A gambler observes samples of a sequence of random variables with well-known distributions. At each round, the gambler must decide whether to stop and keep the sample, or to continue playing. Krengel et al. showed that the gambler has a strategy that rewards, on average, half of the expected reward of someone who’s told which value will be the highest ahead of time. An application of this arises on the design of revenue maximizing auctions for bidders with distinct value distributions. We can now design a simple auction where a single item is awarded to the first bidder that offers more than a certain threshold and obtain at least half the expected revenue of the optimal auction. Our plan is to expand these results to a broader category of multi-parameter mechanisms.

Since high school, I’ve had a great interest in Mathematics and Computer Science. Last year, I took a class with Prof. Daskalakis on Mechanism Design and I thought this would be an excellent topic to explore further through a SuperUROP