A New Approach to Volatility Modeling: The Factorial Hidden Markov Volatility Model

Abstract

A new process-the factorial hidden Markov volatility (FHMV) model-is proposed to model financial returns or realized variances. Its dynamics are driven by a latent volatility process specified as a product of three components: a Markov chain controlling volatility persistence, an independent discrete process capable of generating jumps in the volatility, and a predictable (data-driven) process capturing the leverage effect. An economic interpretation is attached to each one of these components. Moreover, the Markov chain and jump components allow volatility to switch abruptly between thousands of states, and the transition matrix of the model is structured to generate a high degree of volatility persistence. An empirical study on six financial time series shows that the FHMV process compares favorably to state-of-the-art volatility models in terms of in-sample fit and out-of-sample forecasting performance over time horizons ranging from one to one hundred days.

A Neural network enhanced hidden Markov model for tourism demand forecasting

“…Despite being effective in capturing time series non-linearity, particularly in the case of detrend series [25], ANN may be less effective in representing strong seasonal patterns [25,26]. Thus, this study draws upon earlier research [3,6,27,28,29] to develop a new hybrid model. Tourism data have been represented as a trend undergoing gradual alterations alongside a recurrent yearly peak and valley with somewhat dissimilar amplitudes.…”

“…Furthermore, the nonlinear approach applied by [6,27] is adopted for ANN-based modelling of the trend element. Whilst to capture the seasonal patterns, a framework of hidden Markovian model outlined by [28,29] is heavily extended with the multiplicative error model containing four components for the seasonally cyclical patterns, unexpected jump, event amplitude, and a random error term. In this way, the abilities of both ANN and hidden Markovian model are exploited to devise an application-specific technique to model equivalent aspects of tourism data.…”

“…First, this work introduces the multiplicative error model (MEM) structure into the tourism demand forecasting area. MEM has been commonly used in econometrics area in modelling financial volatility [29] and has been proved as a better choice in extracting the error term more cleanly, more consistent with large variability of the time series data and producing better forecasting performance [31] [32] [33]. Second, this work models the factors that impact the seasonal patterns with the idea of the hidden Markovian model, but through an extended format.…”

Modelling asymmetric market volatility with univariate GARCH models: Evidence from Nasdaq-100

“…However, his R-GARCH model does not capture leverage effect in the Nasdaq-100 series. In their recent studyAugustyniak et al (2018) propose Factorial Hidden Markov Volatility model, a new approach, to simulate volatility of the Nasdaq-100 index among other series. Their new approach outperforms other peer methods at modelling and forecasting Nasdaq-100 return series volatility at short and long-run horizons.…”

Markov Regime-Switching in Forecasting Models

“…In attempt to characterize the business environment and the switching between different regimes, Asea and Blomberg (1998) with the same data sets to show the better perfomance of HMM [11]. to 100 days [21].…”