A New Decision Making Strategy: Combining Markov Chain Market Prediction with Kelly Criterion

Abstract

In some cases, solely relying on the traditional Kelly Criterion, which assumes fixed win and lose rates, is insufficient to cope with the constantly changing market environment. Some present attempts to develop an improved decision-making model combine the Kelly Criterion with a Markov chain, but they are either too complex or have limitations. To figure out a better method, this paper carries out research on dynamic position decision-making. A Markov chain model divides the market into two states by analyzing a state transition probability matrix based on historical price movements. These two states represent increasing and declining market conditions, respectively. Then, the Kelly Criterion is used to calculate the optimal position corresponding to each state, enabling state-dependent position sizing. Each step is easy to apply and does not require much calculation, making the method accessible for practical investment decisions. This novel and simple strategy forms a twostep decision-making system including state identification and position optimization. Empirical results from an A-share stock demonstrate that this dynamic strategy outperforms the static Kelly approach, offering improved cumulative returns while maintaining simplicity and ease of implementation.