Computational Models of Algorithmic Trading in Financial Markets.
AbstractToday's trading landscape is a fragmented and complex system of interconnected electronic markets in which algorithmic traders are responsible for the majority of trading activity. Questions about the effects of algorithmic trading naturally lend themselves to a computational approach, given the nature of the algorithms involved and the electronic systems in place for processing and matching orders. To better understand the economic implications of algorithmic trading, I construct computational agent-based models of scenarios with investors interacting with various algorithmic traders. I employ the simulation-based methodology of empirical game-theoretic analysis to characterize trader behavior in equilibrium under different market conditions. I evaluate the impact of algorithmic trading and market structure within three different scenarios. First, I examine the impact of a market maker on trading gains in a variety of environments. A market maker facilitates trade and supplies liquidity by simultaneously maintaining offers to buy and sell. I find that market making strongly tends to increase total welfare and the market maker is itself profitable. Market making may or may not benefit investors, however, depending on market thickness, investor impatience, and the number of trading opportunities. Second, I investigate the interplay between market fragmentation and latency arbitrage, a type of algorithmic trading strategy in which traders exercise superior speed in order to exploit price disparities between exchanges. I show that the presence of a latency arbitrageur degrades allocative efficiency in continuous markets. Periodic clearing at regular intervals, as in a frequent call market, not only eliminates the opportunity for latency arbitrage but also significantly improves welfare. Lastly, I study whether frequent call markets could potentially coexist alongside the continuous trading mechanisms employed by virtually all modern exchanges. I examine the strategic behavior of fast and slow traders who submit orders to either a frequent call market or a continuous double auction. I model this as a game of market choice, and I find strong evidence of a predator-prey relationship between fast and slow traders: the fast traders prefer to be with slower agents regardless of market, and slow traders ultimately seek the protection of the frequent call market.
algorithmic tradinghigh-frequency tradingfrequent call marketmarket makingagent-based modelingsimulation
MetadataShow full item record
Accessibility: If you are unable to use this file in its current format, please select the Contact Us link and we can modify it to make it more accessible to you.