This paper examines whether correlated trading activities of investor clienteles influence comovement patterns in stock returns. We focus on geography- and price-based return comovements identified in the recent literature. First, we show that retail as well as institutional investors specialize and exhibit a greater propensity to buy and sell stocks that have similar characteristics. Specifically, due to local stock preference and a propensity to gamble, investors specialize in local stocks, low-priced stocks, and stocks with lottery features. Next, we show that specialization-induced correlated trading generates return comovements within geography- and price-based categories. Both comovement patterns are stronger among stocks with higher retail concentration and are more difficult to arbitrage. Comovements also get amplified during periods of greater aggregate uncertainty and stronger consumer sentiment. Overall, our results provide direct empirical support for the habitat-based return comovements model proposed in Barberis, Shleifer, and Wurgler (2005).