Does Crypto Sentiment Extremity Widen Estimated Spreads? Evidence Depends on the Specification

Battle Count: 74

The paper addresses spread estimation and sentiment in crypto markets, which are relevant to market making and execution cost modelling. However, the central finding is explicitly negative for trading: the association is specification-dependent, not causally identified, and the author concludes it does not support a trading rule or structural liquidity premium. The methodological contribution (exposing specification sensitivity) is more valuable for research hygiene than for direct strategy development. The Corwin-Schultz estimator itself is a practical tool for spread estimation from daily data.

The individual econometric methods (OLS, logistic regression, Newey-West HAC, circular-shift permutation tests) are standard and well-documented. The complexity lies in the specification ladder design, the Corwin-Schultz estimator implementation with zero-floor handling, the two-part model decomposition, and the careful Holm correction across quintile tests. The replication package reduces practical implementation burden significantly.

Full replication package provided at GitHub commit 56177c9 with cached upstream inputs, environment lock, code-state identifier, file hashes, and deterministic outputs. Reconstruction scripts regenerate all table rows without live API calls. Release checks compare custom HAC calculation with independent library implementation. However, the upstream Fear & Greed index construction cannot be reproduced due to unpublished components.