stored record. Markets such as corners require their own factual market data; a final football score alone is insufficient.

When a fixture is not final, the record remains pending. When the fixture is final but a required fact is missing, or the market cannot be resolved safely, it remains unresolved or awaiting data rather than being converted into a loss.

A combined selection such as double chance plus a goals line is one published prediction. Its legs are settled together under the shared settlement rules, and the combination contributes only one observation to aggregate performance. Individual legs are not counted as separate picks.

The results summary reports the full published sample, historical records, settled outcomes, wins, losses, pushes or voids, pending fixtures and unresolved records. Win rate is calculated as wins divided by wins plus losses. Pushes, voids, half-results, pending fixtures and unresolved entries are excluded from that denominator, and the sample count is displayed alongside the percentage.

ROI, profit, yield and units won are not published because the archive does not contain a complete, auditable record of stakes. Missing metrics are omitted rather than estimated.

Factual claims should be checked against relevant, identifiable sources, with direct HTTPS links preferred over generic homepages. Source relevance is reviewed editorially; the presence of a URL alone is not treated as proof.

For 52 older analyses, 16 have sources that support the recorded rationale in full and 36 are marked partial because the available sources do not independently establish every part of the recorded pick. We do not invent or backfill sources, bookmaker names or capture times. New analyses list their sources, and newly quoted odds record where and when they were captured.

Football is uncertain, data can be incomplete, team news can change and every selection can lose. Historical percentages describe the recorded sample and are not forecasts or claims of profit. Review the transparent prediction results, our editorial policy and responsible gambling guidance.