Trading strategies at optimal frequencies

The dissertation is composed of two chapters, that investigate theoretical and empirical aspects of the process of exchange under asymmetric information in modern financial markets.;The first chapter studies a continuous-time stochastic game of trading activity in a single asset market. The model has the following features. First, informed and liquidity traders optimally control the timing of their order submissions. Second, they continuously choose whether to take or provide liquidity, issuing market or limit orders. Third, uninformed traders learn from the order flow optimally exploiting the information in the limit-order book. I construct an equilibrium in this setting and characterize (i) price formation and (ii) how optimal submission intensities and liquidity supply-demand behavior depend on private information and market conditions. I show that informed traders actively use both market and limit orders in equilibrium resulting in time-varying informed liquidity supply. After an information event, inter-arrival times are positively autocorrelated, the price impact of all order types increases and the limit-order book shows depth unbalances. The findings shed light on empirical and experimental results in the literature and have implications for inference methods with high frequency data.;The model nests the cases in which liquidity provision is decentralized (limit-order markets) and centralized (dealer markets). I find that the speed of information transmission into prices is lowered in the decentralized version.;In the second chapter I exploit a central feature of the theoretical model: since submission intensities are strategic, the time spacing of the data can be used to learn about traders' optimal policies. In the first part of the chapter, I use a unique order-level dataset from the NYSE to test structural restrictions on optimal liquidity provision. Using non parametric methods and bootstrap methods for Markovian processes I find new evidence that traders react to competition in liquidity provision by optimally adjusting the frequency of their limit order placements.;In the second part of the chapter I propose a structural econometric model to analyze general order-level datasets. I show how to identify, exploiting the model structure, fundamentals of the asset value process and hidden states related to information regimes. These tools have potential applications in the design of quantitative trading strategies, the investigation of insider trading episodes and the measurement of volatility in asset pricing.