top of page


Optimal Auction Strategy

Kevin O'Neill, Aolong Li, Enhao Feng, Benjamin Bruce, Jiuqin Wei


Let's say there is a painting up for auction. You and 10 other bidders each place a single bid which remains secret until all are revealed together and the highest bid wins. Whoever placed this bid has to pay this amount for the painting. If you think the painting is worth $10,000, how much do you bid? Certainly not $10,001 or higher, since then you would lose total value if you have to pay for the painting. You could bid $10,000, since that would be a fair trade for the painting. But why not bid $9,000 and see if you can gain total value in the case you still bid enough to win? Or maybe $9,500? (Note: the answer will depend on what you expect others to bid. Your confidence in their valuations is usually modeled by a random variable.)

The goal of this project is to train an AI to bid according to optimal strategy in an auction. The above description of an auction is on the simple side, but there are plenty of possible extensions. For instance, this project may involve finding optimal strategy for a more complicated auction or helping the auctioneer maximize revenue by choosing the right type of auction.

Screen Shot 2022-06-03 at 11.31.35 AM.png
github URL
bottom of page