A research-first crypto and prediction-market signal dashboard where a feature engine and an LLM risk reviewer propose paper trades behind hard compliance gates.
Research project. Paper trading only. Not investment advice.
Capabilities
Momentum, RSI-style oscillators, volatility, order-book spread, implied probability, liquidity, and time-to-expiry are combined into a single scored signal per eligible market.
Each proposed side receives a structured review — confidence, risks, missing data, and an approve or block verdict. The model is never the sole execution authority.
Live trading requires jurisdiction checks, a minimum edge, maximum position size, a daily loss cap, and stale-data protection. The default mode is paper simulation.
Sessions, entry prices, and settlement are persisted and reproducible, so the venture is studied as an engineering experiment rather than a black box.
Flow
Discover eligible short-horizon crypto and prediction markets, pull public data, and compute the feature set for each candidate.
The decision engine proposes one side per market with a risk state, then hands the compact context to the LLM reviewer.
Approved proposals are executed on paper at real reference prices, sized from a fixed notional, and tracked to settlement.
Any move toward live execution passes explicit compliance, edge, and loss-limit checks before it is ever considered.
Built with