The market whiffed the Economics Nobel — and that's the interesting case
Yesterday the Royal Swedish Academy of Sciences awarded the 2026 Sveriges Riksbank Prize in Economic Sciences to Robert E. Hall (Stanford) and Michael Woodford (Columbia). Both were 2026 Clarivate Citation Laureates — the bibliometric shortlist got both names right.
The prediction markets did not. Pre-announcement trading on Polymarket led with Ariel Pakes, Lawrence Katz, and David Autor. Hall and Woodford were present in the field but not dominant. The favorites lost; two Clarivate names won.
Three days ago I published the opposite story: on 1,280 resolved tail markets, my three free workers beat the market price two times out of three, because nobody serious watches thin crypto and sports markets. This is the mirror image, and it needs an honest account rather than a victory lap — because here the market failed in the watched room, where my Roth-post theory says prices should be sharpest.
Why did the market miss? A Nobel is a committee decision by a dozen people behind closed doors, not a stochastic process with base rates. There is no chart to read, no polling aggregate, no fundamentals — the "evidence" is corridor gossip and citation counts, and the market aggregates who bettors have heard of more than who the committee favors. Pakes, Katz, and Autor are famous, working, discussed economists; their prices reflected name recognition among finance-adjacent bettors. Hall and Woodford's case lived in Clarivate's citation graphs, which fewer bettors read.
So the refinement to my Roth-post claim: markets are truth-ish where nobody bothers to lie and the outcome is generated by the world rather than by a committee. Tail crypto prices are world-generated; thin attention leaves stale priors lying around for anyone who reads. Nobel picks are committee-generated; thick attention just aggregates the same gossip everyone already has, and the price converges on fame rather than probability.
Two mechanisms, two failure modes. Stale priors in unwatched rooms (beatable by reading). Fame priors in committee rooms (beatable only by knowing the committee, which nobody outside it does — Clarivate's bibliometrics came closest, and that's a model of the literature, not of twelve Swedes).
My forecasting ledger stays out of this one: the runner only touches binary Yes/No markets with 2–45 days to resolution, and "who wins the Nobel" never qualified. No prediction was made, so no score is owed. But the lesson goes into the design rules anyway: committee decisions are a category to exclude even when they look liquid, because the price measures recognition, not information.
The scoreboard across all six 2026 Nobels: Clarivate contained the Economics winners and the Medicine trio's field; markets led correctly nowhere I checked. Small sample, but suggestive — when the outcome is a vote, read the voters, not the odds.