Skills/09
Quant & Markets
Models that trade, measured honestly.
The work
Prediction-market agents, momentum models and backtesting rigs. Markets are the harshest eval environment there is, which makes them a great place to sharpen agent design.
What I deliver
- Signal research and backtesting infrastructure
- Prediction-market and brokerage API agents
- Risk signals and portfolio monitoring
- Self-improving research loops
Tools
Proof
12 projects.
Recursive Momentum Trader
Claude agents autonomously research a regime-routed stock model
AlpacaJev
A paper-trading bot built to measure how good a typed-judgment model is
Kalshi Trading Pod
Three Claude agents research, backtest and paper-trade Kalshi markets
Polymarket Autonomous Trading Agent
Rust agent that has Claude price prediction markets and sizes bets with Kelly
Stock Universe (Equilibrium)
Survivorship-bias-free top-1000 US equity universe with 10 years of bars
Momentum Acceleration Ranker
Stock ranker scoring weighted returns amplified by price acceleration
Kalshi NBA Scalp Bot
Event-driven bot that scalps Kalshi NBA markets on every made 2 or 3
Self-Improving Market Bot
Karpathy-style autoresearch loop for a quant strategy, gated by CPCV
PennySurge
Case-control study of what actually precedes 100% penny-stock surges
Composite Risk Signal Dashboard
39-signal composite risk model with a Bull/Default/Bear leverage state machine
Kalshi UFC Favorite Bot
Stakes a fixed amount on every UFC favorite an hour before the fight
Model Portfolio Allocation Tool
Blend model-portfolio presets and export allocations in the source CSV format