My Ph.D. dissertation, defended at Boston College in June 2005 and became public thereafter, consists of three chapters. All three chapters use Abel Noser (ANcerno) institutional trading data. For details of the dataset, visit the Abel Noser (ANcerno) Data Page: http://ganghu.org/an In the first chapter, I provide a simple yet previously unexplored explanation for the buy-sell asymmetry phenomenon: it is because previous studies use pre-trade benchmark prices to measure implicit trading costs. Buy-sell asymmetry is mainly driven by mechanical characteristics of measures of implicit trading costs. I further argue that different measures of implicit trading costs serve very different purposes: pretrade measures are suitable for measuring trading costs of investment strategies, and during-trade measures are suitable for measuring execution quality. I show that a pre-trade measure can be decomposed into a market movement component and a during-trade measure, and empirically the market movement component is the dominant component. In the second chapter, I analyze the profitability and informativeness of institutional trading in IPOs, using a large sample of proprietary transaction-level trading data. My results can be summarized as follows. First, institutions sell most of their IPO allocations within the first year after the IPO. Second, institutions realize most of the “money left on the table” for IPO allocations sold within the first year. Third, post-IPO institutional trading outperforms a buy-and-hold investment strategy in IPOs. Fourth, institutional trading has predictive power for subsequent long-run IPO performance, even after controlling for publicly available information. Overall, my results indicate that institutions do possess private information about IPOs and receive considerable compensation for participating in these IPOs. In the third chapter, we analyze the daily trades of a large sample of investment managers. We find that opinion divergence among managers arises in different types of stocks than it does with analysts. We test whether various trading patterns can predict stock returns. When managers trade together, future returns are similar regardless if they are buying or selling, however when managers trade against each other, subsequent returns are low, especially for stocks that are difficult to short. A plot of this disagreement-return relationship resembles a smile. This result is consistent with the hypothesis that opinion divergence can cause an upward bias in prices.