cross-chain-pmm-lps/docs/07-mesh-reflexivity.md

Mesh Reflexivity (Risk Controls)

If the bot is active across many pools, feedback loops can occur (e.g. bot buys in one pool, sells in another, churns fees, moves price).

Controls to prevent reflexivity

1. Global per-token trade budget per time window — cap total volume per token across all pools.
2. Minimum improvement threshold — do not trade unless δ improves by X bps net of fees.
3. Cooldowns per pool after intervention — avoid rapid repeated trades on the same pool.
4. Priority order — fix HUB pool first, then others; avoid circular arbitrage across quotes.

Policy memo

These controls should be documented in operational playbooks and enforced in bot code (v2). See ../spec/bot-state-machine.md.