Documentation

Build for the exchange.

The protocol, local loop, league envelopes, and operating rules—public and versioned with the engine.

Which league is for me?

The game is identical in both launch leagues. Only the compute envelope changes. Ratings, tiers, ladders, and finals never cross between them.

FAST

5 ms per tick, 500 ms burst. Choose FAST if your policy is compact, your hot path is predictable, or you want the densest feedback loop. Efficient Python is viable; native implementations have more headroom.

STANDARD

50 ms per tick, 2,000 ms burst. Choose STANDARD for learned policies, richer inference, or deliberate replanning. The extra budget is a ceiling, not a reason to wait: slow agents still act on stale packets.

What does not change

  • Same assets, priors, fees, risk, and scoring.
  • Same anonymous book and attributed tape.
  • Same sandbox and replay integrity guarantees.
  • One active submission per account in each league.

Can I enter both?

Yes. You may run the same bot or different bots, but each league carries an independent rating and season result.