ICES: Small: New and Better Markets via Automated Market Making

Carnegie Mellon University, Pittsburgh PA

Investigators

Abstract

Automated market makers are algorithmic agents that provide liquidity within markets. They provide a vehicle to create markets - for trade, risk hedging, speculation, or information aggregation - that might not otherwise exist. This project applies to binary-payout markets: prediction markets, weather insurance, sports betting, important options markets, credit default swaps, etc. Research on automated market makers will enable new kinds of markets: combinatorial markets, markets with continuous event spaces, and decision markets (where a principal uses a market to decide on a course of action). The project will also study how to better facilitate trade in existing markets. The PI plans to develop better liquidity-sensitive market makers in a novel framework. He also plans to study settings in which the market maker has a good prior over event likelihoods and to test how sensitive a market maker - even with a perfect prior - should be to unbalanced exposure risk. Finally, the PI plans to test a range of market makers on data from widely-traded options contracts. This will validate the approach beyond prediction markets. In regular prediction markets, the increased liquidity from automated market making improves predictions; in combinatorial ones, it enables the market. One of the most popular uses of prediction markets is internal markets for corporations. Better prediction markets mean that companies will perform better. Prediction markets can be anonymous, enabling people who may be outsiders in traditional deliberative processes to have considerable weight in decision making, as long as they have quality insights. The markets to which the research potentially applies range all the way to the world's largest markets. Even small improvements from this research can thus have vast economic benefits. Better market making increases liquidity, so the social welfare in the system increases due to better allocations. The research focuses on making the market mechanisms better rather than making any one trader's strategy better; thus the benefits will be distributed broadly among market participants. Replacing human market makers with automated ones also has other advantages: higher speed, no human errors, and avoidance of intentional misbehavior. The proposed work will increase the adopters' competitiveness; the first adopters will likely be US based, yielding US competitiveness in particular.