The Auction Round at Prop Trading Firms

The auction round tests the winner's curse, the tendency for the highest of many independent estimates to overshoot the true value.

The auction round tests how you value an asset with uncertain worth and adjust your bid once you learn how many other people are bidding against you, rather than testing whether you can guess a number correctly. Interviewers use it to see whether you understand the winner's curse, the tendency for the highest bid in a group of independent estimates to be an overestimate of the true value, purely because it was the highest of many guesses rather than because it was the most accurate one.

A jar or an asset has a true value unknown to everyone in the room. Each of several bidders submits an independent estimate, and the highest bidder wins the item but pays their own bid. Because every estimate carries some error, the highest bid among a group of, say, ten independent estimates is systematically higher than the average estimate, and often higher than the true value itself, simply because it is the maximum of ten noisy draws rather than a typical one.

The interview version usually strips this down to a specific number: you are told there are 9 other bidders, each estimating a value independently with some spread around the truth, and asked how you would adjust your own bid to account for the fact that winning is itself informative about your estimate being too high.

If your bid wins, that outcome only happens when your estimate was the highest of the group, which is more likely to occur when your estimate has drifted above the true value than when it happens to be exactly correct. This is the winner's curse, and it means a bidder who bids their honest, unadjusted estimate will lose money on average across many such auctions, even though each individual estimate was unbiased before the auction started.

The correction is to shade your bid downward from your raw estimate, with the amount of shading increasing as the number of competing bidders grows and as the uncertainty in each individual estimate grows. With more bidders, the maximum of the group drifts further above the true average, so the winning bid needs a larger downward correction to stay accurate.