Domestic mutual fund investors show obvious short-term speculative behavior. They pay excessive attention to short-term performance, blindly chase funds with outstanding short-term performance, but the returns they earn are not high. Such behavior is related to investors’ lottery preference. They exhibit a preference for assets with a small probability of a large payoff because they overweight the probabilities of such an extremely positive outcome in a payoff distribution. The pricing implications of this preference have been studied in the stock, option, and IPO markets. Here we investigate how lottery preference affects investors’ mutual fund trading behavior.

We use each fund’s maximum monthly return over the previous six months（hereafter, MAX）to represent investors’ lottery preference, and intend to verify that MAX can positively predict fund flows. Our sample covers 852 actively managed open-ended equity funds from year 2005 to 2019. First, we investigate whether MAX can positively predict future fund flows. We find a positive and significant relationship between MAX and future fund flows: a 1% increase in MAX is associated with an incremental flow of 1.43% over the following six months. This is in line with investors’ preference for extreme positive payoffs. Second, we investigate whether buying mutual funds based on MAX is rational. We find that high MAX cannot predict superior future performance: a 1% increase in MAX will cause 0.15% decrease in CAPM alpha. So it is irrational. Third, we investigate the heterogeneity of MAX－Flow relationship across investors with different rationality. We find that the positive MAX－Flow relationship exists in individual investors’ mutual fund flows, but it is not significant in institutional investors’ mutual fund flows. Lastly, we demonstrate that limited attention theory cannot fully explain our MAX－Flow relationship. After controlling the star fund effect, the positive MAX－Flow relationship still holds. The above results are robust.

Our paper makes contributions to at least two areas of the literature: First, we directly prove that lottery preference steadily exists in the mutual fund market, which deepens our understanding of Chinese mutual fund investors’ behavior and expands the research of MAX anomaly. Different from the research based on the stock market, this paper uses the fund flow data to directly observe the trading behavior of fund investors, and constructs a complete research path of “MAX－investors’ trading behavior－fund performance”. It is found that lottery preference only exists in individual fund investors, and institutional investors are not obvious; funds with extreme positive return MAX in the past will have lower future returns, which enriches the research on the influencing factors of fund performance. Second, we find that the extreme return of funds will affect investors’ fund selection behavior, which enriches the research on “fund performance－fund flows”. Previous studies on “fund performance－fund flows” focus on the influence of the average or cumulative value of historical returns on fund flows, but pay less attention to the influence of the distribution characteristics of fund historical income on fund flows. This paper finds that funds with extreme positive returns in the past will get more fund inflows in the future, which is enlightening for the regulatory authorities to supervise the investment behavior of fund managers.