#	Question	My Pick	2×	Pts Available	Pts Earned
01	Predict the podium	Leclerc · Hamilton · Norris	2×	78	30
02	Pole position	Kimi Antonelli	—	10	—
03	Will Aston Martin get both cars into Q2?	No	—	10	10
04	Who qualifies highest (Stroll vs Pérez)?	Sergio Pérez	—	15	—
05	Fastest pitstop (team)	Red Bull Racing	—	25	—
06	Will the Safety Car or VSC be required during the GP?	Yes	—	15	15
07	Which team scores the most points (Racing Bulls vs Alpine)?	Racing Bulls	—	10	10
08	Where does Lewis Hamilton finish?	Podium	—	15	—
09	Fastest lap of the race	Lando Norris	—	18	—
10	Will we see a new Hungarian GP winner?	Yes	—	10	—
TOTAL				206	65

Final: 65 pts · 30% accuracy (3/10 clean hits: Q3, Q6, Q7) — the third-worst round of the season, immediately after its best.

Q1 podium: Norris · Verstappen · Antonelli. One leg of three landed and it landed sideways — Norris was called at P3 and won, so the leg paid face value 15 with no exact-slot bonus, doubled to 30. Leclerc (P1 call, finished P4) and Hamilton (P2 call, finished P5) both missed by a place or two. Verstappen and Antonelli were the two names the sheet never seriously considered, and they took the other two rostrum slots.

Notes

Pre-race decisions:

• 2× on the podium (Q1) for the fifth straight round, on ~21.1 combined EV against ~11.3 for the best binary (Q6).
• Antonelli deliberately omitted from the podium at a payout of 5 — graded the most probable rostrum-getter but judged correctly priced, “the market has priced the favourite correctly; there is no edge there.”
• Q4 (Pérez, 15) flagged in advance as “the value call of the sheet” — taking both the higher payout and the outcome judged more likely, on Spa’s 2.2s qualifying gap over Stroll.
• Q5 retained as a deliberate reversal of the R10 results-file recommendation to drop it, on the argument that 0-for-2 against a ~0.18 prior is unremarkable.
• Q6 identified as “the biggest mispricing here” (Yes at 15 to No at 10) and Q3 as “the safest points on the board.”
• Q10 (Yes) was written explicitly as a hedge on Q1 — “if Antonelli wins and our podium legs suffer, Q10 pays.”

Result:

• The three hits were the three calls the sheet reasoned hardest about, and all three landed on their stated mechanism. Q6 (15) — the Hungaroring produced its neutralisation, as argued. Q3 (10) — Aston Martin split, Alonso P16 into Q2 and Stroll out in Q1. Q7 (10) — Racing Bulls 5 championship points to Alpine’s zero, exactly the two-upgraded-cars thesis that also carried the fantasy sheet. The reasoning is not what failed this round.
• The headline failure is an internal inconsistency, not a bad read. The sheet backed Norris on the podium (Q1) and Norris for fastest lap (Q9) on the strength of McLaren’s new high-downforce floor — then took Antonelli for pole (Q2) at 10 while pricing Norris at 18, and took “new winner: Yes” (Q10) at 10, an outcome that Norris winning makes impossible. Norris took pole and won. One belief, held strongly enough to drive two picks, contradicted by two others on the same sheet: Q2 (18) + Q10 (15) = 33 points forfeited to inconsistency, which would have made this a 98-point round. Before locking a sheet, check that every question is priced off the same set of beliefs.
• Q10 was anti-correlated with our own Q1, and the file said so without noticing. The Norris podium leg was justified in part because “he won here in 2025” — which is precisely what makes “new Hungarian GP winner” false when he wins. The hedge was constructed against Antonelli winning; the live correlation ran the other way and was visible in the document’s own text. The R10 lesson about correlated questions was applied to the wrong pair.
• Hamilton was read correctly and still returned nothing — twice. He qualified P2, 0.012 off pole, vindicating the eight-time-winner/high-downforce case completely. Then a three-place grid penalty put him P5, and he finished P5. That single post-qualifying event blanked the Q1 P2 leg (14) and Q8 (15) together — 29 points of availability to one cause no EV model would have caught. The same penalty is the story of the fantasy round. Correlated exposure to one driver across two questions is the structural risk, whether the driver is fast or not.
• Q5 is now 0-for-3 at 25 pts a round — 75 points of availability, zero returned. Fastest pitstop went to Racing Bulls. The statistical defence still technically holds (0-for-3 against a 0.18 prior has probability 0.55, unremarkable), but that was never the real objection: the 0.18 prior is reputational and unverifiable from any data in this project, and three rounds of paying for it is enough. Adopt the R10 recommendation — take the priced favourite or skip the value play until pitstop data exists.
• Q9 is 0-for-3, and this time the model was right and the pick ignored it. R10 established that fastest lap goes to “a car out of position with a free late stop, not the leader.” The R11 note cited that re-weighting as driving the pick — then selected Norris, the pole-sitter and likely race leader. Fastest lap went to Leclerc (1:22.000, lap 58), who started P2 and finished P4: a car out of position, exactly as the heuristic specified. The heuristic works; it wasn’t followed.
• Q4 extrapolated a single race and lost. Stroll out-qualified Pérez P20 to P22 by 0.663, reversing the 2.2s Spa gap the pick was built on. The market’s 10-vs-15 pricing had it right. One qualifying session is not a trend — the same error class the R11 fantasy sheet avoided by refusing to buy Leclerc off one result.
• The 2× was still optimally placed, but for the first time not uniquely so. Q1 added 15 over its unmultiplied base of 15. Q6 would have added exactly the same 15; Q3 and Q7 would have added 10 each. Un-multiplied the sheet scores 50. The podium policy is 5-for-5 on being a correct home for the multiplier, but this is the first round where it merely tied rather than won — worth noting before the streak is treated as stronger evidence than it is.
• Season predictions total: 944 pts (R01: 97 · R02: 99 · R03: 87 · R04: 90 · R05: 28 · R06: 88 · R07: 56 · R08: 78 · R09: 93 · R10: 163 · R11: 65). 65 ranks 9th of 11, ahead of only R07 (56) and R05 (28), and 30% accuracy is the weakest since R07.
• Verdict: a sheet whose analysis was better than its score. Every question it argued from concrete evidence — the Aston Martin pace deficit, the Hungaroring’s incident profile, the Racing Bulls upgrade — paid. Everything else was lost to three fixable process faults: one belief priced inconsistently across four questions (Q1/Q2/Q9/Q10), one driver carrying two questions into a single grid penalty (Q1/Q8), and two punt questions now 0-for-3 that the previous results file already told us to drop (Q5, Q9). None of these are prediction problems. R12 Zandvoort is a sprint weekend with a new Racing Bulls lineup — resolve the McLaren pace question first and let every question inherit the same answer.