Using High Frequency Market Data For Intraday Portfolio Risk Analysis And Optimization

Asset returns based on low frequency prices (e.g. end-of-day quotes) are still dominating modern portfolio analysis. To make portfolio metrics more relevant intraday and improve precision of estimates, new data frequency needs to be explored. In this presentation we demonstrate how using high frequency market data for portfolio risk management and optimization could improve the classic variance-bias trade-off and bring new insights to strategy backtesting. Since high frequency prices require special handling, we discuss key components of an automatic model pipeline for microstructure noise, price jumps, outliers, fat tails and long-memory. We conclude our presentation with an introduction to high frequency portfolio optimization built on top of intraday portfolio metrics.