Alpha Without Access: Execution Constraints in a Bonding-Curve Market

We report a market in which a positive return is visible in prices yet cannot be realized by a fixed trading policy, and we measure why. On a frozen set of 1,220 tokens approaching the pump.fun graduation boundary, a counterfactual buy that ignores the execution race shows a median gross return of +8.5% to +17.45%. Access to that return disappears within seconds: 69.0% of tokens are unreachable at the earliest resolvable state and 89.7% by eight seconds, a collapse reconstructed exactly from graduation timing. Granting the policy zero cost and entry-state pricing — a conditional optimistic bound, favorable whenever the curve does not retrace below its entry state within the unobserved arrival interval — the optimistic upper bound on expected gross return per token is negative at every observable latency, and its upper 95% confidence limit (hour-cluster bootstrap) stays below zero at every step. The negative sign is not latency alone: reaching more of the population fills more of a subset whose realized economics are worse, an adverse composition we observe rather than a selection mechanism we demonstrate. Observed, accessible, and realizable returns are three distinct quantities here, measured on the same tokens. We identify no owner of the favorable tokens, make no priority claim, and claim nothing beyond the single policy, window, and venue studied. Method note: language-model tools assisted with prose and analysis-code generation under author-defined estimands, policies, and reporting constraints; the author reviewed and takes responsibility for all analyses and claims. Reproducibility artifacts and hashes accompany the record.