post-race

name	post-race
description	Post-race results update for the current F1 round. Looks up actual race results, updates the notebook and diary with points scored and learnings. Use after each race finishes.
argument-hint	[round-number]

Record the results for Round $ARGUMENTS after the race.

All paths below are relative to the project root.

Steps

1. Read the picks diary (analysis/picks-diary-2026.md) to find the pick that was made for this round and the pre-race reasoning.

2. Check the notebook (analysis/picks_analysis.ipynb) to confirm MY_PICKS has an entry for round $ARGUMENTS and see who was picked.

3. Look up the actual race results — use web search to find the full race classification for this round:

   • Search for "F1 2026 [Race Name] Grand Prix results classification"
   • Find the full finishing order (all 20 drivers), DNFs, DNS
   • Determine how many points our picked driver scored
   • Note the fastest lap holder (bonus point if in top 10)

4. Update ACTUAL_POINTS in the notebook — add the entry $ARGUMENTS: <points> with a comment showing position and context.

5. Re-execute the notebook in-place from the analysis/ directory:

   cd analysis && uv run jupyter nbconvert --to notebook --execute --inplace --ExecutePreprocessor.allow_errors=True picks_analysis.ipynb
   

6. Update the diary entry for this round in analysis/picks-diary-2026.md. Replace the _TBD_ placeholders in the Result and Learnings sections with:

   Result section:

   • Actual points scored and finishing position
   • Brief race narrative relevant to our pick (what happened to our driver)
   • How the top 10 finished
   • Notable DNFs/DNS

   Learnings for future rounds section — this is the most important part. Capture:

   • Model calibration: How did predicted vs actual compare? Was the model overconfident/underconfident for any teams? Are practice boosts trustworthy for this tier of team?
   • Pecking order update: Tier the teams based on cumulative results so far (Tier 1/2/3/midfield/backmarker). Note any teams that moved tiers vs last round.
   • Reliability flags: Any new DNFs/mechanical issues to watch. Update or clear previous flags if teams proved reliable.
   • Strategic takeaways: Concrete observations that should influence future picks — which teams/drivers are undervalued or overvalued by the model? Any drivers to avoid?
   • Season tracker: Running points total, delta vs model prediction, drivers used, drivers remaining.

7. Update the optimizer CSV — run cat analysis/optimal_picks_2026.csv and check whether the completed round's status is now marked as COMPLETED.

8. Present a summary to the user:

   • What our pick scored vs expected
   • Key takeaways for future rounds
   • Running season score vs model projection
   • Any flags for the next round's pick

Important notes

• Points system: P1=25, P2=18, P3=15, P4=12, P5=10, P6=8, P7=6, P8=4, P9=2, P10=1, P11+=0. Fastest lap = +1 if in top 10.
• Sprint race points (if applicable): P1=8, P2=7, P3=6, P4=5, P5=4, P6=3, P7=2, P8=1. Add to race points for total.
• Be honest in the learnings — if the pick was bad, say so and explain what we'd do differently
• The diary learnings are the primary mechanism for improving picks round-over-round — be thorough
• If the user provides no round number, infer from the last entry in MY_PICKS