Insights.
Go beyond the result. Follow the patterns. See the evidence.
Explore insights
Explore competitions →League insights
Premier League · 14 signals · 14 clubs in this view
Manchester UnitedChance balance ahead of the points returnInsight Score70/100
A positive observed xG balance sits alongside points below stored forecast expectations. These are different windows: inspect finishing, conceding and individual matches before assigning a cause.
- Points minus expected9/9 current-league matches · 2026-08-22 → 2026-10-18 UTC
Stored LOAD90 forecasts · same covered games - −5.8
- Observed xG difference / match10-match profile · exact dates unavailable
Observed match xG · available averages - +0.26
- Goals scored / matchLast 9 recorded matches · 2026-08-22 → 2026-10-18 UTC
Recorded scores · all competitions - 1.6
Sources, limitations & score
- Deviation
- 62.1
- Recency
- 64.1
- Context
- 100.0
- Quality
- 63.3
Uncalibrated selection heuristic. Not team strength, win probability or statistical significance. Sources keep their own windows; overlapping scorelines and forecast gaps receive reduced context credit.
- Points minus expected — Points 10; xPts 15.8. Saved pre: 3; reconstructed: 6. No claim that the gap must reverse. This window differs from the recorded-results and xG windows.
- Observed xG difference / match — Before next/latest league fixture, not necessarily current UTC. xG for 1.56; xG against 1.3. Coverage counts are unavailable and denominators can differ. Recency receives zero for this undated source; quality is capped at 40.
- Goals scored / match — Premier League 2026-2027: 1.45 goals per team/match, 70 matches · 2026-08-21 → 2026-10-19 UTC. Reference only: the club window spans all recorded competitions; no opponent adjustment.
FulhamChance balance ahead of the points returnInsight Score68/100
A positive observed xG balance sits alongside points below stored forecast expectations. These are different windows: inspect finishing, conceding and individual matches before assigning a cause.
- Points minus expected7/7 current-league matches · 2026-08-24 → 2026-10-17 UTC
Stored LOAD90 forecasts · same covered games - −5.1
- Observed xG difference / match10-match profile · exact dates unavailable
Observed match xG · available averages - +0.31
- Goals scored / matchLast 7 recorded matches · 2026-08-24 → 2026-10-17 UTC
Recorded scores · all competitions - 1.0
Sources, limitations & score
- Deviation
- 63.8
- Recency
- 62.0
- Context
- 100.0
- Quality
- 51.7
Uncalibrated selection heuristic. Not team strength, win probability or statistical significance. Sources keep their own windows; overlapping scorelines and forecast gaps receive reduced context credit.
- Points minus expected — Points 5; xPts 10.1. Saved pre: 2; reconstructed: 5. No claim that the gap must reverse. This window differs from the recorded-results and xG windows.
- Observed xG difference / match — Before next/latest league fixture, not necessarily current UTC. xG for 1.61; xG against 1.3. Coverage counts are unavailable and denominators can differ. Recency receives zero for this undated source; quality is capped at 40.
- Goals scored / match — Premier League 2026-2027: 1.45 goals per team/match, 70 matches · 2026-08-21 → 2026-10-19 UTC. Reference only: the club window spans all recorded competitions; no opponent adjustment.
BrightonWinning, and exceeding expectationsInsight Score68/100
A winning run and points above forecast expectations stand out together. Separate the recorded results from the chance balance; the points comparison covers a different set of league matches.
- Wins in a rowLast 7 recorded matches · 2026-08-23 → 2026-10-18 UTC
Recorded scores · all competitions - 4
- Points minus expected7/7 current-league matches · 2026-08-23 → 2026-10-18 UTC
Stored LOAD90 forecasts · same covered games - +4.6
- Goals scored / matchLast 7 recorded matches · 2026-08-23 → 2026-10-18 UTC
Recorded scores · all competitions - 3.3
- Observed xG difference / match10-match profile · exact dates unavailable
Observed match xG · available averages - +0.74
Sources, limitations & score
- Deviation
- 62.8
- Recency
- 64.1
- Context
- 100.0
- Quality
- 51.7
Uncalibrated selection heuristic. Not team strength, win probability or statistical significance. Sources keep their own windows; overlapping scorelines and forecast gaps receive reduced context credit.
- Wins in a row — A run of 10 is a lower bound. Related result sequences share one evidence family.
- Points minus expected — Points 16; xPts 11.4. Saved pre: 2; reconstructed: 5. No claim that the gap must reverse. This window differs from the recorded-results and xG windows.
- Goals scored / match — Premier League 2026-2027: 1.45 goals per team/match, 70 matches · 2026-08-21 → 2026-10-19 UTC. Reference only: the club window spans all recorded competitions; no opponent adjustment.
- Observed xG difference / match — Before next/latest league fixture, not necessarily current UTC. xG for 2.01; xG against 1.27. Coverage counts are unavailable and denominators can differ. Recency receives zero for this undated source; quality is capped at 40.
ArsenalWinning, and exceeding expectationsInsight Score67/100
A winning run and points above forecast expectations stand out together. Separate the recorded results from the chance balance; the points comparison covers a different set of league matches.
- Wins in a rowLast 9 recorded matches · 2026-08-21 → 2026-10-18 UTC
Recorded scores · all competitions - 3
- Points minus expected9/9 current-league matches · 2026-08-21 → 2026-10-18 UTC
Stored LOAD90 forecasts · same covered games - +5.0
- Goals scored / matchLast 9 recorded matches · 2026-08-21 → 2026-10-18 UTC
Recorded scores · all competitions - 1.7
- Observed xG difference / match10-match profile · exact dates unavailable
Observed match xG · available averages - +0.71
Sources, limitations & score
- Deviation
- 53.6
- Recency
- 64.1
- Context
- 100.0
- Quality
- 63.3
Uncalibrated selection heuristic. Not team strength, win probability or statistical significance. Sources keep their own windows; overlapping scorelines and forecast gaps receive reduced context credit.
- Wins in a row — A run of 10 is a lower bound. Related result sequences share one evidence family.
- Points minus expected — Points 24; xPts 19. Saved pre: 3; reconstructed: 6. No claim that the gap must reverse. This window differs from the recorded-results and xG windows.
- Goals scored / match — Premier League 2026-2027: 1.45 goals per team/match, 70 matches · 2026-08-21 → 2026-10-19 UTC. Reference only: the club window spans all recorded competitions; no opponent adjustment.
- Observed xG difference / match — Before next/latest league fixture, not necessarily current UTC. xG for 1.56; xG against 0.85. Coverage counts are unavailable and denominators can differ. Recency receives zero for this undated source; quality is capped at 40.
ChelseaPoints ahead. Chance balance tells another story.Insight Score52/100
Points exceed the stored forecasts, while the observed xG profile has no positive balance. Results and chances tell different stories; this does not establish that a reversal is due.
- Points minus expected7/7 current-league matches · 2026-08-24 → 2026-10-17 UTC
Stored LOAD90 forecasts · same covered games - +3.7
- Observed xG difference / match10-match profile · exact dates unavailable
Observed match xG · available averages - −0.30
Sources, limitations & score
- Deviation
- 46.3
- Recency
- 46.5
- Context
- 80.0
- Quality
- 42.5
Uncalibrated selection heuristic. Not team strength, win probability or statistical significance. Sources keep their own windows; overlapping scorelines and forecast gaps receive reduced context credit.
- Points minus expected — Points 13; xPts 9.3. Saved pre: 2; reconstructed: 5. No claim that the gap must reverse. This window differs from the recorded-results and xG windows.
- Observed xG difference / match — Before next/latest league fixture, not necessarily current UTC. xG for 1.2; xG against 1.5. Coverage counts are unavailable and denominators can differ. Recency receives zero for this undated source; quality is capped at 40.
Each signal belongs to one category. Categories sum to All signals; stories combine related evidence across categories.
Results vs forecasts14 signals · 14 clubs
Select a row to explore the evidence. Scroll to see all columns →
Manchester United
Points below expected
−5.8
9/9 covered matches · current leagues
Forecast66
The points return is below the stored match forecasts on the same covered games. This is a results-versus-expectations gap, not a guarantee of improvement.
Stored LOAD90 forecasts · same covered games · 9/9 current-league matches · 2026-08-22 → 2026-10-18 UTC. Points 10; xPts 15.8. Saved pre: 3; reconstructed: 6. No claim that the gap must reverse. This window differs from the recorded-results and xG windows.
- Deviation
- 62.1
- Recency
- 96.2
- Context
- 40.0
- Quality
- 60.0
Uncalibrated selection heuristic. Not team strength, win probability or statistical significance. Sources keep their own windows; overlapping scorelines and forecast gaps receive reduced context credit.
9/9 covered matches · current leagues
- Points
- 10.0
- xPts
- 15.8
Points and xPts cover identical matches. xPts uses stored LOAD90 forecasts, including historical reconstructions; not observed xG. Saved: 3; reconstructed: 6.
Full club analysis →
Arsenal
Points above expected
+5.0
9/9 covered matches · current leagues
Forecast63
Arsenal
+5.0
Points above expected
9/9 covered matches · current leagues
Points24.0xPts19.0
Results have delivered more points than the stored match forecasts expected. The gap describes those results; it does not establish that a reversal is due.
Stored LOAD90 forecasts · same covered games · 9/9 current-league matches · 2026-08-21 → 2026-10-18 UTC. Points 24; xPts 19. Saved pre: 3; reconstructed: 6. No claim that the gap must reverse. This window differs from the recorded-results and xG windows.
- Deviation
- 53.6
- Recency
- 96.2
- Context
- 40.0
- Quality
- 60.0
Uncalibrated selection heuristic. Not team strength, win probability or statistical significance. Sources keep their own windows; overlapping scorelines and forecast gaps receive reduced context credit.
9/9 covered matches · current leagues
- Points
- 24.0
- xPts
- 19.0
Points and xPts cover identical matches. xPts uses stored LOAD90 forecasts, including historical reconstructions; not observed xG. Saved: 3; reconstructed: 6.
Full club analysis →
Manchester City
Points above expected
+5.0
8/9 covered matches · current leagues
Forecast61
Results have delivered more points than the stored match forecasts expected. The gap describes those results; it does not establish that a reversal is due.
Stored LOAD90 forecasts · same covered games · 8/9 current-league matches · 2026-08-23 → 2026-10-17 UTC. Points 21; xPts 16. Saved pre: 3; reconstructed: 5. No claim that the gap must reverse. This window differs from the recorded-results and xG windows.
- Deviation
- 57.7
- Recency
- 93.1
- Context
- 40.0
- Quality
- 48.9
Uncalibrated selection heuristic. Not team strength, win probability or statistical significance. Sources keep their own windows; overlapping scorelines and forecast gaps receive reduced context credit.
8/9 covered matches · current leagues
- Points
- 21.0
- xPts
- 16.0
Points and xPts cover identical matches. xPts uses stored LOAD90 forecasts, including historical reconstructions; not observed xG. Saved: 3; reconstructed: 5.
Full club analysis →
Fulham
Points below expected
−5.1
7/7 covered matches · current leagues
Forecast63
The points return is below the stored match forecasts on the same covered games. This is a results-versus-expectations gap, not a guarantee of improvement.
Stored LOAD90 forecasts · same covered games · 7/7 current-league matches · 2026-08-24 → 2026-10-17 UTC. Points 5; xPts 10.1. Saved pre: 2; reconstructed: 5. No claim that the gap must reverse. This window differs from the recorded-results and xG windows.
- Deviation
- 63.8
- Recency
- 93.1
- Context
- 40.0
- Quality
- 45.0
Uncalibrated selection heuristic. Not team strength, win probability or statistical significance. Sources keep their own windows; overlapping scorelines and forecast gaps receive reduced context credit.
7/7 covered matches · current leagues
- Points
- 5.0
- xPts
- 10.1
Points and xPts cover identical matches. xPts uses stored LOAD90 forecasts, including historical reconstructions; not observed xG. Saved: 2; reconstructed: 5.
Full club analysis →
Brighton
Points above expected
+4.6
7/7 covered matches · current leagues
Forecast61
Results have delivered more points than the stored match forecasts expected. The gap describes those results; it does not establish that a reversal is due.
Stored LOAD90 forecasts · same covered games · 7/7 current-league matches · 2026-08-23 → 2026-10-18 UTC. Points 16; xPts 11.4. Saved pre: 2; reconstructed: 5. No claim that the gap must reverse. This window differs from the recorded-results and xG windows.
- Deviation
- 57.5
- Recency
- 96.2
- Context
- 40.0
- Quality
- 45.0
Uncalibrated selection heuristic. Not team strength, win probability or statistical significance. Sources keep their own windows; overlapping scorelines and forecast gaps receive reduced context credit.
7/7 covered matches · current leagues
- Points
- 16.0
- xPts
- 11.4
Points and xPts cover identical matches. xPts uses stored LOAD90 forecasts, including historical reconstructions; not observed xG. Saved: 2; reconstructed: 5.
Full club analysis →
Nottingham Forest
Points below expected
−4.6
7/7 covered matches · current leagues
Forecast61
The points return is below the stored match forecasts on the same covered games. This is a results-versus-expectations gap, not a guarantee of improvement.
Stored LOAD90 forecasts · same covered games · 7/7 current-league matches · 2026-08-22 → 2026-10-18 UTC. Points 5; xPts 9.6. Saved pre: 2; reconstructed: 5. No claim that the gap must reverse. This window differs from the recorded-results and xG windows.
- Deviation
- 57.5
- Recency
- 96.2
- Context
- 40.0
- Quality
- 45.0
Uncalibrated selection heuristic. Not team strength, win probability or statistical significance. Sources keep their own windows; overlapping scorelines and forecast gaps receive reduced context credit.
7/7 covered matches · current leagues
- Points
- 5.0
- xPts
- 9.6
Points and xPts cover identical matches. xPts uses stored LOAD90 forecasts, including historical reconstructions; not observed xG. Saved: 2; reconstructed: 5.
Full club analysis →
Chelsea
Points above expected
+3.7
7/7 covered matches · current leagues
Forecast56
Results have delivered more points than the stored match forecasts expected. The gap describes those results; it does not establish that a reversal is due.
Stored LOAD90 forecasts · same covered games · 7/7 current-league matches · 2026-08-24 → 2026-10-17 UTC. Points 13; xPts 9.3. Saved pre: 2; reconstructed: 5. No claim that the gap must reverse. This window differs from the recorded-results and xG windows.
- Deviation
- 46.3
- Recency
- 93.1
- Context
- 40.0
- Quality
- 45.0
Uncalibrated selection heuristic. Not team strength, win probability or statistical significance. Sources keep their own windows; overlapping scorelines and forecast gaps receive reduced context credit.
7/7 covered matches · current leagues
- Points
- 13.0
- xPts
- 9.3
Points and xPts cover identical matches. xPts uses stored LOAD90 forecasts, including historical reconstructions; not observed xG. Saved: 2; reconstructed: 5.
Full club analysis →
Hull City
Points above expected
+3.6
7/7 covered matches · current leagues
Forecast56
Hull City
+3.6
Points above expected
7/7 covered matches · current leagues
Points9.0xPts5.4
Results have delivered more points than the stored match forecasts expected. The gap describes those results; it does not establish that a reversal is due.
Stored LOAD90 forecasts · same covered games · 7/7 current-league matches · 2026-08-22 → 2026-10-17 UTC. Points 9; xPts 5.4. Saved pre: 2; reconstructed: 5. No claim that the gap must reverse. This window differs from the recorded-results and xG windows.
- Deviation
- 45.0
- Recency
- 93.1
- Context
- 40.0
- Quality
- 45.0
Uncalibrated selection heuristic. Not team strength, win probability or statistical significance. Sources keep their own windows; overlapping scorelines and forecast gaps receive reduced context credit.
7/7 covered matches · current leagues
- Points
- 9.0
- xPts
- 5.4
Points and xPts cover identical matches. xPts uses stored LOAD90 forecasts, including historical reconstructions; not observed xG. Saved: 2; reconstructed: 5.
Full club analysis →
Bournemouth
Points below expected
−2.6
9/9 covered matches · current leagues
Forecast54
The points return is below the stored match forecasts on the same covered games. This is a results-versus-expectations gap, not a guarantee of improvement.
Stored LOAD90 forecasts · same covered games · 9/9 current-league matches · 2026-08-23 → 2026-10-18 UTC. Points 12; xPts 14.6. Saved pre: 3; reconstructed: 6. No claim that the gap must reverse. This window differs from the recorded-results and xG windows.
- Deviation
- 27.9
- Recency
- 96.2
- Context
- 40.0
- Quality
- 60.0
Uncalibrated selection heuristic. Not team strength, win probability or statistical significance. Sources keep their own windows; overlapping scorelines and forecast gaps receive reduced context credit.
9/9 covered matches · current leagues
- Points
- 12.0
- xPts
- 14.6
Points and xPts cover identical matches. xPts uses stored LOAD90 forecasts, including historical reconstructions; not observed xG. Saved: 3; reconstructed: 6.
Full club analysis →
Liverpool
Points above expected
+3.1
9/9 covered matches · current leagues
Forecast55
Results have delivered more points than the stored match forecasts expected. The gap describes those results; it does not establish that a reversal is due.
Stored LOAD90 forecasts · same covered games · 9/9 current-league matches · 2026-08-23 → 2026-10-17 UTC. Points 18; xPts 14.9. Saved pre: 3; reconstructed: 6. No claim that the gap must reverse. This window differs from the recorded-results and xG windows.
- Deviation
- 33.2
- Recency
- 93.1
- Context
- 40.0
- Quality
- 60.0
Uncalibrated selection heuristic. Not team strength, win probability or statistical significance. Sources keep their own windows; overlapping scorelines and forecast gaps receive reduced context credit.
9/9 covered matches · current leagues
- Points
- 18.0
- xPts
- 14.9
Points and xPts cover identical matches. xPts uses stored LOAD90 forecasts, including historical reconstructions; not observed xG. Saved: 3; reconstructed: 6.
Full club analysis →
Ipswich
Points above expected
+2.9
7/7 covered matches · current leagues
Forecast53
Results have delivered more points than the stored match forecasts expected. The gap describes those results; it does not establish that a reversal is due.
Stored LOAD90 forecasts · same covered games · 7/7 current-league matches · 2026-08-22 → 2026-10-17 UTC. Points 9; xPts 6.1. Saved pre: 2; reconstructed: 5. No claim that the gap must reverse. This window differs from the recorded-results and xG windows.
- Deviation
- 36.3
- Recency
- 93.1
- Context
- 40.0
- Quality
- 45.0
Uncalibrated selection heuristic. Not team strength, win probability or statistical significance. Sources keep their own windows; overlapping scorelines and forecast gaps receive reduced context credit.
7/7 covered matches · current leagues
- Points
- 9.0
- xPts
- 6.1
Points and xPts cover identical matches. xPts uses stored LOAD90 forecasts, including historical reconstructions; not observed xG. Saved: 2; reconstructed: 5.
Full club analysis →
Tottenham
Points below expected
−2.5
7/7 covered matches · current leagues
Forecast53
Tottenham
−2.5
Points below expected
7/7 covered matches · current leagues
Points6.0xPts8.5
The points return is below the stored match forecasts on the same covered games. This is a results-versus-expectations gap, not a guarantee of improvement.
Stored LOAD90 forecasts · same covered games · 7/7 current-league matches · 2026-08-22 → 2026-10-19 UTC. Points 6; xPts 8.5. Saved pre: 2; reconstructed: 5. No claim that the gap must reverse. This window differs from the recorded-results and xG windows.
- Deviation
- 31.3
- Recency
- 99.5
- Context
- 40.0
- Quality
- 45.0
Uncalibrated selection heuristic. Not team strength, win probability or statistical significance. Sources keep their own windows; overlapping scorelines and forecast gaps receive reduced context credit.
7/7 covered matches · current leagues
- Points
- 6.0
- xPts
- 8.5
Points and xPts cover identical matches. xPts uses stored LOAD90 forecasts, including historical reconstructions; not observed xG. Saved: 2; reconstructed: 5.
Full club analysis →
Newcastle
Points below expected
−2.1
7/7 covered matches · current leagues
Forecast49
The points return is below the stored match forecasts on the same covered games. This is a results-versus-expectations gap, not a guarantee of improvement.
Stored LOAD90 forecasts · same covered games · 7/7 current-league matches · 2026-08-23 → 2026-10-17 UTC. Points 8; xPts 10.1. Saved pre: 2; reconstructed: 5. No claim that the gap must reverse. This window differs from the recorded-results and xG windows.
- Deviation
- 26.3
- Recency
- 93.1
- Context
- 40.0
- Quality
- 45.0
Uncalibrated selection heuristic. Not team strength, win probability or statistical significance. Sources keep their own windows; overlapping scorelines and forecast gaps receive reduced context credit.
7/7 covered matches · current leagues
- Points
- 8.0
- xPts
- 10.1
Points and xPts cover identical matches. xPts uses stored LOAD90 forecasts, including historical reconstructions; not observed xG. Saved: 2; reconstructed: 5.
Full club analysis →
Crystal Palace
Points below expected
−1.4
9/9 covered matches · current leagues
Forecast49
The points return is below the stored match forecasts on the same covered games. This is a results-versus-expectations gap, not a guarantee of improvement.
Stored LOAD90 forecasts · same covered games · 9/9 current-league matches · 2026-08-22 → 2026-10-18 UTC. Points 10; xPts 11.4. Saved pre: 3; reconstructed: 6. No claim that the gap must reverse. This window differs from the recorded-results and xG windows.
- Deviation
- 15.0
- Recency
- 96.2
- Context
- 40.0
- Quality
- 60.0
Uncalibrated selection heuristic. Not team strength, win probability or statistical significance. Sources keep their own windows; overlapping scorelines and forecast gaps receive reduced context credit.
9/9 covered matches · current leagues
- Points
- 10.0
- xPts
- 11.4
Points and xPts cover identical matches. xPts uses stored LOAD90 forecasts, including historical reconstructions; not observed xG. Saved: 3; reconstructed: 6.
Full club analysis →About these insights
Read the signal. Keep the context.
Insights describe what is recorded. Use club and match analysis to explore what comes next.
Insight Score · experimental editorial interest. This version uses 0.35D + 0.25R + 0.20C + 0.20Q. D measures heuristic deviation, R observation recency, C breadth of evidence families and Q data quality. These starting weights are uncalibrated. It is neither team strength, win probability, statistical significance nor forecast confidence. Related result signals share one family; different sources are not automatically statistically independent. Unknown dates receive zero recency; unknown xG coverage limits its quality component. League averages are context references, not opponent-adjusted estimates.
Check the window
Result signals use the last 10 recorded matches; away signals use the last 8 away matches. At least 5 are required. Missing and special results are excluded. A 10+ streak is a lower bound.
Read the chart
Bars run oldest to newest. Height shows goals scored or conceded, capped at 6; zero keeps a marker. Teal / amber / red means win / draw / loss. Faded bars mean the event did not happen. Open the evidence for full scores.
Separate the sources
xG uses observed Playing style averages before the next or latest league fixture; coverage can differ between inputs. xPts comes from stored LOAD90 forecasts on identical covered games, including labelled historical reconstructions.
Types alternate for variety. Values are ordered within their own metric, without a combined ranking. These descriptive signals do not change the LOAD90 forecast.
Explore methodology →