← Back to ProofPitchProofPitchTerms of Use Privacy Policy Cookie Policy

About ProofPitch & How It Works

Last updated: 9 October 2026

ProofPitch is a football analytics service. It uses statistical and AI-assisted modelling to help you explore matches, understand probability estimates and review forecasts against their outcomes. It is available as a web app (PWA) at app.proofpitch.online and as a Telegram Mini App with the bot @proofpitch_bot.

Our forecasts are optional analytical recommendations. You choose whether and how to use them. We do not accept wagers, hold betting balances or settle bets.

How the analysis works

The model estimates team strength from historical match information, gives more weight to recent matches, estimates expected goals and derives probabilities for possible outcomes. It uses historical calibration to assess how reliable its estimates have been and compares them with bookmaker market information where available.

The current beta does not include confirmed line-ups or injury information in its model inputs. Missing information, data errors and unexpected events can affect an assessment. The method and the inputs may change as the service develops.

Understanding ratings

Strong and Good mark forecasts that the model considers worth examining more closely. Watching marks forecasts with no edge over the price yet, lower confidence or unresolved concerns, including limited data or disagreement with the market. A rating is not a promise, a command or a prediction of guaranteed success.

Understanding probability

A probability estimate describes uncertainty. An estimated 70% chance does not make an outcome certain, and it does not promise exactly seven successes in the next ten matches. Judge estimates across a large, comparable sample, not a single result or a selected streak.

An open record

Our approach is to record forecasts before kick-off and to make their history and results available for review. When you inspect a result, check the original forecast, its recording time, the forecast type, the sample size and the period. A record identifier or hash helps to check a record, but it is not by itself independent proof of when the record was created.

Two kinds of results are shown separately. Retrospective forecasts were calculated after the matches had been played, using the model as it is now. Forecasts published before the match were recorded before kick-off. Historical performance does not establish future accuracy, and a selected subset of results does not describe the full record.

Commercial relationships

The service contains partner links, and a partner may compensate us for qualifying referrals. Partner links do not influence forecasts, ratings or the model. See the Affiliate & Advertising Disclosure. Paid plans are not available at the moment; if we introduce them, they will give access to the described service, not to a particular match outcome or financial return.

Questions: support [at] proofpitch [dot] online.

18+. Football analytics involves uncertainty. Betting involves a risk of loss.