Altora Analytics
Five minutes on how this site works, so you can judge the record for yourself rather than take our word for anything.
We build statistical models for sports and markets, publish what they predict before the event, and keep every result on the record afterwards, wins and losses alike. We are not here to sell you an edge. We are here to try and prove one, out in the open.
Each row on a card shows the fixture, the kick-off time and league, the model's pick, the bookmaker odds at the time we logged it, and the model's probability. The probability is the product. "Model 64%" means: in the model's view this outcome happens 64 times in 100. It does not mean "this will win". A well-built model is wrong about a 64% pick more than a third of the time, on purpose.
These are model outputs, not tips. Whether a probability is worth acting on depends on the price, and most days the answer is no.
A prediction only becomes a tracked bet when it clears rules that were written down before the season: minimum probability or value thresholds per market, frozen in public. Those rules earned their place by passing an honest test: the strategy had to be profitable on both halves of the historical data independently, with at least thirty bets in each half. One lucky stretch is not evidence.
Everything that fails the gates is still published and still settled. We show the misses because a record with the misses removed is not a record.
When a match is cancelled, a player withdraws, or a prediction was published under a wrong setting on our side, the row is marked VOID and excluded from the accuracy numbers, but it stays visible. Corrections are listed on their own page with reasons. Nothing is deleted. An exclusion you cannot see is a deletion with better manners, so every exclusion is disclosed.
Closing-line value asks: after we published at the early price, did the market move towards our number by kick-off? Beating the closing line consistently is the strongest early evidence a model sees something real, long before profit curves mean anything. We record opening and closing prices for every tracked market and publish the comparison, including when it is against us.
Tracked bets are settled at a flat £10 stake so the arithmetic is checkable by anyone. Sports records are model tracking, not our bankroll. The trading side is different: three systematic books run live on demo accounts, verified against the broker's own history, and the weekly profit and loss is published every Monday whether it is green or red.
Every sport's page links a CSV of the complete history. The methodology page explains each model. The research log records what we tried, including the ideas that failed and were dropped. If you spot an error, we would genuinely like to know.
18+ only. This is analysis, not financial advice, and no outcome is guaranteed. If you bet, bet what you can afford to lose, and if it stops being fun, stop: BeGambleAware.org.