||Abstract: It has been shown in many papers (Peters E. E. (1994), Zhou W.X, Sornette D. (2003), etc.) that the Efficient Market Hypothesis (EMH) fails as a valid model of financial markets. The Fractal Market Hypothesis (FMH) is in a place as a more general alternative way to the EMH. The FMH can be formed on the following parameter: agents’ investment horizons. It lead us to conclude that a financial market is more stable when we adopt this fractal character in structures of agent’s investment horizons. For computer simulations, the Brock and Hommes model (Brock & Hommes (1998)) is modified. This adjusted model shows that various frequency distributions on agents’ investment horizons lead to different returns behavior. The FMH focuses on matching of demand and supply of agents’ investment horizons in the financial market. It is the cornerstone that holds financial markets together. The EMH assumes the market is at equilibrium. The FMH on the other hand asserts that investors have an information differently based on temporal attributes. Since all investors in the market have different time investment horizons, the market remains stable. Our simulations of probability distributions of agents’ investment horizons demonstrate that many investment horizons ensure stability of the financial market. The behavior of the model under dynamical changes of agents’ trading strategies is analyzed. It is demonstrated that choosing the strategy qualitatively close to the best one (herding) leads to the failure of the fractal structure of financial market and thus to higher returns volatility.