Market Fear & Greed Analysis

THE CONTRARIAN PREMISE. Sentiment indices rest on the idea that extremes mark turning points: when everyone is fearful the selling is largely done, and when everyone is greedy the buying is. It is an old idea in market folklore, it has some support in the behavioural finance literature on investor sentiment, and it is much weaker as a timing tool than as a description - extremes can persist far longer than a contrarian can hold.

THE FOUR INPUTS AND THEIR WEIGHTS. Nominally: news sentiment 30%, market-cap change around 24 hours 30%, Bitcoin dominance 20%, stablecoin net flows 20%.

NOW EXAMINE WHAT THAT ACTUALLY MEASURES, because this is the lesson.

• MARKET-CAP CHANGE is 100 x (current total / anchor total - 1), using the eligible provider observation nearest 24 hours earlier within 23-25 hours. Totals must be positive and finite. Provider times cannot follow fetch or read times, and each source must be no more than 2 hours older than its fetch; the current source must also be within 2 hours of this read. Exact duplicates keep their earliest fetch; conflicting totals at one source time are ineligible; equal-distance anchors choose the earlier source. The observed comparison replaces the provider-reported percentage when verified; otherwise that percentage remains an explicitly UNVERIFIED FALLBACK, never a verified contraction or expansion. Both figures and the actual window are published separately. Verification checks the snapshot pair, not the provider's market census. Market cap reflects price, supply and changing universe composition; contemporaneous correlation with price is partly mechanical, while any predictive use needs an out-of-sample test.
• BITCOIN DOMINANCE IS A LEVEL, not a measured rise or fall. Its contribution to a fear score is a convention: even an independently measured rise could mean risk aversion or a Bitcoin-led advance.
• 'STABLECOIN NET FLOW' HERE IS AGGREGATE SUPPLY CHANGE, over about 24 hours. It moves only through issuance and redemption: selling crypto into existing stablecoins passes them between holders and leaves the aggregate untouched, so the popular 'dry powder' reading is unavailable from it. Exchange deposits and withdrawals or wallet flows are a different quantity which CAN carry positioning information, and this analysis does not measure them. See the stablecoin-liquidity analysis for the full argument.
• NEWS SENTIMENT is keyword-derived, inheriting every limitation of lexicon scoring.

THE WEIGHT-REDISTRIBUTION RULE. Every unavailable or unusable component is dropped, and the remaining nominal weights are scaled proportionally to sum to 100%. Missing news does not transfer its full weight to market cap; missing market cap, dominance or stablecoin flow is not scored neutral. The score publishes the weights actually used and the reasons for dropped components. Two readings of 65 can therefore represent different measurements.

A GENERAL DISCIPLINE. Whenever a composite redistributes weight on missing data, ask what the index BECOMES in the degraded case - and whether the output tells you which case produced it. If it does not, the number and its label are not the same object from one reading to the next.