Penny Auctions Are Predictable

We study user behavior and the predictability of penny auctions, auction sites often criticized for misrepresenting themselves as lowprice auction marketplaces. Using a 166-day trace of 134,568 auctions involving 174 million bids on DealDash, the largest penny auction site in service, we show that a) both the timing and source of bids are highly predictable, and b) users are easily classi similar to ed into clear behavioral groups by their bidding behavior, and such behaviors correlate highly with the eventual pro similar to tability of their bidding strategies. This suggests that penny auction sites are vulnerable to modeling and adversarial attacks.