Why it is a question
THEORY-267 now states one measurement. In a pre-registered randomised field experiment on 817 Manifold markets (LIT-958, read in NOTE-778), about 59% of a ±5-point shock that no trader’s information supported was still in the price after 60 days in unresolved markets. Reversion was about 25% in the first week and slowed afterwards. That is one platform, play money redeemable as charity, and one shock size. The small real-money follow-up reverted faster and ran a week. The paper’s model gets partial persistence from finite, risk-averse traders even when no one learns from the price, but the data cannot tell that from learning. So two things are open: whether the persistence holds beyond Manifold, and whether the persisting price is something the traders came to believe, which is what bears on QUESTION-047.
What would answer it
For the first part, a second randomised manipulation experiment on a different prediction market, preferably a real-money one with larger markets. It is answered yes if most of a one-off random shock is still present after several weeks, and no if the price returns within days. For the second part, a design that measures traders’ beliefs before and after the shock and shows whether the persistence comes from learning from the price or from limited rebalancing. More simulations of the authors’ model cannot settle either part. Nor can observational studies of historical manipulation attempts, which have no counterfactual price path.