May 2025 – Jul 2025 · 2 mos · On-site

• Conducted high-frequency market microstructure analysis on Binance tick-level crypto data, using Hayashi-Yoshida covariance as a baseline. Addressed challenges of asynchronicity, irregular sampling, and intraday seasonality.
• Developed noise filtration algorithms to mitigate bid-ask bounce and rapid order book clearing, extracting efficient price signals and uncovering latent lead-lag dynamics in high-volume, volatile regimes with concentrated alpha opportunities.
• Statistically modeled lead-lag event distributions, leveraging assumptions of normality in price gaps to quantify lagging asset movements while incorporating time lags, volume surges, and trading fees for precise alpha estimation.
• Designed shape-based Dynamic Time Warping with DP state optimization (reduced from O(n*m) to O(n+m)), using the metric as a reward function to capture temporal distortions and arbitrage opportunities across fragmented markets.
• Built real-time, state-dependent metrics using lagged market observables to assess arbitrage conviction, predicting profit probabilities in lead-lag market pairs for live alpha quantification.