This paper studies daily institutional trading and market efficiency using a natural experiment- the split-share structure reform in the Chinese stock market between 2005 and 2007.The reform converts previously non-tradable shares to tradable shares,after their owners transfer a negotiated amount of shares to the tradable shareholders.(1) Prior to the reform announcement date T0,only passive domestic institutions trade in the right direction: they bought (sold) an abnormal amount of tradable shares in companies that end up announcing high (low) compensation ratio.This suggests that passive domestic institutions possess and trade on private information about company-specific reform details.(2) On the first trading day T1 after the reform announcement,both passive domestic institutions and individual investors sold a large and abnormal amount of stocks with high unrealized gains.This pattern persists for the next one month.Our results show that investors can be subject to the disposition effect for both informational and non-informational reasons.(3) The disposition effect has significant price impact on stock price.The opening and closing market prices on T1 for tradable shares of reform companies are both significantly below the fair values determined by the compensation ratios and market return over the non-trading period [T0,T1].The abnormal returns on T1 are more negative for the reform companies with large unrealized gains.(4) A trading strategy that buys reform companies on T1 and shorts the market earns about 6% return over the next month.This strategy is especially profitable for reform companies with large unrealized gains.(5) Active domestic institutions and qualified foreign institutions take advantage of the stock undervaluation created by the disposition effect.They are large net buyers of reform companies with large unrealized gains.