# England 1-2 Argentina: Postgame Audit ## Audit Verdict - Fixture: England vs Argentina - Scope: regulation time; England score first - Frozen C4: `1-1`, `1-2`, `0-1`, `2-1` - Lock timestamp: `2026-07-15T15:14:47Z` - Final score: England `1-2` Argentina - C4 settlement: **EXACT HIT** - Hit node: `1-2`, node 2 of four equal public predictions - Locked posterior for `1-2`: `14.142851%` - Stability of `1-2`: `8/8` declared forks This is a successful exact-score observation under the immutable V2.1 pregame record. It is not evidence by itself that Turtle is calibrated, generally predictive, or profitable. ## Chronology and Integrity The prediction was frozen at `11:14:47 AM EDT`, before the scheduled `3:00 PM EDT` kickoff. The original `artifact_hashes_v2_1_final.csv` remains unchanged. At postgame audit time its SHA-256 digest was `f839db025c72f78d1baeb271b33dff7e406b5b6b6d97cd8be3d1d945303a61cd`, and all 33 recorded pregame entries matched their stored hashes and byte counts. The postgame files are a separate append-only layer. They do not change the candidate set, node order, model probabilities, theorem, lock timestamp, or pregame signatures. ## What Turtle Got Right The final score landed exactly on `1-2`, the second-ranked score cell. Its independent count-kernel probability was `8.792810%`; the declared result, spread, total, BTTS, clean-sheet, and family terms raised its normalized posterior to `14.142851%`. It survived every declared sensitivity fork. The realized terminal state also matched these pregame threads: 1. Argentina win was the largest result bucket at `46.247531%`. 2. Argentina by one was assigned `32.515618%` in the spread thread. 3. Exactly three goals was the largest total bucket at `32.379100%`. 4. BTTS yes and no clean sheet were each assigned `60.144409%`. 5. England scoring and Argentina scoring were assigned `73.124689%` and `80.787569%`, respectively. Those components coherently identify the terminal family: Argentina wins by one, both teams score, and the match contains three goals. ## What Turtle Did Not Get Right The spread thread's largest bucket was draw at approximately `37.1%`, not Argentina by one at `32.5%`. The correct spread branch was present and selected, but it was not the thread mode. The model did not predict the match path. England led through Anthony Gordon in the 55th minute; Enzo Fernandez equalized in the 85th, and Lautaro Martinez won the match in the 90+2nd. Postgame reporting described England retreating after taking the lead. That late game-state compression is an explanation of how the selected terminal node arrived, not evidence that the pregame model forecast the timing or tactical sequence. The C4 carried `50.742775%` of normalized model mass. Almost half of the model's mass remained outside the four disclosed nodes. England's comparable-opponent realm also remained a conditional sparse pass with `n_eff=4.09341334`. Therefore, this hit must not be used to erase the documented uncertainty. ## Market Exercise Settlement The pregame market comparison was a zero-dollar paper exercise. No personal profit or loss belongs in this public audit. Three broad positions cleared Turtle's conservative pregame threshold and later settled correctly: Argentina regulation win YES, over 2.5 regulation goals YES, and England regulation win NO. Argentina to score was classified as thin rather than qualified. The exact `1-2` market settled YES, but it had been rejected before kickoff by the conservative uncertainty buffer: raw edge `+0.06143`, conservative net edge `-0.01857`. The favorable outcome does not retroactively turn that rejected paper position into a valid pregame recommendation. BTTS yes and under 3.5 also settled correctly after failing the declared value gate; they remain rejected under the original rule. ## Match Record The operator-supplied FOX final panel recorded Argentina with 64% possession, 14 shots, 6 shots on target, 540 passes, and 6 corners. England recorded 36% possession, 6 shots, 3 shots on target, 280 passes, and 1 corner. These postgame statistics support a territorial-control account of the result, but they were not pregame model inputs and cannot be used to justify the lock retroactively. Independent result and scorer corroboration: - AP: https://apnews.com/article/afa13ed9fa933f8b75bd56eb16546031 - Reuters: https://www.reuters.com/sports/soccer/englands-bid-hang-blown-apart-by-argentina-masterclass-2026-07-15/ - Guardian live report: https://www.theguardian.com/football/live/2026/jul/15/england-v-argentina-world-cup-2026-semi-final-live ## Carry-Forward Work No rule is rewritten merely because the card hit. The next research item is to measure late-game state transitions when a lower-possession side leads: retreat depth, possession compression, substitution state, and the hazard of conceding after minute 75. That study must be built from a historical sample before it can become a pregame modifier. ## Final Statement The faithful claim is narrow: the immutable regulation-time C4 contained the final score exactly, with `1-2` ranked second internally and retained in all eight declared stability forks. This is one exact hit under a conditional model, not proof of calibration, repeatability, or profitability.