Intelligent Probabilistic Risk Forecasting With Applications to Algorithmic Trading and Portfolio Optimization

A novel volatility forecasting approach is explored with applications in algorithmic trading and portfolio optimization. The mathematical algorithms applied to algorithmic trading will be the Kalman filter and Hidden Markov Models. The Kalman filter will be applied to construct a pairs trading strategy. The trading strategy using a Kalman filter is then extended to take advantage of a Hidden Markov model to identify different asset price regions. Three pairs of trading approaches, KFIVF explored in [15], DDIVF first explored in [16], and DDIVF-HMM introduced in [2], will be compared. The second topic that is discussed in detail in chapter 4 is portfolio optimization using Data-Driven exponential moving average (DD-EWMA). The volatility forecasting models are used to study the generalized dynamic portfolio optimization using intelligent probabilistic forecasts based on the data-driven t distribution of the portfolio returns distribution. [1].