Expected goals
Home xG / Away xG
Bundesliga - 28321
Current lifecycle fixture
Kickoff
2026-08-29 00:00:00
VS

Intelligence overview
Low confidence, strongest supported indicator: Over 1.5 goals.
Expected goals
Home xG / Away xG
Model
analysis-v1
live-local
Publication
published
Public view hides drafts
Confidence
Medium confidence
Review context
AI MATCH ANALYSIS
The model projects an away-lean for Hoffenheim (46% probability) versus Köln (27%), with a combined expected goals of 3.5. High probabilities for Over 1.5 (95%) and BTTS (84%) point to an open game; confidence is low (48) so the prediction should be treated as a statistical leaning rather than a robust forecast.
Prediction classification: away_lean. Hoffenheim carries the highest single outcome probability at 46%, with the model placing the match in an open, attacking category rather than a narrow defensive contest. The model’s low confidence score (48) and a 19-point gap between the highest and the next-best outcome mean the away-lean is meaningful but not definitive.
The statistical inputs indicate Hoffenheim has the stronger recent trend in the data used by this model, contributing to the away-lean. There is limited match-sample depth (one match used per side in the provenance), which reduces the robustness of form conclusions. Use the trend signal as directional rather than conclusive.
Expected goals are close between the sides but assign a slight away scoring edge. Köln’s expected home goals are 1.7 while Hoffenheim’s expected away goals are 1.8 — a near parity that points to both teams having scoring capacity. The model does not indicate a clear home defensive advantage.
The combined expected goals (3.5) and market probabilities point to an open game. The model strongly favours over 1.5 goals and sees a solid likelihood that both teams score.
Over 1.5
Over 1.5 goals probability is 95% — the strongest market signal from the model, indicating multiple goals expected in the match.
Over 2.5
Over 2.5 goals probability is 67% — a majority probability that the game will produce three or more goals, supporting an aggressive goals projection.
BTTS
Both teams to score probability is 84% — strong statistical support that both sides will find the net.
Expected goals
FC Köln: 1.7
TSG Hoffenheim: 1.8
TSG Hoffenheim
Slight away scoring edge
Expected away goals 1.8 gives Hoffenheim a marginal advantage in expected scoring output on the road.
FC Köln
Home attacking projection
Köln’s expected home goals of 1.7 indicate the home side is projected to contribute to a high-scoring contest rather than sit back.
Low sample size and limited match data
Provenance shows only one match used per side (matchesUsedHome = 1, matchesUsedAway = 1). This small sample weakens confidence in form and trend signals.
Model confidence and calibration
Overall confidence score is 48 (Low). The model’s known limitations include weaker draw prediction calibration and variable performance by league/season.
Narrow probabilistic margin
Top outcome (away win 46%) is only 19 percentage points above the next-best outcome grouping, so results could still plausibly fall to draw or home win.
Final Verdict
The model leans to an away win for Hoffenheim (46%) while signalling an open, goal-rich fixture (combined xG 3.5; Over 1.5 = 95%; BTTS = 84%). Limitations in sample depth (one match per side used) and a low confidence score mean this should be treated as a probabilistic inclination, not a definitive outcome. Key practical takeaways: expect both teams to contribute to the scoring and a higher-than-average goal count; the match outcome remains uncertain within the model’s low confidence band.
Confidence language: Low confidence (score 48) — statistical leaning rather than a firm prediction. This remains an analytical view, not a guaranteed selection.
Analysis Engine V1 calculates the probabilities. The AI writer explains the verified inputs and does not alter numbers, add odds, or claim certainty.
Source datasets: live-local. Generated 2026-07-14T16:38:51.296Z.
Exact model percentages, never bookmaker odds.
Model-estimated probability, not odds.
Model-estimated probability, not odds.
Model-estimated probability, not odds.
Model-estimated probability, not odds.
Model-estimated probability, not odds.
Model-estimated probability, not odds.
A probability is an estimated chance from the local model context. It is not an outcome promise and is not bookmaker odds.
Market intelligence
A separate explanation layer for match markets. These signals do not change Analysis Engine V1 and are not bookmaker odds.
V1 expects 3.5 total goals. Local team samples average 3.37 total goals in matches involving these sides.
The winner market remains separated from goals markets, so a high goal signal does not automatically mean a strong home or away call.
Sportmonks returned corner statistics in the access audit. Per-team corner trends are not available for this fixture yet.
Sportmonks returned yellow cards, red cards and ball-possession statistics in the access audit. Match-specific trend import is pending.
Risk review
Short caution signals that explain why the analysis remains measured. These do not change the V1 prediction.
Draw tendency sample
26%
Combined draw share across both team samples.
V1 draw estimate shown for transparency.
Confidence caution
Standard
Still not a guarantee or betting recommendation.
Manual publication remains required.
Historical sample context
68 matches
Current-season form will appear separately after Sportmonks current-season imports are expanded.
Longer-term profile from completed local datasets: 2025/2026. Current-season signals appear separately when Sportmonks match data is available.
FC Köln form
PPG 0.94 - GF 49 - GA 63
TSG Hoffenheim form
PPG 1.79 - GF 65 - GA 52
Home team signal
Points profile
0.94 PPG
7W 11D 16L sample
Goals for
1.44
49 scored across local sample
Goals against
1.85
63 conceded across local sample
Away team signal
Points profile
1.79 PPG
18W 7D 9L sample
Goals for
1.91
65 scored across local sample
Goals against
1.53
52 conceded across local sample
Match timeline
Goals, cards, VAR and substitutions appear only when verified Sportmonks event data exists locally.