Editorial Policy
Most of what Odds Uni publishes is produced by software, not typed by a person. This page says how that works, who is accountable for it, and what happens when it is wrong.
Two kinds of content
Model output — predictions, probabilities, confidence ratings and the statistics shown beside them. These are generated automatically from data, on a schedule, without a person approving each one. There are far too many fixtures a day for human sign-off on each to be honest, so we do not claim it.
Written pages — this page, the methodology, the legal pages, and anything explanatory. These are written and reviewed by the people who run Odds Uni.
We do not present one as the other. A model's reasoning is labelled as a model's reasoning.
Where the data comes from
Schedules, scores, team form, head-to-head records, standings and box scores come from third-party sports data providers. Market prices, where shown, are reference data from licensed bookmakers with whom we have no affiliation and through whom you cannot bet on this site.
Provider data can be late, incomplete or wrong, and sometimes is. Where a figure is not available for a fixture, the page says so rather than substituting an estimate. A blank is information; a fabricated number is not.
How a prediction is published — and then left alone
A day's predictions are written once, at publication, and read back unchanged from then on. They are never regenerated after the fact.
This matters more than it sounds. If a prediction were recomputed when you loaded the page after kick-off, it would quietly become a better prediction than the one we actually made, and every accuracy figure built on it would be fiction. The price shown against a pick is the price at publication, not a price fetched later.
The modelling itself — what goes in, how probability and confidence are derived — is set out in the Methodology.
How results are graded
Every published prediction is settled against the final score and counted, win or loss. Nothing is quietly dropped for being embarrassing, and there is no “void” category used to hide a miss — voids exist only where a fixture was abandoned or a market was genuinely not resolvable.
The aggregate, with the sample size behind every figure, is published at Results. It is generated from the graded record, not typed in.
Corrections
Two kinds happen, and they are handled differently.
A provider amends the underlying data. Box scores are revised after the final whistle more often than most people expect. When that changes a settlement, the affected predictions are re-graded and the published record moves with them. The record is meant to be right, not stable.
We got something wrong ourselves — a mis-stated statistic, a broken page, an error in a written explanation. Email support@oddsuni.com with the URL. Substantive changes to a written page are noted on the page.
We do not delete a prediction because it lost. A prediction that has been published is part of the record.
What we will not publish
- A guaranteed outcome, a “sure bet”, a lock, or any claim that a prediction cannot lose. Sports are uncertain and our own record says so.
- An accuracy figure that is not computed from the graded record.
- A statistic, injury, lineup, head-to-head record or weather detail we do not have data for. If the data is missing, the page says the data is missing.
- Financial, investment or betting advice. We publish analysis; the decision is yours.
Independence
Odds Uni is not a sportsbook and is not owned by one. Where advertising appears, it is labelled and it does not influence what the models produce or what the record says. No advertiser, operator or partner can buy a prediction, change a published result, or have a losing pick removed.
Who is responsible
The Odds Uni team is responsible for everything published here, including the automated output. “The model did it” is not a defence, and we do not use it as one. Reach us through Contact.