"This thesis consists of two essays on evaluating mutual fund performance and its predictability. In the first essay, I study the ex ante predictability of 12 well-known predictors for fund performance from investors' perspective. The 12 predictors cover three major categories: fund characteristics, fund performance, and holding-based activeness measures, which are constructed using real-time information. For performance evaluation, I exploit two types of fund picking strategies with either rule-based approach or machine learning methods and find that utilizing machine learning can deliver superior real-time economic gains for investors with fund short-term performance being the primary driver underlying predictability. Specifically, using variable selection methods such as LASSO and elastic net at individual predictor level can generate annual 1.3%-1.7% real-time alphas after adjusting for standard risk factors. The essay further examines whether real-world investors react to those well-known predictors when evaluating mutual fund performance. Using a novel approach to decomposing fund returns, I find that conditional on investors' usage of CAPM, investors react to the components of CAPM alpha implied by predictors in different ways, and investor reaction to predictive information embedded in predictors is stronger within aggressive growth funds. These results provide empirical support for Gârleanu and Pedersen (2018) and suggest ex ante predictability exists not due to lack of investor reaction but as the compensation for employing costly algorithms to identify skilled managers.The second essay examines how decision-making hierarchy in team-managed U.S. equity mutual funds affects their performance and risk-taking behavior. Employing a unique hand-collected dataset, we find that vertically-managed funds with lead managers earn 75 bps per year lower Fama-French five-factor alpha than their horizontally-managed counterparts. Moreover, vertically-managed funds hold less concentrated portfolios and are exposed to lower residual risk, thus showing signs of inferior security selection ability. Using mutual fund industry as a laboratory, the second essay provides evidence supporting a horizontal decision-making structure in organizations functioning in an uncertain expectation environment. These results echo similar mechanisms as in recent cross-country studies on the benefits of democratic form of government for country's economic growth"--