The author suggests approaches to modeling volatility of returns of financial assets, different from the existing higher level of accuracy when out-of-sample prediction (with the formal proof on the basis of procedure -the Model Confidence Set) by taking into account the dynamics of diversification of market potential, able to describe the transformation mechanism of clustering of volatility of returns on micro-level clustering of volatility of returns on the macro level, the example of the Russian financial market. Comparison of different approaches to modeling diveraification potential based on the model families MEWMA, OGARCH, DCC and realized covariation it was found that the best quality of forecasting volatility of financial assets yield in most cases is provided by using the DCC model to calculate the index of diversification potential. This is true for stocks, stock indices and random financial portfolios. It is better to calculate diversification potential based on the OGARCH model to predict the volatility of the profitability of Markovitzefficient financial portfolios. The results obtained can be used by private investors and financial institutions to predict the volatility of financial asset returns. Financial regulators can use the diversification potential index as an indicator of macroeconomic risks in general.