🤓 Latest Submissions
Autopilot
Autopilot is an autonomous paper-trading and market-intelligence platform built for prediction markets. It continuously ingests live market data, identifies cross-market pricing inefficiencies, and ranks opportunities with a deterministic decision pipeline before any simulated order is placed. The system is designed around risk-first automation. Each opportunity is evaluated across multiple gates, including edge quality, liquidity depth, settlement confidence, and policy constraints. If a check fails, execution is blocked with explicit reasons, so operators can see exactly why a trade was rejected instead of relying on black-box behavior. Autopilot integrates data and execution logic for markets such as Kalshi and Polymarket, then enriches decision quality with external intelligence signals. It remains paper-only by design, enabling safe experimentation while preserving institutional discipline. A core differentiator is explainability and observability. Operators get a live terminal-style dashboard to monitor opportunity streams, risk state, agent activity, and execution outcomes in real time. Every action is logged with structured audit events so runs are reproducible and reviewable. The architecture is modular and test-driven: ingestion, scoring, agent logic, risk checks, execution simulation, and telemetry are separated into clear layers. This makes the platform robust, easier to debug, and fast to extend with new models or market strategies. In short, Autopilot combines AI-driven discovery, strict risk governance, and operator-grade visibility to make autonomous prediction-market trading research practical, transparent, and reliable.
31 May 2026
