Ping Liu

Latest Research Highlights

1. Prediction Markets as Data Marketplaces: The Price of Information under Automated Market Making

   With Zhan Pang

   A prediction market can be viewed as a crowdsourced data marketplace: traders supply information through their trades, and the prices they produce serve as forecasts for data users. The automated market maker's liquidity is the price of that information: under the logarithmic market scoring rule it equals the informed trader's expected profit per unit of information acquired and the market maker's expected net payout per unit of information reflected in the final price. The arbitrageur who corrects the price sees the order but not whether it came from the informed trader or the noise trader. Informed orders scale with liquidity while noise orders do not, so liquidity changes how well the arbitrageur can tell the two apart, and a deeper market need not deliver more information. The market maker weighs this when choosing liquidity and deciding whether to open the market, with information acquisition endogenous.

2. Rational Expectations Equilibrium with an Automated Market Maker

   With Zhan Pang

   In a Grossman–Stiglitz equilibrium with costly information acquisition, an automated market maker's liquidity affects price informativeness only through how many traders choose to become informed. In a liquid market, deeper liquidity damps price movements toward zero or one, raises the value of the signal, and so draws more informed traders in. Noise trading shelters them only up to a point: extreme noise carries the price to certainty and leaves the signal nothing to add. The market maker earns an expected profit in a shallow market and bears an expected loss in a deep one.

3. Automated Market Making and Liquidity Provision in Prediction Markets

   With Zhan Pang

   Under the logarithmic market scoring rule, liquidity is what the market maker pays, in expectation, for each unit of uncertainty a trade resolves. As the sole buyer of information, the market maker prices like a monopsonist and sets liquidity below the value it places on accuracy.

4. Put Credit Rating Agency's Money Where Its Mouth Is

   With Zhan Pang and Alexei Tchistyi

   Revise and Resubmit, Management Science

   The optimal contract for a rating agency that must both learn a bond's quality and report it truthfully involves no direct payment; instead, the agency is given the opportunity to profit from its own knowledge and earns only when its rating proves accurate. A chooser option on credit default swaps, mediated by a clearinghouse, implements the contract.

All papers