A trader’s edge lives in timing and position management, and most of what improves both is visible in their own order flow. A trading persona is a working model of how one trader behaves, estimated from their trades and used for two things: surfacing the few facts that would change their decision, and recommending strategies that fit or correct how they actually trade.

An LLM agent holding raw credentials is the ultimate confused deputy. OAP — Orbit’s implementation of the Agent Proxy Pattern — gives every trading agent a scoped, capped, revocable mandate instead: every order checked by a deterministic policy engine, every action receipted, loss provably bounded.

What separates a market’s forecast from a committee’s opinion is stake. This paper decomposes event prices into the populations behind them — expressive bettors, hedgers, arbitrageurs, informed flow — and shows how conviction-weighted judgment, Bayes, and Kelly compose into a discipline for trading events.

An event contract is a digital option — but the replication arguments that price ordinary derivatives stop working where events begin. This paper maps the boundary between replication and estimation, and builds the pricing and CVaR-hedging architecture that lives on the far side of it.