21% vs 68%
How much of each index is just yesterday’s price
Trailing price returns explain about 68% of the Fear & Greed Index and about 21% of the Perception Index. A sentiment score that restates price adds nothing to it; this test measures exactly that.
METHOD · Regression of daily index values on trailing BTC returns, matched windows for both indices.
~3 days
How far ahead the index moves before sentiment shifts
Shifts in the Perception Index lead broader market-sentiment shifts by roughly three days (t = 2.45). This is a measured lead on sentiment, and it is the only lead-time claim we make.
METHOD · Cross-correlation with significance testing across the full daily history.
Feb 2018
How far back the daily history runs, same formula
One score every day since February 2018, computed from the archived corpus with the same formula that runs today. Every historical value can be re-derived from stored articles.
METHOD · Full corpus archived; the index is recomputable from raw scored articles for any date.
r ≈ −0.02
The null result we publish anyway
Across 455 analyzed earnings calls, the numeric directness score (Evade-o-Meter) showed no relationship with subsequent returns. We report it as a communication-style descriptor, never a signal. Management tone cohorts did separate in the same sample, and the sample is small (66 calls in the cautious bucket), so tone ships as a measurement with that caveat attached.
METHOD · 455 earnings calls, 37 companies, returns measured at 7 and 30 days; full write-up in the earnings backtest.