UEFA Champions League·Tue 20 Oct, 12:00 PDT

Manchester CityvAEK Athens FC

Manchester City

83%

Draw

11%

AEK Athens FC

Got%OnIt gives Manchester City a 83% chance of winning. That is roughly 4 chances in 5.

Estimate available
Evidence: tested on past results
Usable Kalshi price

Worth knowing: Injury information unavailable, so absences are unknown (not zero).

Deep Dive Every figure on this page is Deep Dive's: Core plus availability and schedule-load adjustments. Compare the modes.

Why

What the estimate is built from, and what argues against it.

Used by the model

Deep Dive starts from Core's estimate (past matches from 8 competitions (Premier League, La Liga, Bundesliga, Serie A, Ligue 1, Champions League, Europa League, Conference League): the goals each side scored and conceded over the last three seasons, with recent matches counting more, and the home advantage seen in this competition, plus the quality of the chances created (expected goals) where it was recorded) and adjusts the expected goals for reported absences, the confirmed line-up and goalkeeper where known, and each side's schedule load, each adjustment with a stated uncertainty band. When nothing defensible applies it equals Core.
Matches behind the estimate: Manchester City 151, AEK Athens FC 17 (across 8 competitions).
On these inputs the model expects about 3.1 goals for Manchester City and 0.8 for AEK Athens FC.
Manchester City won 6 of the 6 most recent matches the model used, scoring 2.7 and conceding 0.7 a game.
AEK Athens FC won 4 of the 6 most recent matches the model used, scoring 1.8 and conceding 1.3 a game.

What argues against it

Injury information unavailable, so absences are unknown (not zero).
Deep Dive's availability and schedule adjustments are stated priors with bands, not yet validated on settled matches; the band beside each figure shows how far they could move it.
The model tends to give draws slightly too little chance.

Not used by the model · context

Manchester City are 8th of 36 after 1 games; AEK Athens FC are 16th.
Manchester City have won 1 and lost 0; AEK Athens FC have won 1 and lost 0.

Recent form descriptive · from the matches the model used

Manchester City

8W 1D 1L in the last 10 (2.50 points a match) · goals 23–8 · xG 18.7–13.2 (10 of 10 with xG) · 7 wins in a row

Last 5: 5W 0D 0L, goals 12–3, xG 9.7–7.2.

This season's 7 matches carry 12 % of this side's weight in the fit; the last 5 carry 8 %; effective sample ≈ 110 of 151 matches.

AEK Athens FC

7W 1D 2L in the last 10 (2.20 points a match) · goals 21–10 · xG 23.6–11.1 (10 of 10 with xG) · 2 wins in a row

Last 5: 3W 0D 2L, goals 8–6, xG 9.3–6.3.

This season's 1 match carry 12 % of this side's weight in the fit; the last 5 carry 48 %; effective sample ≈ 13 of 17 matches.

Core applies no separate form or momentum term: these matches shape its strength estimate only through the weights above (half-life 365 days, refit at every snapshot). Deep Dive adds availability and schedule-load adjustments on top of that estimate, not a form adjustment.

Recent matches the model used

Manchester City

W3–0at Liverpool11 Oct
W5–3v Sunderland20 Sept
W1–0at Manchester United13 Sept
W2–0at FC Porto8 Sept
W1–0v Coventry5 Sept
W4–1at Crystal Palace28 Aug

AEK Athens FC

W1–0v Lask Linz8 Sept
W3–1v Rayo Vallecano16 Apr
L0–3at Rayo Vallecano9 Apr
L0–2v Celje19 Mar
W4–0at Celje12 Mar
W3–2v Universitatea Craiova18 Dec

It does not use league position, form tables or any market price. The availability and schedule facts it used are listed below with their labels; the rest is context for you to weigh.

What Got%OnIt knewprediction made Mon 12 Oct, 09:35 PDT

The frozen prediction was made Mon 12 Oct, 09:35 PDT (evidence cutoff Mon 12 Oct, 09:35 PDT; snapshot 989, soccer-deepdive-v1). Each item below is dated by when Got%OnIt observed it. Only the evidence set is a model input; everything else is context, whenever it arrived.

Used by the modelevidence set used by the modelSun 11 Oct, 13:55 PDT7044 finished UEFA Champions League matches (6964 with xG); latest kicked off 2026-10-11T19:00:00+00:00
Available at prediction timefixture scheduleMon 12 Oct, 09:30 PDTkick-off 2026-10-20T19:00:00+00:00, status NS
Current information — not included in frozen predictionleague tableMon 12 Oct, 21:55 PDTreceived
Current information — not included in frozen predictionseason figures (Manchester City)Mon 12 Oct, 21:55 PDTreceived
Current information — not included in frozen predictionseason figures (AEK Athens FC)Mon 12 Oct, 21:55 PDTreceived
Current information — not included in frozen predictionKalshi quotes on this pageMon 12 Oct, 22:58 PDTlatest observation of the quotes shown
Observation time unknowninjury reporttime unknownnot collected for this match
Observation time unknownstarting line-upstime unknownnot collected for this match

Scoring evidence10 + 7 matches observed · model estimate

Observed — results as they happened

Manchester City

Manchester City: 10 finished matches (10 of them in the model's evidence set).
Scored 1.6 and conceded 1.4 a game; scored in 80%, kept a clean sheet in 30%.
Both teams scored in 50%; over 1.5 goals in 100%, over 2.5 in 60%.

W2–0at FC Porto8 Sept · in model evidence
L1–2v Real Madrid17 Mar · in model evidence
L0–3at Real Madrid11 Mar · in model evidence
W2–0v Galatasaray28 Jan · in model evidence
L1–3at Bodo/Glimt20 Jan · in model evidence
W2–1at Real Madrid10 Dec · in model evidence

AEK Athens FC

AEK Athens FC: 7 finished matches (1 of them in the model's evidence set).
Scored 1.4 and conceded 0.9 a game; scored in 71%, kept a clean sheet in 43%.
Both teams scored in 43%; over 1.5 goals in 57%, over 2.5 in 57%.

W1–0v Lask Linz8 Sept · in model evidence
W4–0v Levski Sofia26 Aug · not in model evidence
D0–0at Levski Sofia18 Aug · not in model evidence
L1–2v Antwerp30 Aug · not in model evidence
L0–1at Antwerp22 Aug · not in model evidence
D2–2v Dinamo Zagreb19 Aug · not in model evidence

observed results as the provider reported them, counted by Got%OnIt; frequencies from a sample this small are descriptive, not a probability; a row 'in the model's evidence' is one of the finished matches the frozen estimate was computed from

Model estimate — frozen Mon 12 Oct, 09:35 PDT

Total goals

0 goals	2%
1 goal	7%
2 goals	15%
3 goals	20%
4 goals	20%
5 goals	15%
6 or more	21%
Over 1.5	91%
Over 2.5	75%
Over 3.5	55%
Both teams score	53%

Manchester City

Expected goals	3.13
Scores at least once	96%
Scores 2 or more	82%
Clean sheet	45%

AEK Athens FC

Expected goals	0.81
Scores at least once	55%
Scores 2 or more	19%
Clean sheet	4%

model-estimated probabilities frozen at the prediction instant (independent Poisson score grid); not observed frequencies and not recomputed since

Pregame comparison

What each source said before kick-off, on one scale: the chance of each result in percent: Got%OnIt's estimate, the bookmaker estimate and the Kalshi reference price. Agreement between them is not proof of accuracy.

Outcome	Got%OnIt	Books	Kalshi
Manchester City	83%	—	84%
Draw	11%	—	10%
AEK Athens FC	6%	—	7%

No bookmaker prices collected for this match.Kalshi prices seen 33 min ago.Percentages are rounded; market medians need not total exactly 100%.

Got%OnIt and the markets broadly agree on this match.

Kalshi market choices3 Kalshi contracts · 115 modelled

Every modelled market first (exact figure, simulation and whether Kalshi lists it), then every Kalshi contract matched to this match, with both sides of each: open a contract to see what YES and NO win on, the price of that side and whether Build can use it. Got%OnIt shows an estimate only where it has one; “Limited evidence” marks a market type that has not been historically evaluated. The Kalshi figure beside each question is the stored midpoint; the YES and NO prices are the resting asks a buyer of that side would pay, before fees.

Every modelled market 115 from the grid · 6 on Kalshi · 52 simulated

Deep Dive · frozen snapshot (Core + adjustments) The exact grid figure is the headline for every grid-expressible market; the simulation column is its Monte Carlo estimate with one standard error. Every market here settles on the 90-minute score (plus stoppage time): extra time and penalty shoot-outs never count. No market including extra time is modelled or listed; the registry's result rule for every competition is regulation time.

Match result

Manchester City to win82.6%82.6% ± 0.1yes · YES ask 84¢

Draw11.2%11.2% ± 0.1yes · YES ask 13¢

AEK Athens FC to win6.2%6.2% ± 0.1yes · YES ask 9¢

Double chance

Manchester City or draw limited evidence93.8%93.8% ± 0.1yes · NO ask 95¢

AEK Athens FC or draw limited evidence17.4%17.4% ± 0.1yes · NO ask 17¢

Not a draw limited evidence88.8%88.8% ± 0.1yes · NO ask 94¢

Draw no bet

Manchester City to win, stake returned on a draw limited evidence93.0%93.0% ± 0.1no

AEK Athens FC to win, stake returned on a draw limited evidence7.0%7.0% ± 0.1no

Total goals

Over 0.5 goals in total limited evidence97.9%98.0% ± 0.0no

Under 0.5 goals in total limited evidence2.1%2.0% ± 0.0no

Over 1.5 goals in total limited evidence90.5%90.5% ± 0.1no

Under 1.5 goals in total limited evidence9.5%9.5% ± 0.1no

Over 2.5 goals in total75.3%75.4% ± 0.1no

Under 2.5 goals in total24.7%24.6% ± 0.1no

Over 3.5 goals in total limited evidence55.5%55.7% ± 0.2no

Under 3.5 goals in total limited evidence44.5%44.3% ± 0.2no

Over 4.5 goals in total limited evidence35.9%36.1% ± 0.2no

Under 4.5 goals in total limited evidence64.1%63.9% ± 0.2no

Over 5.5 goals in total limited evidence20.5%20.6% ± 0.1no

Under 5.5 goals in total limited evidence79.5%79.4% ± 0.1no

Both teams to score

Both teams to score limited evidence53.1%53.0% ± 0.2no

At least one team fails to score limited evidence46.9%47.0% ± 0.2no

Team goals

Manchester City over 0.5 goals limited evidence95.6%95.7% ± 0.1no

Manchester City under 0.5 goals limited evidence4.4%4.3% ± 0.1no

Manchester City over 1.5 goals limited evidence82.0%81.9% ± 0.1no

Manchester City under 1.5 goals limited evidence18.0%18.1% ± 0.1no

Manchester City over 2.5 goals limited evidence60.6%60.9% ± 0.2no

Manchester City under 2.5 goals limited evidence39.4%39.1% ± 0.2no

Manchester City over 3.5 goals limited evidence38.3%38.4% ± 0.2no

Manchester City under 3.5 goals limited evidence61.7%61.6% ± 0.2no

AEK Athens FC over 0.5 goals limited evidence55.4%55.3% ± 0.2no

AEK Athens FC under 0.5 goals limited evidence44.6%44.7% ± 0.2no

AEK Athens FC over 1.5 goals limited evidence19.4%19.3% ± 0.1no

AEK Athens FC under 1.5 goals limited evidence80.6%80.7% ± 0.1no

AEK Athens FC over 2.5 goals limited evidence4.8%4.8% ± 0.1no

AEK Athens FC under 2.5 goals limited evidence95.2%95.2% ± 0.1no

AEK Athens FC over 3.5 goals limited evidence0.9%0.9% ± 0.0no

AEK Athens FC under 3.5 goals limited evidence99.1%99.1% ± 0.0no

Winning margin

Manchester City to win by 3 or more limited evidence44.3%44.5% ± 0.2no

Manchester City to win by 2 or more limited evidence64.9%65.0% ± 0.2no

Manchester City to win by 1 or more limited evidence82.6%82.6% ± 0.1no

Manchester City not to lose by 1 or more limited evidence93.8%93.8% ± 0.1no

Manchester City not to lose by 2 or more limited evidence98.3%98.3% ± 0.0no

Manchester City not to lose by 3 or more limited evidence—99.6% ± 0.0no

AEK Athens FC to win by 3 or more limited evidence—0.4% ± 0.0no

AEK Athens FC to win by 2 or more limited evidence1.7%1.7% ± 0.0no

AEK Athens FC to win by 1 or more limited evidence6.2%6.2% ± 0.1no

AEK Athens FC not to lose by 1 or more limited evidence17.4%17.4% ± 0.1no

AEK Athens FC not to lose by 2 or more limited evidence35.1%35.0% ± 0.2no

AEK Athens FC not to lose by 3 or more limited evidence55.7%55.5% ± 0.2no

Clean sheet

Manchester City clean sheet grid44.6%44.7% ± 0.2no

AEK Athens FC clean sheet grid4.4%4.3% ± 0.1no

Correct scoreexact · no Kalshi correct-score contract listed yet (Kalshi lists them on match day)

Manchester City goals down, AEK Athens FC goals across. every figure is a cell, or a sum of cells, of the frozen score grid of the selected mode's snapshot; the "6+" row and column sum the grid beyond five goals; the simulation column is the run's own count of the same cells (run 2416, 100,000 draws).

Manchester City \ AEK Athens FC	0	1	2	3	4	5	6+
0	2.1%	1.4%	0.6%	0.2%	0.0%	0.0%	0.0%
1	6.0%	5.0%	2.0%	0.5%	0.1%	0.0%	0.0%
2	9.5%	7.7%	3.1%	0.8%	0.2%	0.0%	0.0%
3	10.0%	8.0%	3.2%	0.9%	0.2%	0.0%	0.0%
4	7.8%	6.3%	2.5%	0.7%	0.1%	0.0%	0.0%
5	4.9%	4.0%	1.6%	0.4%	0.1%	0.0%	0.0%
6+	4.4%	3.5%	1.4%	0.4%	0.1%	0.0%	0.0%

Manchester City 3–0 AEK Athens FC10.0%10.0% ± 0.1no

Manchester City 2–0 AEK Athens FC9.5%9.4% ± 0.1no

Manchester City 3–1 AEK Athens FC8.0%8.2% ± 0.1no

Manchester City 4–0 AEK Athens FC7.8%7.8% ± 0.1no

Manchester City 2–1 AEK Athens FC7.7%7.6% ± 0.1no

Manchester City exact goalsexact · not listed on Kalshi

Manchester City 0 goals4.4%4.3% ± 0.1no

Manchester City 1 goal13.6%13.7% ± 0.1no

Manchester City 2 goals21.4%21.1% ± 0.1no

Manchester City 3 goals22.3%22.5% ± 0.1no

Manchester City 4 goals17.5%17.4% ± 0.1no

Manchester City 5 goals11.0%11.1% ± 0.1no

Manchester City 6+ goals9.8%9.8% ± 0.1no

AEK Athens FC exact goalsexact · not listed on Kalshi

AEK Athens FC 0 goals44.6%44.7% ± 0.2no

AEK Athens FC 1 goal36.0%35.9% ± 0.2no

AEK Athens FC 2 goals14.5%14.5% ± 0.1no

AEK Athens FC 3 goals3.9%3.9% ± 0.1no

AEK Athens FC 4 goals0.8%0.8% ± 0.0no

AEK Athens FC 5 goals0.1%0.1% ± 0.0no

AEK Athens FC 6+ goals0.0%0.0% ± 0.0no

Winning margin (exact goal difference)exact · Kalshi lists half-line spreads only

AEK Athens FC by 6 or more0.0%0.0% ± 0.0no

AEK Athens FC by exactly 50.0%0.0% ± 0.0no

AEK Athens FC by exactly 40.1%0.1% ± 0.0no

AEK Athens FC by exactly 30.3%0.4% ± 0.0no

AEK Athens FC by exactly 21.4%1.3% ± 0.0no

AEK Athens FC by exactly 14.5%4.5% ± 0.1no

Draw (goal difference 0)11.2%11.2% ± 0.1no

Manchester City by exactly 117.7%17.6% ± 0.1no

Manchester City by exactly 220.6%20.5% ± 0.1no

Manchester City by exactly 318.1%18.2% ± 0.1no

Manchester City by exactly 412.7%12.7% ± 0.1no

Manchester City by exactly 57.4%7.4% ± 0.1no

Manchester City by 6 or more6.0%6.1% ± 0.1no

Mass outside the stored grid: 0.0420 %.

Manchester City v AEK Athens FC: the exact column is the frozen figure Build and the Lab use where the market is admitted; “grid” marks a figure read from the frozen grid for a market outside this competition’s evaluated scope. On Kalshi means a classified contract exists for this fixture, with the side (YES on the question or NO on its mirror) and its resting ask before fees; it is not a statement of value.

Match winner

AEK Athens FC to win (90 minutes)

Estimate available · Evidence: tested on past results · Current Kalshi quote

YES on AEK Athens FC to WinAEK Athens FC wins in 90 minutes

6%9¢

Conservative chance below Build's default 50% floor · 127 contracts at the ask · quote 33 min old

The NO side of this contract is a double chance, listed below.

Manchester City to win (90 minutes)

83%

84%

Estimate available · Evidence: tested on past results · Current Kalshi quote

YES on Manchester City to WinManchester City wins in 90 minutes

83%84¢

Eligible for Build under current settings · 936 contracts at the ask · quote 33 min old

The NO side of this contract is a double chance, listed below.

Draw (90 minutes)

11%

10%

Estimate available · Evidence: tested on past results · Current Kalshi quote

YES on DrawThe match is drawn after 90 minutes

11%13¢

Conservative chance below Build's default 50% floor · 100 contracts at the ask · quote 33 min old

The NO side of this contract is a double chance, listed below.

Double chanceNO on a match-result contract

NO on DrawManchester City or AEK Athens FC wins: one of the two teams must win · Not a draw

89%94¢

Limited evidence — in the Champions League only the match result is evaluated; not in default Build, included when a limited-evidence group is ticked under Markets to consider · 110 contracts at the ask · quote 33 min old

NO on AEK Athens FC to WinManchester City wins or the match is drawn · Home or draw

94%95¢

Limited evidence — in the Champions League only the match result is evaluated; not in default Build, included when a limited-evidence group is ticked under Markets to consider · 589 contracts at the ask · quote 33 min old

NO on Manchester City to WinAEK Athens FC wins or the match is drawn · Away or draw

17%17¢

Limited evidence · Conservative chance below Build's default 50% floor · 90 contracts at the ask · quote 33 min old

A double chance wins whenever the named result does not happen; its figure is one minus Got%OnIt’s frozen estimate for that result. It inherits the match-result evidence (including the known tendency to under-predict draws). Default Build admits it where that inherited calibration passed the M3B rule — Premier League and La Liga (all three), Bundesliga except NO on Draw — and never in the Champions League; each row states its own status.

Available prediction types: 11 the app can estimate · 30 listed on Kalshi but not estimated

Both teams to score — supported: exact from the 90-minute score grid (Dixon-Coles corrected Poisson); limited calibration evidence (no numeric allowance) · this type can appear in Lab and Build; each contract's own estimate, evidence and price are checked separately · in this competition only the 90-minute match result has been evaluated (Phase 8A: over/under 2.5 not supported); this family is priced from the same score grid as limited evidence — in Build only when limited-evidence groups are ticked, never High Confidence
Clean sheet — supported: exact from the 90-minute score grid (Dixon-Coles corrected Poisson) (= opponent under 0.5); no Kalshi contract
Correct score (exact 90-minute score) — supported: exact from the 90-minute score grid (Dixon-Coles corrected Poisson) (one cell per score, the full grid on the match page); Kalshi correct-score contracts are fetched and classified (listed on match day); limited calibration evidence (no numeric allowance) · this type can appear in Lab and Build; each contract's own estimate, evidence and price are checked separately · in this competition only the 90-minute match result has been evaluated (Phase 8A: over/under 2.5 not supported); this family is priced from the same score grid as limited evidence — in Build only when limited-evidence groups are ticked, never High Confidence
Double chance — supported: exact from the 90-minute score grid (Dixon-Coles corrected Poisson); traded on Kalshi as the NO side of the match-result contracts · this type can appear in Lab and Build; each contract's own estimate, evidence and price are checked separately · in this competition only the 90-minute match result has been evaluated (Phase 8A: over/under 2.5 not supported); this family is priced from the same score grid as limited evidence — in Build only when limited-evidence groups are ticked, never High Confidence
Draw no bet — supported: exact from the 90-minute score grid (Dixon-Coles corrected Poisson) (conditional on no draw; void on a draw); no Kalshi contract
Exact goal difference / winning margin — supported: exact from the 90-minute score grid (Dixon-Coles corrected Poisson) (shown on the match page); no Kalshi contract beyond the half-line spreads
Exact team goals — supported: exact from the 90-minute score grid (Dixon-Coles corrected Poisson) (marginal of one axis, shown on the match page); no Kalshi contract
Match result (home / draw / away) and the NO sides (double chances) — supported: exact from the 90-minute score grid (Dixon-Coles corrected Poisson) · this type can appear in Lab and Build; each contract's own estimate, evidence and price are checked separately · YES on the match result is evaluated in this competition; the NO sides (double chances) are limited evidence
Team goals over / under (half lines 0.5–3.5) — supported: exact from the 90-minute score grid (Dixon-Coles corrected Poisson); Kalshi lists team-total contracts on match day only (verified 2026-10-12), so they appear in Lab and Build late · this type can appear in Lab and Build; each contract's own estimate, evidence and price are checked separately · in this competition only the 90-minute match result has been evaluated (Phase 8A: over/under 2.5 not supported); this family is priced from the same score grid as limited evidence — in Build only when limited-evidence groups are ticked, never High Confidence
Total goals over / under (half lines 0.5–5.5) — supported: exact from the 90-minute score grid (Dixon-Coles corrected Poisson) · this type can appear in Lab and Build; each contract's own estimate, evidence and price are checked separately · in this competition only the 90-minute match result has been evaluated (Phase 8A: over/under 2.5 not supported); this family is priced from the same score grid as limited evidence — in Build only when limited-evidence groups are ticked, never High Confidence
Winning margin / handicap (half lines) — supported: exact from the 90-minute score grid (Dixon-Coles corrected Poisson) · this type can appear in Lab and Build; each contract's own estimate, evidence and price are checked separately · in this competition only the 90-minute match result has been evaluated (Phase 8A: over/under 2.5 not supported); this family is priced from the same score grid as limited evidence — in Build only when limited-evidence groups are ticked, never High Confidence
Anytime goalscorer / two or more goals — supported by simulation only: player allocation of simulated team goals (expected minutes, scoring share, availability); shown with Monte Carlo standard errors; Kalshi lists no player contract today (series fetched; classified only once their rules are verified on live contracts)

Listed on Kalshi; not estimated by Got%OnIt (shown so nothing is silently substituted):

First goalscorer — no event-order model: the simulation allocates goals per match, not in time order · needs an evaluated player model (the 100,000-draw player allocation is an experimental overlay with no chronological evaluation; Kalshi's player contracts also settle including any extra time)
First team to score — no event-order model · needs a model of when or how goals happen (halves, first goal) or of non-goal events (corners, shots), with its own data and a leak-free chronological evaluation
First-half both teams to score — no half-time model · needs a model of when or how goals happen (halves, first goal) or of non-goal events (corners, shots), with its own data and a leak-free chronological evaluation
First-half correct score — no half-time model · needs a model of when or how goals happen (halves, first goal) or of non-goal events (corners, shots), with its own data and a leak-free chronological evaluation
First-half result — no half-time model: the engines estimate 90-minute goals only · needs a model of when or how goals happen (halves, first goal) or of non-goal events (corners, shots), with its own data and a leak-free chronological evaluation
First-half spread — no half-time model · needs a model of when or how goals happen (halves, first goal) or of non-goal events (corners, shots), with its own data and a leak-free chronological evaluation
First-half total goals — no half-time model · needs a model of when or how goals happen (halves, first goal) or of non-goal events (corners, shots), with its own data and a leak-free chronological evaluation
League of the tournament winner — no tournament model · needs a season / tournament simulator over every remaining fixture (two-legged ties also need extra time and penalties)
Method of finish — no extra-time or penalty model · needs settlement or data insufficient for a defensible probability
Method of victory / advancing — no extra-time or penalty model; the result rule is regulation time · needs settlement or data insufficient for a defensible probability
Other — contract semantics not read · needs settlement or data insufficient for a defensible probability
Player assists — assists are not modelled · needs an evaluated player model (the 100,000-draw player allocation is an experimental overlay with no chronological evaluation; Kalshi's player contracts also settle including any extra time)
Player shots — no shot model (per-player shot data is not stored) · needs an evaluated player model (the 100,000-draw player allocation is an experimental overlay with no chronological evaluation; Kalshi's player contracts also settle including any extra time)
Player shots on target — no shot model · needs an evaluated player model (the 100,000-draw player allocation is an experimental overlay with no chronological evaluation; Kalshi's player contracts also settle including any extra time)
Player to score or assist — assists are not modelled · needs an evaluated player model (the 100,000-draw player allocation is an experimental overlay with no chronological evaluation; Kalshi's player contracts also settle including any extra time)
Qualification — no tournament model · needs a season / tournament simulator over every remaining fixture (two-legged ties also need extra time and penalties)
Round qualification — no tournament model · needs a season / tournament simulator over every remaining fixture (two-legged ties also need extra time and penalties)
Season leader — no season simulation · needs a season / tournament simulator over every remaining fixture (two-legged ties also need extra time and penalties)
Season top finish — no season simulation · needs a season / tournament simulator over every remaining fixture (two-legged ties also need extra time and penalties)
Second-half both teams to score — no half-by-half model · needs a model of when or how goals happen (halves, first goal) or of non-goal events (corners, shots), with its own data and a leak-free chronological evaluation
Second-half result — no half-by-half model · needs a model of when or how goals happen (halves, first goal) or of non-goal events (corners, shots), with its own data and a leak-free chronological evaluation
Second-half spread — no half-by-half model · needs a model of when or how goals happen (halves, first goal) or of non-goal events (corners, shots), with its own data and a leak-free chronological evaluation
Second-half total goals — no half-by-half model · needs a model of when or how goals happen (halves, first goal) or of non-goal events (corners, shots), with its own data and a leak-free chronological evaluation
Team corners — no corners model · needs a model of when or how goals happen (halves, first goal) or of non-goal events (corners, shots), with its own data and a leak-free chronological evaluation
Team shots — no shot model · needs a model of when or how goals happen (halves, first goal) or of non-goal events (corners, shots), with its own data and a leak-free chronological evaluation
Team shots on target — no shot model · needs a model of when or how goals happen (halves, first goal) or of non-goal events (corners, shots), with its own data and a leak-free chronological evaluation
To advance (knockout tie) — no two-leg / extra-time / penalty model · needs a season / tournament simulator over every remaining fixture (two-legged ties also need extra time and penalties)
Total corners — no corners model (corner counts are not part of the evidence) · needs a model of when or how goals happen (halves, first goal) or of non-goal events (corners, shots), with its own data and a leak-free chronological evaluation
Tournament finalist — no tournament model · needs a season / tournament simulator over every remaining fixture (two-legged ties also need extra time and penalties)
Tournament winner — no season or tournament simulation · needs a season / tournament simulator over every remaining fixture (two-legged ties also need extra time and penalties)

a family is supported only when its contract semantics are read by the classifier and its event is a function of the 90-minute score grid (or, for player events, of the simulation's goal allocation); nothing is supported because it exists on Kalshi every supported family is served by both modes from the selected mode's own snapshot; prices are the same under both.

Recent developments

Verified changes observed since the frozen prediction, from the provider pages Got%OnIt collects. They are information for you; none of them changes the frozen estimate.

No changes detected in the sources checked since the prediction (standings, season figures).

No injury list had been collected by the prediction instant, so additions since then cannot be told from the first report.

Team situationManchester City 3.2 d rest (short) · AEK Athens FC 2 absent regulars

Schedule, workload and availability of both clubs across every stored competition, as known at Mon 12 Oct, 23:31 PDT. Each fact carries its evidential label. Deep Dive applies the availability and schedule-load facts as stated adjustments with bands (Compare the modes shows each one); Core uses none of them.

Used by Core (and by Deep Dive's base)

finished results of the same competition inside the history window (goals, and expected goals where stored), weighted by age CONFIRMED
the two teams' identities (attack / defence strengths), the target's home / away status, kick-off and season labels CONFIRMED

Core (the configured headline engine) uses none of the cross-competition or player facts below; its inputs are the finished results above. Deep Dive applies the availability and schedule-load facts as stated adjustments with uncertainty bands and is recorded as its own identity; the walk-forward experiments on this evidence (report §14) did not justify changing Core itself.

Availability and schedule load (Deep Dive adjustments) and context

Not used by Core: cross-competition schedule, rest days and congestion; player minutes, rotation and absences; reported injuries and suspensions; international duty; line-ups; standings. Deep Dive uses rest days and congestion, reported absences and line-ups from this list as adjustments with stated bands; standings and international duty remain context.

Manchester City

3.2 days since the previous match (short rest) · 2 in 7 days, 3 in 14, 3 in 30 · 2 still to play before kick-off CONFIRMED

Next: Premier League A Aston Villa +3.7 d

Recent matches, all competitions (30 days)

Sun 11 OctPLA LiverpoolW 3–0counted as workload
Wed 14 OctUCLH Paris Saint GermainNScounted as scheduled workload
Sat 17 OctPLH IpswichNScounted as scheduled workload

Upcoming (14 days)

Sat 24 OctPLA Aston Villa+3.7 d
Wed 28 OctEFLH Brighton+8 d
Sat 31 OctPLH Brighton+10.8 d

Players

Last stored line-up: Premier League, Sun 11 Oct (11 starters, 20 in the squad); 65 line-ups stored.

Regulars absent from the last squad: P. Foden (M) — reason unknown CONFIRMED ESTIMATED

Line-up changes last match: 1 · last XI average minutes in 7 days: 0 CONFIRMED

Manchester City: the last XI's minutes in 7 and 14 days and international appearances
Last XI	Pos	Min 14 d	International (14 d)
I. Ndiaye	M	56	—
R. Cherki · regular	M	90	—
Matheus Nunes	D	90	—
Rúben Dias · regular	D	90	—
G. Donnarumma · regular	G	90	—
E. Haaland · regular	F	80	—
A. Semenyo · regular	M	90	—
E. Fernández	M	80	—
E. Anderson · regular	M	90	—
N. O'Reilly	D	66	—
M. Guéhi · regular	D	90	—

International duty in 14 days: none in the stored national-team competitions CONFIRMED

AEK Athens FC

6 days since the previous match · 1 in 7 days, 1 in 14, 1 in 30 · 1 consecutive away · 1 still to play before kick-off CONFIRMED

Recent matches, all competitions (30 days)

Wed 14 OctUCLA Shakhtar DonetskNScounted as scheduled workload

Upcoming (14 days)

No match stored in the next 14 days.

Players

Last stored line-up: UEFA Champions League, Tue 8 Sept (11 starters, 21 in the squad); 13 line-ups stored.

Regulars absent from the last squad: T. Strakosha (G) — reason unknown; O. Pineda (M) — reason unknown CONFIRMED ESTIMATED

Line-up changes last match: 2 · last XI average minutes in 7 days: 0 CONFIRMED

AEK Athens FC: the last XI's minutes in 7 and 14 days and international appearances
Last XI	Pos	International (14 d)
S. Pilio · regular	D	—
R. Marin · regular	M	—
A. Brignoli	G	—
A. Koita · regular	M	—
H. Moukoudi · regular	D	—
L. Jovic · regular	F	—
F. Relvas · regular	D	—
L. Majer	M	—
L. Rota · regular	D	—
B. Varga	F	—
M. Vitalis	M	—

International duty in 14 days: none in the stored national-team competitions CONFIRMED

Reported injuries and suspensions REPORTED

Injury information unavailable: no injury report has been observed for this match, so absences are unknown, not zero.

Starting line-ups UNKNOWN

Lineups not confirmed. The provider publishes no projected line-ups, so no probable XI is shown here.

Unknown or unavailable

Travel distances and time-zone changes: unavailable. The provider gives venue names only: no coordinates, so distances and time-zone changes are not estimated; 'abroad' counts away matches in a competition of another country. UNKNOWN
International friendlies and competitions of confederations that are not stored (Africa, Asia): a club player's minutes there are unknown, so the international figures above are a lower bound. UNKNOWN
A player's fitness when no injury report names them: a missing entry is not a fit player. UNKNOWN
The reason a regular was missing from the last squad list: an absence is observed, never read as an injury. UNKNOWN
Whether a reported suspension applies to this competition: the provider does not state the scope. UNKNOWN
Projected line-ups: Only when a full XI was observed before kick-off; the provider publishes no projected line-ups, so 'probable' line-ups do not exist here. UNKNOWN

Labels: CONFIRMED = directly supported by stored records; REPORTED = provider-reported, not independently confirmed; ESTIMATED = derived from observed patterns; UNKNOWN = insufficient evidence; never read as zero or as fit. Situation situation-v1; schedule schedule-features-v1; players workload-features-v1.

Player availabilityInjury information unavailable

Deep Dive uses the reported absences and the confirmed line-up listed here as stated adjustments with bands (see Compare the modes); Core does not.

Injury information unavailable: we haven't collected an injury report for this match yet.

A player listed here is reported by the provider as out or doubtful; the list does not say who would have started.

Starting lineupsLineups not confirmed

Lineups not confirmed yet. They are usually published about an hour before kickoff; we start checking 90 minutes before the game.

This season8th v 16th

This season so far, as published by the provider. Context only; not an input to the estimate.

Manchester City

8th of 36 · 3 pts · W1 D0 L0 from 1 games · scored 2, conceded 0

Home W0 D0 L0 (scored 0, conceded 0) · Away W1 D0 L0 (scored 2, conceded 0)

Clean sheets 1 · failed to score 0 in 1 games

Last five league games W (most recent last)

AEK Athens FC

16th of 36 · 3 pts · W1 D0 L0 from 1 games · scored 1, conceded 0

Home W1 D0 L0 (scored 1, conceded 0) · Away W0 D0 L0 (scored 0, conceded 0)

Clean sheets 3 · failed to score 1 in 3 games

Last five league games W (most recent last)

Table Last checked 2 h ago; the provider last marked the table updated on 12 Oct.

Player scoring (simulation)E. Haaland 60% anytime

Simulation · player layer on Deep Dive's goals Who the simulation makes score: each player’s share of the team’s goals is estimated from past scoring and expected minutes, then drawn 100,000 times with Deep Dive's adjusted team goals (absent players' shares reassigned). No Kalshi player market is listed for this fixture; simulation only.

Manchester City

availability REPORTED scoring share ESTIMATED XI not confirmed · 77% of goals assigned

Player	Anytime	2+	xG ± SE	Start
E. Haaland F	59.8% ± 0.2	23.0% ± 0.1	0.91 ± 0.003	83%
R. Cherki M	31.7% ± 0.1	5.7% ± 0.1	0.38 ± 0.002	89%
A. Semenyo M	30.3% ± 0.1	5.1% ± 0.1	0.36 ± 0.002	86%
N. O'Reilly D	15.2% ± 0.1	1.3% ± 0.0	0.17 ± 0.001	51%
M. Guehi D	8.8% ± 0.1	0.4% ± 0.0	0.09 ± 0.001	79%
J. Doku F	7.8% ± 0.1	0.3% ± 0.0	0.08 ± 0.001	4%
J. Gvardiol D	6.2% ± 0.1	0.2% ± 0.0	0.06 ± 0.001	60%
R. Dias D	4.4% ± 0.1	0.1% ± 0.0	0.04 ± 0.001	86%

AEK Athens FC

availability REPORTED scoring share ESTIMATED XI not confirmed · 97% of goals assigned

Player	Anytime	2+	xG ± SE	Start
R. Marin M	19.7% ± 0.1	2.1% ± 0.0	0.22 ± 0.001	91%
L. Jovic F	14.1% ± 0.1	1.1% ± 0.0	0.15 ± 0.001	89%
A. Koita F	13.0% ± 0.1	0.9% ± 0.0	0.14 ± 0.001	100%
B. Varga F	12.1% ± 0.1	0.8% ± 0.0	0.13 ± 0.001	67%
H. Moukoudi D	4.2% ± 0.1	0.1% ± 0.0	0.04 ± 0.001	95%
Zini F	3.3% ± 0.1	0.1% ± 0.0	0.03 ± 0.001	26%
D. Kutesa M	2.2% ± 0.0	0.0% ± 0.0	0.02 ± 0.000	19%
M. Gacinovic M	1.9% ± 0.0	0.0% ± 0.0	0.02 ± 0.000	16%

Run 2416 (mc-v1:dc+players, 100,000 draws), player inputs as of 12/10/2026, 09:35:45. A player absent from the injury report keeps full availability (UNKNOWN is never read as fit or as out); unassigned goal share goes to no player.

Core | Deep DiveDeep Dive selected

Deep Dive and Core agree to within half a point on the match result; the applied adjustments are small. Lineups not confirmed.

Compare modesCore and Deep Dive side by side

Market	Core	Deep Dive	DifferenceΔ
Manchester City to win	82.6%	82.6% 74%–89%	-0.0 pts
Draw	11.2%	11.2% 7%–16%	+0.1 pts
AEK Athens FC to win	6.2%	6.2% 3%–11%	-0.1 pts
Over 3.5 goals in total	56.4%	55.5% 43%–68%	-0.9 pts
Both teams to score	53.9%	53.1% 47%–60%	-0.8 pts
Over 2.5 goals in total	76.0%	75.3% 65%–84%	-0.7 pts
AEK Athens FC over 1.5 goals	20.0%	19.4% 15%–24%	-0.6 pts
Manchester City over 1.5 goals	82.4%	82.0% 74%–88%	-0.4 pts

Difference = Deep Dive − Core, in percentage points. The small range beside a Deep Dive figure is the range if context assumptions change: the span of that figure if every adjustment assumption sat at either end of its stated band (availability and schedule load together). It is not a statistical confidence interval. The Core figures come from that mode’s own stored snapshot (#903); nothing is recomputed.

What changed the estimate?2 adjustments applied

Player availability · Manchester City×1.000 (+0.0 %), band ×1.000–×1.000REPORTED
no reported absence with a measurable scoring share · Lineups not confirmed (expected line-up used)
Player availability · AEK Athens FC×1.000 (+0.0 %), band ×1.000–×1.000REPORTED
no reported absence with a measurable scoring share · Lineups not confirmed (expected line-up used)
Schedule load · Manchester City×0.990 (-1.0 %), band ×0.840–×1.168CONFIRMED
short rest (own) 1.00 → -3.7 %; congestion 14d (own) 1.00 → 1.3 %; congestion 14d (opponent) -1.00 → 2.1 %; regulars absent last (own) 1.00 → -0.8 %; regulars absent last (opponent) 2.00 → 1.7 %; xi minutes 7d centred (own) -1.00 → -1.6 %; xi minutes 7d centred (opponent) -1.00 → 0.7 %; xi changes last share (own) 0.09 → -0.1 %; xi changes last share (opponent) 0.18 → -0.6 %
Schedule load · AEK Athens FC×0.979 (-2.1 %), band ×0.839–×1.142CONFIRMED
short rest (opponent) 1.00 → 3.4 %; congestion 14d (own) -1.00 → -1.3 %; congestion 14d (opponent) 1.00 → -2.1 %; regulars absent last (own) 2.00 → -1.5 %; regulars absent last (opponent) 1.00 → 0.8 %; xi minutes 7d centred (own) -1.00 → -1.6 %; xi minutes 7d centred (opponent) -1.00 → 0.7 %; xi changes last share (own) 0.18 → -0.2 %; xi changes last share (opponent) 0.09 → -0.3 %

Availability constants are stated priors with bands (half of a missing player’s recent scoring share, band a quarter to three quarters; +3 % opponent goals per missing defensive starter, band 1.5–4.5 %). Schedule-load coefficients are walk-forward medians whose bands include zero. 2 adjustment(s) moved the figures.

Freshness. Deep Dive inputs as of 12 Oct 16:35 UTC; line-ups not yet published, expected XIs used (as of 14 h ago; newest input 22 h ago before that instant).

Core: team strength, scoring levels, home advantage and the low-score correction from results and expected goals; no line-ups, injuries or schedule inputs. Deep Dive: Core plus reported absences, confirmed or expected line-ups, the goalkeeper and schedule load, each with an uncertainty band; equal to Core when nothing defensible applies. Paired forward record so far: 0 settled fixture(s); neither mode is called more accurate until the pre-registered rule is met.

How this estimate was builtscorelines, team strength, evidence, calibration

Why this prediction

Deep Dive · frozen snapshot (Core + adjustments) What the frozen snapshot itself says: its probabilities, its expected goals and the evidence it was built from. Experimental adjustments are kept in their own section below.

Match result · headline and control

Outcome	Headline	Control	Control − headline
Manchester City win	82.6%	53.1%	−29.5 pts
Draw	11.2%	24.2%	+13.0 pts
AEK Athens FC win	6.2%	22.7%	+16.5 pts

Headline: soccer-deepdive-v1. Control: soccer-poisson-goals-v1 — goals-only control produced beside the working engine for matched comparison (Phase 5B); not a second opinion to average with.

Expected score

3–1 is the expected score (rounded expected goals 3.13 and 0.81). Manchester City 3.13 · AEK Athens FC 0.81 · total 3.94 expected goals.

Most likely scorelines (Manchester City–AEK Athens FC)

3–010.0%
2–09.5%
3–18.0%
4–07.8%
2–17.7%

From the frozen score grid (0–10 goals each; grid mass 0.9996).

Team strength derived from the frozen model inputs

Manchester City

Expected goals: 3.13
Scoring vs average: ×1.65
Conceding vs average: ×0.61

AEK Athens FC

Expected goals: 0.81
Scoring vs average: ×0.61
Conceding vs average: ×1.65

Manchester City (home): expected to score ×1.65 the competition's average for a home side and to concede ×0.61 the average.
AEK Athens FC (away): expected to score ×0.61 the competition's average for a away side and to concede ×1.65 the average.
Recent 261 finished matches of this competition averaged 1.90 home and 1.32 away goals (home advantage observed ×1.44).

ratios of the frozen expected goals to the competition's recent average for a side at that venue; each combines one side's attack with the other's defence as the frozen model multiplied them, since the snapshot stores no separate parameters; the competition average is a display reference, not a model input.

Evidence behind it

Manchester City: 151 matches across 8 competitions (150 with xG)
AEK Athens FC: 17 matches across 8 competitions (17 with xG)
Evidence pool: 7044 matches across 8 competitions over seasons 2023, 2024, 2025, 2026 (Premier League, La Liga, Bundesliga, Serie A, Ligue 1, Champions League, Europa League, Conference League)
xG coverage: 6964 with, 80 without · weight 0.5
Recency: half-life 365 days · home advantage as seen in this competition

Data freshness

Snapshot 14 h old · evidence cutoff Mon 12 Oct, 09:35.
Latest match in the evidence kicked off Sun 11 Oct; its result was observed Sun 11 Oct, 13:55.

Model version

soccer-deepdive-v1 · deepdive-v1 · digest 872d05eb3e7c0e3d · regulation_time_90_minutes · scope default_catalogue

Warnings and limitations (11)

xG stored for 6964 of 7044 evidence matches; others entered as goals only
Deep Dive = Core + player availability + schedule load; the availability constants are stated priors with bands (not fitted), the schedule-load coefficients are walk-forward medians whose bands include zero (artifact verdict RETAIN_FROZEN_ENGINE, pooled ΔLL +0.0021); every band is carried into the market bands
a missing injury report is not fitness, an expected XI is not a confirmed XI, and an unexplained absence has an UNKNOWN reason: nothing is inferred beyond the stored observations
home advantage, team strength and competition scoring levels are the base's; Deep Dive never re-estimates them (no double counting)
paired forward evaluation against Core: 0 settled fixture(s) when this snapshot was made; Deep Dive may be called more accurate than Core only when the pooled paired ΔLL (Deep Dive − Core, 1X2) is below zero with a bootstrap 95 % interval below zero over at least 200 settled paired fixtures; until then both modes are two views of the same match, never a ranking
EXPERIMENTAL candidate (status experimental, headline false): the keep rule is adopt-as-headline only if pooled ΔLL vs soccer-poisson-goals-xg-v1 < 0 with a bootstrap 95 % interval below zero over the full evaluation seasons and no competition with ΔLL > +0.005 and an interval above zero (research/ENGINE-V2-MULTI-COMPETITION.md); the headline engine of the product is unchanged
evidence spans several competitions (2: 518, 3: 534, 39: 1199, 61: 971, 78: 963, 135: 1198, 140: 1217, 848: 444); a club's strength is its league-group level plus its own fitted terms; clubs from leagues that are not stored (outside:uefa) share one league-strength bucket and are shrunk toward it, so their individual strength rests on their UEFA matches only
draw calibration, home-advantage and league-strength terms are fitted per competition / league group from past results and xG only; no market data, lineups or injuries enter
Over/Under, BTTS, team totals and spreads are derived from the same score grid and stored for every competition including UEFA Champions League; derivable is not validated — only the chronological evaluation in research/ENGINE-V2-MULTI-COMPETITION.md says anything about them, and nothing is forward-validated
historical chronological evaluation is not forward validation; settled forward 1X2 snapshots for this model and competition when this snapshot was made: 0
cross-league target: the two clubs come from different league groups; their relative strength depends on the fitted league-strength terms (Phase 8A found the frozen engine biased here)

Calibration per market (3 measured of 48)

Market	Frozen	Documented calibration
1x2:home	82.6%	over-stated by 2.0 pts on 58 matches (this competition); conservative 75.6%
1x2:draw	11.2%	over-stated by 3.5 pts on 125 matches (this competition); conservative 4.9%
1x2:away	6.2%	over-stated by 6.6 pts on 35 matches (this competition); conservative 0.0%
double_chance:1x	93.8%	limited evidence: no numeric allowance
double_chance:x2	17.4%	limited evidence: no numeric allowance
double_chance:12	88.8%	limited evidence: no numeric allowance
btts:yes	53.1%	limited evidence: no numeric allowance
btts:no	46.9%	limited evidence: no numeric allowance
total_goals:over:0.5	97.9%	limited evidence: no numeric allowance
total_goals:under:0.5	2.1%	limited evidence: no numeric allowance
total_goals:over:1.5	90.5%	limited evidence: no numeric allowance
total_goals:under:1.5	9.5%	limited evidence: no numeric allowance
total_goals:over:2.5	75.3%	limited evidence: no numeric allowance
total_goals:under:2.5	24.7%	limited evidence: no numeric allowance
total_goals:over:3.5	55.5%	limited evidence: no numeric allowance
total_goals:under:3.5	44.5%	limited evidence: no numeric allowance
total_goals:over:4.5	35.9%	limited evidence: no numeric allowance
total_goals:under:4.5	64.1%	limited evidence: no numeric allowance
total_goals:over:5.5	20.5%	limited evidence: no numeric allowance
total_goals:under:5.5	79.5%	limited evidence: no numeric allowance
team_total:home:over:0.5	95.6%	limited evidence: no numeric allowance
team_total:home:under:0.5	4.4%	limited evidence: no numeric allowance
team_total:home:over:1.5	82.0%	limited evidence: no numeric allowance
team_total:home:under:1.5	18.0%	limited evidence: no numeric allowance
team_total:home:over:2.5	60.6%	limited evidence: no numeric allowance
team_total:home:under:2.5	39.4%	limited evidence: no numeric allowance
team_total:home:over:3.5	38.3%	limited evidence: no numeric allowance
team_total:home:under:3.5	61.7%	limited evidence: no numeric allowance
team_total:away:over:0.5	55.4%	limited evidence: no numeric allowance
team_total:away:under:0.5	44.6%	limited evidence: no numeric allowance
team_total:away:over:1.5	19.4%	limited evidence: no numeric allowance
team_total:away:under:1.5	80.6%	limited evidence: no numeric allowance
team_total:away:over:2.5	4.8%	limited evidence: no numeric allowance
team_total:away:under:2.5	95.2%	limited evidence: no numeric allowance
team_total:away:over:3.5	0.9%	limited evidence: no numeric allowance
team_total:away:under:3.5	99.1%	limited evidence: no numeric allowance
spread:home:-2.5	44.3%	limited evidence: no numeric allowance
spread:away:+2.5	55.7%	limited evidence: no numeric allowance
spread:home:-1.5	64.9%	limited evidence: no numeric allowance
spread:away:+1.5	35.1%	limited evidence: no numeric allowance
spread:home:-0.5	82.6%	limited evidence: no numeric allowance
spread:away:+0.5	17.4%	limited evidence: no numeric allowance
spread:home:+0.5	93.8%	limited evidence: no numeric allowance
spread:away:-0.5	6.2%	limited evidence: no numeric allowance
spread:home:+1.5	98.3%	limited evidence: no numeric allowance
spread:away:-1.5	1.7%	limited evidence: no numeric allowance
draw_no_bet:home	93.0%	limited evidence: no numeric allowance
draw_no_bet:away	7.0%	limited evidence: no numeric allowance

Measured gaps come from the leak-free retrospective cohort (A1 §5.1, §6; Champions League from Phase 8A); positive = the model over-stated the YES event. A conservative figure subtracts the over-statement and one standard error and never credits an under-statement. Nothing here changes the headline.

0 settled forward snapshots of this engine in this competition so far; too few to calibrate on, so the historical evaluation carries the figures.

Simulation100,000 draws · max gap 0.24 pts

Context · not a model input The exact probabilities are the headline; the simulation measures sampling uncertainty and powers the player and joint figures.

100,000 draws · seed 1363750586759496967 · mc-v1:dc+players · numpy PCG64 · 0.52 s

Match result · exact and simulated

Outcome	Exact (headline)	Simulated ± SE	Gap
Manchester City win	82.6%	82.6% ± 0.1	+0.06 pts
Draw	11.2%	11.2% ± 0.1	−0.05 pts
AEK Athens FC win	6.2%	6.2% ± 0.1	−0.00 pts

Largest gap to the exact grid over 50 markets: 0.24 pts (team_total:home:over:2.5); 0 outside three standard errors.

Clean sheets

Manchester City: 44.7% ± 0.2 (exact 44.6%)
AEK Athens FC: 4.3% ± 0.1 (exact 4.4%)

Goal difference (Manchester City minus AEK Athens FC)

<-50.0% ± 0.0
-50.0% ± 0.0
-40.1% ± 0.0
-30.4% ± 0.0
-21.3% ± 0.0
-14.5% ± 0.1
011.2% ± 0.1
117.6% ± 0.1
220.5% ± 0.1
318.2% ± 0.1
412.7% ± 0.1
57.4% ± 0.1
>56.1% ± 0.1

Correct score · top 5

Score	Simulated ± SE	Exact
Manchester City 3–0 AEK Athens FC	10.0% ± 0.1	10.0%
Manchester City 2–0 AEK Athens FC	9.4% ± 0.1	9.5%
Manchester City 3–1 AEK Athens FC	8.2% ± 0.1	8.0%
Manchester City 4–0 AEK Athens FC	7.8% ± 0.1	7.8%
Manchester City 2–1 AEK Athens FC	7.6% ± 0.1	7.7%

Expected goals (simulated)

Manchester City: mean 3.14 · variance 3.14 · ± 0.006
AEK Athens FC: mean 0.81 · variance 0.81 · ± 0.003

A Poisson variance equal to its mean is what the headline engine assumes; the simulation reproduces it, it does not test it.

Experimental adjustments (not in the headline)previous headline engine · 216 settled

Experimental · not in the headline Deep Dive already applies the availability and schedule-load adjustments with stated bands (see the modes block); no overlay is computed on it, so nothing here applies twice. Candidate engines are shown for information only. Forward evaluation of these adjustments: 216 settled so far.

Previous headline engine soccer-poisson-goals-xg-v1 · previous headline — frozen working engine (Phase 5F), kept for comparison; both identities keep their own forward records

Outcome	Headline	Candidate	Change
Manchester City win	82.6%	56.3%	−26.2 pts
Draw	11.2%	22.9%	+11.7 pts
AEK Athens FC win	6.2%	20.7%	+14.5 pts
Expected goals	—	1.82 – 1.01	−1.32 / +0.20

previous headline engine shown for comparison; the configured headline is the engine above. Snapshot 384 · poisson-xg-blend-v1 · digest 2814d9275461c8a3 · cutoff 11/10/2026, 09:30:10.

Detailed numberssnapshot, contracts, bookmakers, evidence

Model snapshot

Probabilities: home 82.6% · draw 11.2% · away 6.2%
Expected goals: 3.13 – 0.81
Manchester City eligible matches: 151 (150 with xG)
AEK Athens FC eligible matches: 17 (17 with xG)
Competition matches in the evidence set: 7044 by season 2023:2156, 2024:2280, 2025:2279, 2026:329
xG weight: 0.5
Cutoff: 12/10/2026, 09:35:45
Maturity: historically evaluated (primary target); forward calibration: none yet; probabilities come from chronological historical evaluation; no forward snapshot has settled yet, so nothing is forward-calibrated
Flags: xG stored for 6964 of 7044 evidence matches; others entered as goals only

Model limitations recorded with the snapshot

Deep Dive = Core + player availability + schedule load; the availability constants are stated priors with bands (not fitted), the schedule-load coefficients are walk-forward medians whose bands include zero (artifact verdict RETAIN_FROZEN_ENGINE, pooled ΔLL +0.0021); every band is carried into the market bands
a missing injury report is not fitness, an expected XI is not a confirmed XI, and an unexplained absence has an UNKNOWN reason: nothing is inferred beyond the stored observations
home advantage, team strength and competition scoring levels are the base's; Deep Dive never re-estimates them (no double counting)
paired forward evaluation against Core: 0 settled fixture(s) when this snapshot was made; Deep Dive may be called more accurate than Core only when the pooled paired ΔLL (Deep Dive − Core, 1X2) is below zero with a bootstrap 95 % interval below zero over at least 200 settled paired fixtures; until then both modes are two views of the same match, never a ranking
EXPERIMENTAL candidate (status experimental, headline false): the keep rule is adopt-as-headline only if pooled ΔLL vs soccer-poisson-goals-xg-v1 < 0 with a bootstrap 95 % interval below zero over the full evaluation seasons and no competition with ΔLL > +0.005 and an interval above zero (research/ENGINE-V2-MULTI-COMPETITION.md); the headline engine of the product is unchanged
evidence spans several competitions (2: 518, 3: 534, 39: 1199, 61: 971, 78: 963, 135: 1198, 140: 1217, 848: 444); a club's strength is its league-group level plus its own fitted terms; clubs from leagues that are not stored (outside:uefa) share one league-strength bucket and are shrunk toward it, so their individual strength rests on their UEFA matches only
draw calibration, home-advantage and league-strength terms are fitted per competition / league group from past results and xG only; no market data, lineups or injuries enter
Over/Under, BTTS, team totals and spreads are derived from the same score grid and stored for every competition including UEFA Champions League; derivable is not validated — only the chronological evaluation in research/ENGINE-V2-MULTI-COMPETITION.md says anything about them, and nothing is forward-validated
historical chronological evaluation is not forward validation; settled forward 1X2 snapshots for this model and competition when this snapshot was made: 0
cross-league target: the two clubs come from different league groups; their relative strength depends on the fitted league-strength terms (Phase 8A found the frozen engine biased here)
xG stored for 6964 of 7044 evidence matches; others entered as goals only
draws are known to be under-predicted by the engine on every historically evaluated league; no correction is applied
UEFA Champions League: only the 90-minute match result is evaluated (Phase 8A: over/under 2.5 not supported, BTTS encouraging); totals, margins, team goals, both-teams-to-score, correct scores and match-result NO positions are priced from the same score grid as limited evidence — research only, never High Confidence

Matches the model used (most recent first)

the matches the model actually used (finished before the cutoff, observed locally before the cutoff); goals and xG as stored.

Manchester City	Opponent	Venue	Score	xG	Result
11 Oct	Liverpool	away	3–0	1.36–0.94	W
20 Sept	Sunderland	home	5–3	2.82–3.57	W
13 Sept	Manchester United	away	1–0	1.10–1.00	W
8 Sept	FC Porto	away	2–0	2.32–0.28	W
5 Sept	Coventry	home	1–0	2.12–1.37	W
28 Aug	Crystal Palace	away	4–1	2.21–0.66	W

AEK Athens FC	Opponent	Venue	Score	xG	Result
8 Sept	Lask Linz	home	1–0	1.71–0.72	W
16 Apr	Rayo Vallecano	home	3–1	2.78–1.80	W
9 Apr	Rayo Vallecano	away	0–3	1.11–2.31	L
19 Mar	Celje	home	0–2	0.92–1.21	L
12 Mar	Celje	away	4–0	2.81–0.28	W
18 Dec	Universitatea Craiova	home	3–2	3.41–1.34	W

Kalshi contracts and the model

3 contracts across 2 families matched to this fixture. mid of the Kalshi YES bid and ask at the latest stored observation — a market reference, not an executable price. The NO ask is 1 − YES bid on the same book; resting sizes are contracts at the best level when observed ("?" = not reported, never zero).

Match winner	Model	Market ref.	Diff	YES bid / ask	NO ask	Resting (YES / NO)	Status
AEK Athens FC to win (90 minutes) KXUCLGAME-26OCT20MCIAEK-AEK	6.2%	7.0%	−0.8 pts	0.05 / 0.09	0.95	127 / 589	Comparable seen 33 min ago historically evaluated (primary target); forward calibration: none yet
Manchester City to win (90 minutes) KXUCLGAME-26OCT20MCIAEK-MCI	82.6%	83.5%	−0.9 pts	0.83 / 0.84	0.17	936 / 90	Comparable seen 33 min ago historically evaluated (primary target); forward calibration: none yet

Draw	Model	Market ref.	Diff	YES bid / ask	NO ask	Resting (YES / NO)	Status
Draw (90 minutes) KXUCLGAME-26OCT20MCIAEK-TIE	11.2%	9.5%	+1.7 pts	0.06 / 0.13	0.94	100 / 110	Comparable seen 33 min ago historically evaluated (primary target); forward calibration: none yet

Bookmaker consensus

median over bookmakers of each bookmaker's margin-adjusted implied probability (proportional normalisation of its own complete outcome set), at the latest page observed by Got%OnIt at or before the instant — a market reference, never a fair value. Margin adjustment: proportional: p_i = (1/odd_i) / sum_j (1/odd_j) over the bookmaker's complete outcome set from one page. bookmaker Match Winner and Goals Over/Under settle on 90 minutes plus stoppage (no extra time, no penalties), the same event as the score grid and Kalshi's regulation-time contracts; postponed or abandoned matches are voided by bookmakers but may be settled at a fair price by Kalshi, so a comparison is exact only for matches that reach a normal 90-minute result. Observations older than 24 h before the instant are not used; fewer than 3 bookmakers give no figure. Instant: 12/10/2026, 23:31:01.

Match result	Consensus	Range	Raw median	Odds median	Bookmakers	Model	Diff
1x2:home	—	—	—	—	0 only 0 bookmaker(s) quote the complete set (minimum 3)	82.6%	—
1x2:draw	—	—	—	—	0 only 0 bookmaker(s) quote the complete set (minimum 3)	11.2%	—
1x2:away	—	—	—	—	0 only 0 bookmaker(s) quote the complete set (minimum 3)	6.2%	—

No pages observed.

Total goals, 2.5 line	Consensus	Range	Raw median	Odds median	Bookmakers	Model	Diff
total_goals:over:2.5	—	—	—	—	0 only 0 bookmaker(s) quote the complete set (minimum 3)	—	—
total_goals:under:2.5	—	—	—	—	0 only 0 bookmaker(s) quote the complete set (minimum 3)	—	—

No pages observed. UEFA club competition: only 90-minute 1X2 is historically evaluated; totals are market data only

Sports evidence trail

availability is decided by Got%OnIt's observation time only; provider stamps are audit information; once the match has started the view is pinned to kickoff. Instant 12/10/2026, 23:31:01.

Dataset	State	Observed	Notes
Injuries	not_observed	—
Lineups	not_observed	—	before the window
Standings	ok	12/10/2026, 21:55:01 (Last checked 2 h ago)	fresh; provider stamp 11/10/2026
Season statistics · Manchester City	ok	12/10/2026, 21:55:27 (Last checked 2 h ago)	fresh
Season statistics · AEK Athens FC	ok	12/10/2026, 21:55:42 (Last checked 2 h ago)	fresh

Every stored market probability of the snapshot

Derived from the frozen score grid; "supported" is a numerical statement, not validation. Only families marked historically evaluated carry evidence; none is forward-calibrated yet.

Key	Probability	Support	Historical evaluation
1x2:home	82.6%	supported	primary_target
1x2:draw	11.2%	supported	primary_target
1x2:away	6.2%	supported	primary_target
double_chance:1x	93.8%	supported	not_evaluated
double_chance:x2	17.4%	supported	not_evaluated
double_chance:12	88.8%	supported	not_evaluated
btts:yes	53.1%	supported	supported_not_validated
btts:no	46.9%	supported	supported_not_validated
total_goals:over:0.5	97.9%	supported	not_evaluated
total_goals:under:0.5	2.1%	supported	not_evaluated
total_goals:over:1.5	90.5%	supported	not_evaluated
total_goals:under:1.5	9.5%	supported	not_evaluated
total_goals:over:2.5	75.3%	supported	encouraging
total_goals:under:2.5	24.7%	supported	encouraging
total_goals:over:3.5	55.5%	supported	not_evaluated
total_goals:under:3.5	44.5%	supported	not_evaluated
total_goals:over:4.5	35.9%	supported	not_evaluated
total_goals:under:4.5	64.1%	supported	not_evaluated
total_goals:over:5.5	20.5%	supported	not_evaluated
total_goals:under:5.5	79.5%	supported	not_evaluated
team_total:home:over:0.5	95.6%	supported	not_evaluated
team_total:home:under:0.5	4.4%	supported	not_evaluated
team_total:home:over:1.5	82.0%	supported	not_evaluated
team_total:home:under:1.5	18.0%	supported	not_evaluated
team_total:home:over:2.5	60.6%	supported	not_evaluated
team_total:home:under:2.5	39.4%	supported	not_evaluated
team_total:home:over:3.5	38.3%	supported	not_evaluated
team_total:home:under:3.5	61.7%	supported	not_evaluated
team_total:away:over:0.5	55.4%	supported	not_evaluated
team_total:away:under:0.5	44.6%	supported	not_evaluated
team_total:away:over:1.5	19.4%	supported	not_evaluated
team_total:away:under:1.5	80.6%	supported	not_evaluated
team_total:away:over:2.5	4.8%	supported	not_evaluated
team_total:away:under:2.5	95.2%	supported	not_evaluated
team_total:away:over:3.5	0.9%	supported	not_evaluated
team_total:away:under:3.5	99.1%	supported	not_evaluated
spread:home:-2.5	44.3%	supported	not_evaluated
spread:away:+2.5	55.7%	supported	not_evaluated
spread:home:-1.5	64.9%	supported	not_evaluated
spread:away:+1.5	35.1%	supported	not_evaluated
spread:home:-0.5	82.6%	supported	not_evaluated
spread:away:+0.5	17.4%	supported	not_evaluated
spread:home:+0.5	93.8%	supported	not_evaluated
spread:away:-0.5	6.2%	supported	not_evaluated
spread:home:+1.5	98.3%	supported	not_evaluated
spread:away:-1.5	1.7%	supported	not_evaluated
spread:home:+2.5	—	unsupported_tail_mass	not_evaluated
spread:away:-2.5	—	unsupported_tail_mass	not_evaluated
draw_no_bet:home	93.0%	supported	not_evaluated
draw_no_bet:away	7.0%	supported	not_evaluated

Deep Dive · Snapshot 989 · soccer-deepdive-v1 · digest 872d05eb3e7c0e3d · data as of 12/10/2026, 23:31:01