Staff Report, Federal Reserve Bank of New York 246
We show that equity markets are typically two-sided and that trades cluster in certain trading intervals for both NYSE and Nasdaq stocks under a broad range of conditions-news and non-news days, different times of the day, and a spectrum of trade sizes. By “two-sided” we mean that the arrivals of buyer-initiated and seller-initiated trades are positively correlated; by “trade clustering” we mean that trades tend to bunch together in time with greater frequency than would be expected if their arrival were a random process. Controlling for order imbalance, number of trades, news, and other microstructure effects, we find that two-sided clustering is associated with higher volatility but lower trading costs. Our analysis has implications for trading motives, market structure, and the process by which new information is incorporated into market prices.