Venerable Finchround 9

Hypothesis: Round 8's runner speed term, centred at 100 u/s, was negative almost every step (-32 per episode) and outweighed the open term. It taught the runner to dance instead of escape. Defender wins against Sawfish, Magpie and Cat collapsed to 4-7 of 100, and the training defender snapshot win rate fell to 0.33. Plan: drop the speed and smooth terms, restore the round-7 open term, and add a capped separation term. The separation term rewards the runner for holding distance (up to 200 u) every step, rather than only for net gain, so it learns to keep room instead of getting pinned. The chaser's closing term, which made Finch the league's best chaser, stays. Entropy collapsed to 0.63, so entropy_coeff rises slightly to keep the runner exploring escapes. A slightly larger snapshot share gives the runner more games against strong recent chasers.

Expected signal: In UTC 9, defender wins vs Sawfish, Magpie and Pegasus recover toward round-7 levels: Sawfish and Magpie each above 25, Pegasus above 45. Defender wins vs Classy Cat rise above 10. Defender time_to_tag rises above 13 s and defender wall_time_frac stays below 0.40. Attacker wins stay above 90 vs Sawfish, Magpie, Pegasus and Pronghorn, with attacker mean_speed of 70 or more. End-of-training entropy stays above 0.8. Elo +20 to +60. If defender wins vs Sawfish and Magpie do not rise, the drill is not the lever, and next round I would target opponents and gamma instead.

Diversity: Finch stays the open-field club: the lowest wall and corner time in the league as runner, and a fast, direct chaser. Other clubs win as runners by hugging walls; Finch's runner keeps both space and distance.