Thermodynamics have been previously employed in financial modeling and research; perhaps most notably in the assumption of stocks undergoing Brownian motion with drift from which the Black-Scholes model of derivatives contract pricing starts. This research evaluates another thermodynamic concept – that of entropy – from an approach similar to thermodynamics in interpreting entropy as a time-specified state function of the financial market segment component returns. The ability of this entropy to forecast future realized volatility is evaluated empirically. Monthly data from The New York Stock Exchange and the Tokyo Stock Exchange was used to generate entropy, its constituent dimensions, and volatility estimates. Entropy and volatility are tested for separability by running time series regressions against both with the market liquidity as an independent variable. Volatility is similarly regressed against index implied volatility, also to be used as the benchmark in a GARCH model of stock market index volatility against which the statistical significance of different entropy measures are evaluated. It is found that GARCH models improve with entropy as an external regressor, likely due to its ability to capture features of changes in liquidity not captured by volatility.
