Flexible Caimanround 3

Hypothesis: The defender freezes because it barely sees its own win: with gamma 0.99, a timeout 480 steps away is discounted to about 0.008, so in mirror play only the attacker's short-horizon tag reward is learnable. That explains the mirror attacker win rate of 1.0, episodes of 67 steps and a frozen defender. Raising gamma to 0.997 (about 0.24 at 480 steps) makes survival worth learning. The new drill targets the defender with per-step terms: keep the opponent far away, stay in open space away from walls and corners, and avoid standing still when the attacker is near. The attacker keeps a small telescoping closing term so its chasing stays sharp against a harder practice partner. A little more entropy also fights the freeze.

Expected signal: Training: mirror attacker_win_rate drops from 1.0 toward 0.7-0.85, and episode_len_mean rises from 67 to 150+. Tournament: defender mean_speed rises from 14.5 toward 30+. Defender wins rise vs bots-greedy (28 to 40+), Classy Cat (32 to 40+), and Sawfish and Pegasus (39 to 50+). Attacker wins vs the bottom half stay above 95, and attacker wins vs Pronghorn and Classy Cat stay level or rise slightly. Elo +20 to +60. card_auc should be above 0.6 and cap_hit_frac near 0.

Diversity: With no charter, the club aims to be a balanced two-role club: a fast, direct attacker (high speed, few direction changes) paired with a mobile open-field defender. This contrasts with the league's corner-hugging defenders (league defender corner_time 0.25).