Intraday VaR: A copula-based approach

The availability of high-frequency trading data and developments in computing technology make it possible to evaluate Intraday Value-at-Risk (IVaR), a useful risk management tool for investors and regulators. In this study, we propose a new model to evaluate and predict the IVaR for stocks based on irregularly spaced intraday transaction data filtered by volume events. Our model obtains the joint distribution of volume durations and corresponding returns with a copula function, and IVaR predictions are generated via Monte Carlo simulations based on the estimated model. Backtesting results for 30 randomly selected Chinese stocks show that our proposed model outperforms a well-established alternative model in offering more precise IVaR over 30-and 60-min horizons. We also find that the mean of the IVaR results exhibits a W-shaped intraday pattern.