DFB-Pokal Predictions — Model Probabilities

Research, not instructions — the model states a probability, never a selection.
MatchKickoff (UK)Model 1X2Most likely score
SG Sonnenhof Großaspach crestSG Sonnenhof Großaspach v AugsburgAugsburg crest Tue 27 Oct, 17:00 — —
Hamburger SV crestHamburger SV v Eintracht FrankfurtEintracht Frankfurt crest Tue 27 Oct, 17:00 — —
VfL Bochum crestVfL Bochum v FSV Mainz 05FSV Mainz 05 crest Tue 27 Oct, 17:00 — —
Rot-Weiss Essen crestRot-Weiss Essen v RB LeipzigRB Leipzig crest Tue 27 Oct, 17:00 — —
Borussia Dortmund crestBorussia Dortmund v Werder BremenWerder Bremen crest Tue 27 Oct, 19:45 — —
Borussia Monchengladbach crestBorussia Monchengladbach v Karlsruher SCKarlsruher SC crest Tue 27 Oct, 19:45 — —
Elversberg crestElversberg v SV Darmstadt 98SV Darmstadt 98 crest Tue 27 Oct, 19:45 — —
1. FC Köln crest1. FC Köln v Union BerlinUnion Berlin crest Tue 27 Oct, 19:45 — —
SC Freiburg crestSC Freiburg v VfL WolfsburgVfL Wolfsburg crest Wed 28 Oct, 17:00 — —
SC Paderborn crestSC Paderborn v VfB StuttgartVfB Stuttgart crest Wed 28 Oct, 17:00 — —
Dynamo Dresden crestDynamo Dresden v FC Schalke 04FC Schalke 04 crest Wed 28 Oct, 17:00 — —
SSV Jeddeloh crestSSV Jeddeloh v Hertha BerlinHertha Berlin crest Wed 28 Oct, 17:00 — —
Holstein Kiel crestHolstein Kiel v TSG HoffenheimTSG Hoffenheim crest Wed 28 Oct, 19:45 — —
Hannover 96 crestHannover 96 v 1. FC Kaiserslautern1. FC Kaiserslautern crest Wed 28 Oct, 19:45 — —
Bayer Leverkusen crestBayer Leverkusen v 1. FC Nürnberg1. FC Nürnberg crest Wed 28 Oct, 19:45 — —
Bayern Munich crestBayern Munich v 1. FC Magdeburg1. FC Magdeburg crest Wed 28 Oct, 19:45 — —
Read the full picture. Every row links to the fixture's prediction view: expected goals, fair odds, price movement and the grounded team-news brief. The track record grades every published call in the open.

Frequently asked questions

How are these dfb-pokal predictions made?
A Dixon-Coles + ELO statistical model, trained continuously on a growing results archive, produces expected goals per team and win/draw/win probabilities for every fixture. The model is then priced against the real odds market — the market always owns the final number.
Are these betting tips?
No — they are model probabilities published as research. Probabilities carry uncertainty and no outcome is certain; judge the process on the public track record, which grades every published call against the closing line.
How do I know the model is any good?
The track record page publishes calibration (predicted vs observed frequencies), closing-line value on every call, and a held-out walk-forward backtest — process metrics first, never performance claims.
Prices captured in the scheduled market snapshot, 26 October at 08:07 UK time. · engine slate assembled 16 h ago · this page fetched the feed just now. Every price shown is a real captured bookmaker price with its own capture time.