We develop a dynamic model of a market with two specialized sides: traders posting quotes (“market makers”) and traders hitting quotes (“market takers”).Traders monitor the market to seize profit opportunities, generating high frequency liquidity cycles. Monitoring decisions by market-makers and markettakers are self-reinforcing, generating multiple equilibria with differing liquidity levels and duration clustering. The trading rate is typically maximized when makers and takers are charged different fees or even paid rebates. The model yields several empirical implications regarding the determinants of make/take fees, the trading rate, the bid-ask spread, and the effects of algorithmic trading on liquidity and welfare.