Three different questions, three different composites
- Composite Sentiment
- What do current volatility, credit, positioning, breadth, and macro observations look like when normalized into a 100-point descriptive gauge? It uses 15 inputs across five point-weighted pillars.
- Composite Valuation
- Where do five selected U.S. equity valuation measures sit in their rolling or fixed reference distributions? It is an equal-weight average of five percentiles.
- Fragility Score
- How stacked are structural preconditions for a fast, forced-selling-driven decline? It uses five fixed-weight pillars, five-session median smoothing, and a separate stress activation state.
Liquidity and policy observations, economic health, market prices, and other indicators shown in the dashboard remain separate readings unless a composite explicitly lists them. A related indicator should not be assumed to be part of a score merely because it appears on the same tab.
Normalization and missing data
Percentile-style normalization makes most inputs with different units comparable; the Fragility registry also explicitly uses a z-score-to-percentile transform and module outputs that are already on a 0–100 scale. Direction matters: a high VIX is treated differently from a high breadth reading, and the detailed pages explain those choices. Composite implementations also expose data-integrity and confidence fields rather than hiding degraded inputs.
Fallbacks, caches, delayed publications, stale observations, provider revisions, and missing history are real limitations. The sentiment and valuation APIs identify fallback factors and alerts. Fragility excludes unavailable indicators and renormalizes the remaining weights rather than silently filling gaps. A score with incomplete data is not equivalent to a fully live score.
Source families
The app combines public market and economic feeds. The exact source and observation date can differ by indicator; the dashboard's source labels are the authority for a particular reading.
- Federal Reserve Economic Data (FRED) for many macroeconomic, rates, credit, and market series.
- Yahoo Finance for selected market and volatility history.
- Multpl for Shiller CAPE and S&P 500 price-to-book history.
- The Wall Street Journal market data for the current S&P 500 forward P/E input.
- Cboe, AAII, CFTC, CNN Fear & Greed, FinViz, Treasury Fiscal Data, and other clearly labeled provider feeds for individual indicators.
These are data sources, not endorsements. Market data can be delayed, and some provider pages may restrict access or change their formats. The app retains cached history for selected series to make comparisons possible, but cached data is not guaranteed to be current.
What these scores cannot do
- They do not forecast a precise market date, price, return, or crash probability.
- They do not provide individualized advice or account for objectives, tax status, time horizon, or risk capacity.
- They do not establish causation from correlations or historical percentile ranks.
- They cannot eliminate the risk of regime change, data error, publication delay, or an event outside the measured inputs.
Market Signal is for informational and research purposes only. Nothing on the dashboard is financial advice. Review the Privacy Policy and the dashboard's disclaimer before using the service.