Daniel and Moskowitz (2016) document the momentum crashes in the US stock market: the returns to momentum experience infrequent but persistent strings of negative returns. Based on their findings, this paper investigates the phenomenon of momentum crashes in A share market. We confirm the existence of the momentum effect and the crash of momentum strategy in A Share market using weekly data. The crash happens together with, if not slightly before the decline of the whole market. When the market starts to decrease after a long period of expansion, the beta of the winner portfolio (high beta) remains high, and the beta of the loser portfolio (low beta) remains low. The momentum strategy of buying historical winners and selling historical losers is equivalent to buying a high-beta portfolio and selling a low- beta portfolio. When the market decreases, the high-beta portfolio will decrease more than the low-beta portfolio, and the momentum strategy will suffer a great loss. The crash can be avoided if we allow an investor to dynamically allocate her assets between a risky asset and a risk-free asset based on an optimization problem that maximizes the Sharpe ratio of the portfolio.