Where uncertainty is high, structure becomes the advantage.
I am drawn to problems where the data is noisy, the structure is hidden, and the decision still matters. Markets are the cleanest version of that challenge: uncertainty, incentives, timing, feedback, and consequences compressed into a live system.
My default mode is to make ideas falsifiable: turn the question into data, the data into a model, the model into a test, and the test into a decision rule that can be challenged.
Quant systems built as instruments.
Queue-Aware Market Making Simulator
A tick-level exchange simulator for testing quoting policies against queue position, adverse selection, inventory risk, fees, latency, and realistic fill dynamics.
Market questions written as papers.
Paper-style research built around falsifiable claims, market realism, careful validation, and a clear path from statistical evidence to tradable decision.
Queue Priority, Adverse Selection, and the True Cost of a Fill
A tick-level study estimating fill probability and expected adverse selection from order-book state, latency, fees, and queue position.
Learning Stable Cross-Asset Signals Under Regime Drift
A walk-forward study of lead-lag, residual, and macro-sensitive features with purged validation, turnover controls, and live-like costs.
Arbitrage-Free Volatility Surface Forecasting Around Scheduled Events
A volatility paper modelling skew and term-structure dislocations before earnings and macro releases, with constrained calibration and hedged P&L attribution.
Optimal Execution Under Transient Impact and Liquidity Uncertainty
A control study comparing static and adaptive execution policies under spread, impact decay, fill risk, urgency, and post-trade slippage.
Research that survives markets.
I treat a signal as an investment case, not a chart pattern: a precise hypothesis, a clean experiment, a known capacity limit, and a measured path from prediction to execution.
Write the trade before the test
Define the market mechanism, forecast horizon, universe, benchmark, failure condition, and economic reason the edge should exist before looking for evidence.
Make the backtest difficult to pass
Use clean time alignment, purged splits, out-of-sample periods, simple baselines, and stress by regime, liquidity, turnover, and capacity.
Translate signal into implementable P&L
Measure spread, fees, queue priority, borrow, latency, partial fills, market impact, and risk sizing so statistical edge is judged as tradeable economics.
Know when to scale, retrain, or retire
A model is not finished at research approval. It needs monitoring, drawdown attribution, drift checks, risk limits, and explicit criteria for reducing or removing risk.