winners-curse
Winner's Curse
Overview
The winner's curse is the systematic tendency, in a common-value auction — one where every bidder is estimating the same uncertain value from noisy private signals — for the winner to be precisely the bidder who most over-estimated that value. Estimates scatter around the truth; the highest estimate is, by construction, the most optimistic one; and the auction hands the prize to whoever holds it. So the act of winning is itself evidence that you were too optimistic. Winning is bad news. A bidder who fails to account for this — who bids their honest unconditional estimate — will win exactly the deals they most overpriced and lose the ones they priced sanely, earning low or negative returns even while "winning."
The term was coined by three Atlantic Richfield (ARCO) engineers analyzing offshore oil-lease auctions, where firms repeatedly won tracts and then earned poor returns:
"If it is true, as common sense tells us, that a lease winner tends to be the bidder who most overestimates reserves potential, it follows that the 'successful' bidders may not have been so successful after all." — Capen, Clapp & Campbell (1971), Journal of Petroleum Technology
The counter-intuitive core: the more bidders you face, the MORE you must shade your bid, not less. With more competitors, the winning estimate is a more extreme order statistic — a bigger outlier above the truth — so conditioning on winning implies a larger over-estimate to correct. The correction is not "bid a bit under your estimate"; it is "estimate the value assuming you already know you won (i.e. assuming you were the most optimistic), then bid off that discounted number."
Compose with neighbors. Use nash-equilibrium first to model the auction as a game and see why symmetric rational bidders must all shade (unilateral honesty is dominated). Use expected-value-and-kelly after setting a curse-corrected ceiling — EV to check the bid is still positive-value net of the adverse selection, Kelly to size it if the auction repeats (ad auctions, VC deal flow). Use batna-zopa instead of this skill when the value is largely private and negotiated bilaterally — but bring the winner's-curse ceiling into that BATNA as your reservation price whenever a common-value component remains. Use anchoring as a guard: a rival's aggressive opening bid, a banker's "guide price," or your own pre-auction estimate all anchor you upward past the ceiling — set the ceiling before you see those numbers.