New: the review after every match

An analysis does not end at kick-off. After the match, the question is: was the model right, and if not, why not? For that, from today the match page has a Review tab after the final whistle, for everyone, with no sign-in needed.

What the review shows

• The opportunity: three bars: our model, the market (Pinnacle without margin, at the time of the analysis) and the value we recalculate after the final whistle from the match's xG. Plus a sentence such as “would have won in 4 out of 10 such matches”.
• xG before and after: the expected goals of the final analysis next to the xG that the match actually produced.
• Influences: what spoke for the opportunity, what spoke against it and what was neutral.
• How the match went: Match facts, Finishing and goalkeepers, Momentum.

No AI writes the introductory sentence. It is built from fixed building blocks, in all languages.

Won does not mean well calculated

Bayer 04 Leverkusen – SC Freiburg, October 18. The final analysis put the home win at odds of 1.63 at 62%, the market at 59%. Leverkusen won 2:1, and the opportunity came in. Recalculated from the match's xG, however, the home win was only at about 45%. Freiburg came closer than the score suggests.

The opportunity in the review: our model, the market and the value after the final whistle.

The other way round: NEC Nijmegen – Fortuna Sittard on the same day. An opportunity on the home win at odds of 1.76, and NEC reached 3.20 to 0.44 xG. Recalculated, that would have won in 9 out of 10 such matches. Sittard drew level in added time, at 90+3, final score 2:2. So a loss does not automatically mean that the assessment was poor.

For whom and when

• Everyone can see the review, even without signing in, and it does not count against your quota.
• It is available for every finished match with a final analysis since October 1, even if there was no opportunity.
• It appears as soon as it has been calculated, usually just under two hours after kick-off.
• At first it is marked as “provisional”, because the match data is often still corrected during the night. We normally check it the day after; if reliable data is still missing then, it stays provisional a little longer. Once checked, it stays as it is.

A single match proves nothing, either for the model or against it. The review shows you how an opportunity came about and how the match fitted with it. How the opportunities perform over many matches is shown in the performance record.