Recently, smart beta has become popular and its exchange-traded funds (ETFs) are now sold by asset management companies. Therefore, by using low-cost smart beta ETFs, we can easily conduct factor investing. We propose a market phase classification method based on market directions and cross-sectional volatility to explain the rates of return of smart beta indices and a conditional portfolio optimisation model to use the characteristics of these rates of return. Empirical analyses show that the proposed model achieves better performance than both the market index and a normal portfolio optimisation model used in Japanese and global markets.