This thesis examines the cross-sectional dynamic performance of the US stock markets through Principal Components analysis (PCA). We examine the annual and semi-annual performance, from 1928 to 2015, of the portfolios obtained from the top five principal components from past returns. These capture seventy percent of the variation in assets returns. The first principal component has over ninety percent annual correlation with the market. The second, third, fourth and fifth principal components have persistent characteristics different from the market. A Sharpe Ratio of 0.79 and maximum drawdown of 0.27 could be obtained by investing in a combination of the principal portfolios, compared to a Sharpe ratio of 0.40 for the market and a maximum drawdown of 0.80. The four-factors (market, size, value and momentum) and betting against Beta regressions show a significantly positive alpha, whence portfolio performance cannot be explained by these factors. Importantly, the composite portfolios exposure to industry sectors that are most affected by downturns tends to decline before the downturns take place.