♔ Do the bots really play in period?

Method. Each bot played 150 self-play games at its serving temperature. The same analyzer computed identical, move-sequence-based metrics on the bot games and on random samples of the historical era corpora (Lumbra's Gigabase, OTB games) — no ECO tags, no cherry-picking. Resignation is adjudicated from the model's own win-probability head, since humans resign lost positions and policy networks otherwise grind to mate.

Honest residuals. Two systematic gaps remain, both explainable. First moves lean 1.e4 beyond the historical rate: the bots are fine-tuned from Maia-2, which was trained on modern online chess — one epoch of era data moves its opening prior a long way, but not all the way. Bot games run ~25 plies longer than historical games, with fewer draws: the bots resign (adjudicated from the model's own win-probability head) but never agree to draws — and the draw-by-agreement, including the famous "grandmaster draw," is a large feature of real tournament chess, especially in the Soviet era. Neither gap affects the in-period style signal above.