Momentum, volume and investor sentiment study for u.s. technology sector stocks—A hidden markov model based principal component analysis

In this paper, we study the impact of momentum, volume and investor sentiment on U.S. tech sector stock returns using Principal Component Analysis-Hidden Markov Model (PCA-HMM) methodology. Price and volume are two well-known aspects in general equilibrium model. Momentum effect arises from the determination of prices in the market equilibrium. By studying momentum, volume and investor sentiment, we intuitively connect theoretical finance model with modern behavior finance topic. Instead of predicting future stock returns using machine learning models and doing comparisons, we apply the PCA-HMM method to reveal the hidden force in the financial and macroeconomic time series to calibrate different regimes. Combining the traditional financial study methods with modern machine learning techniques, we show investor sentiment effect show the primary effect on tech sector stock return which outweighs volume effect and momentum effect. The volume effect also has ineligible impact on stock return. The investor sentiment effect and volume effect show most impact on tech stocks with large or medium market shares. In contrast, momentum effect has very trivial correlation with tech sector stock return, from both stock level and individual state level. We also discuss the underlying mechanisms behind above findings according to tech sectors' unique characteristics, as well as raise risk management concerns. Using such PCA-HMM method, we reveal the unique patterns in tech sector stock returns. The PCA-HMM method can especially help us to identify those edge cases under which market behaves irregularly.