Leipzig v Man City

The three cover every possible outcome.

Man City are the most likely winners, with a 55% chance. Leipzig have a 23% chance and a draw 22%. The probabilities favor City, but leave room for either side to take points.

The largest separation is European pedigree: City have the much stronger long-run continental rating, 1897 to Leipzig’s 1711. That points to a side with greater strength across recent European seasons, and it is the main reason the matchup tilts toward City. Leipzig’s recent European form has been less convincing, with losses outweighing their win across their last three outings.

City have been in better European form, winning more often across their last three matches, and their scoring form also supports the choice. Defensive solidity is another advantage for City, while their higher domestic league standing reinforces the comparison. Leipzig have conceded heavily in recent European matches, while City have scored more freely, though City have also let in five goals across that span.

Conclusion

Leipzig’s ability to tighten a defence that has conceded eight goals in its last three European matches could change the expected result. If they limit City’s scoring, their own recent goalscoring form gives them a chance to make the match more competitive.

This analysis is written by our models from the probabilities and factors published above. The numbers are the prediction, the prose only explains them.

Model context

Goal rates

1.20 - 1.94

European rating, Leipzig

1711

European rating, Man City

1897

First published

28 Oct 2026, 08:15

The model works by first estimating how often each side scores, then turning those scoring rates into the result probabilities above. The rates are shown so you can see which side is expected to create more. Read them as that comparison, not as a score prediction: the numbers sit lower than real goal counts.

The European rating is a dedicated ladder walked over Champions League and Europa League matches, qualifying included. It carries over between seasons rather than resetting, so the numbers are comparable across years. See the full season board →

Model: C_hybrid_2015 z_shrink=0.90 draw_coupling=0.03 train_seasons=2015-2026