Stress dashboard

The stress dashboard asks one question of every market: how far from itself is it today? We take each market’s Big-6 personality features, subtract the lifetime mean, divide by the lifetime standard deviation, and compute the length of the resulting vector. That length is the stress distance — zero means “exactly at structural identity,” higher means “unusually far from normal for this market.”

Magnitude isn’t enough. We also project the Big-6 z-scores onto a risk-off axis (high vol / high VoV / high kurt / negative skew = +; opposite = -) to get a signed stress. Positive means the market is leaning risk-off; negative means it’s drifting the other way. Combine the two and you can read: “this market is far from itself, and the direction is risk-off / risk-on.”

18 live markets · readings as of 2026-05-22, 2026-09-11, 2026-11-04, 2026-11-05. The stress label self-normalizes to each market’s own history, so an “Extreme” reading on NG means “extreme for NG,” not “extreme absolute magnitude.”

Markets are ranked top-to-bottom by stress distance. Bar colors show the self-normalized bucket (Normal / Elevated / Stressed / Extreme). The right-end chevron shows directional lean — red right-arrow is risk-off, green left-arrow is risk-on.

Full table with per-market stress distance, percentile rank, bucket label, signed lean, and the two most extreme Big-6 drivers. Click a market to open its personality page.

Two numbers per market, one label.

• Distance — the magnitude in z-space. Zero means the market sits exactly at its lifetime average Big-6 position. Higher means further. There is no “good” or “bad” distance; it just says “unusual for this market.”
• Signed — the same reading projected onto the risk-off axis. Positive means the market is leaning into stress (high vol, fat tails, negative skew). Negative means the opposite — a market that’s unusually calm, upside-skewed, or quiet.
• Label — the distance’s percentile against this market’s own history. Normal = below the 50th; Elevated = 50-80; Stressed = 80-95; Extreme = 95-100. Self-normalized, so an Extreme reading always means “extreme for this market.”

How to read a row. PL is Extreme (99th percentile distance) with a +12.7 signed score — it is unusually far from itself today and the direction is clearly risk-off. ES is Normal (21st pctile) with a signed score of -1.4 — close to its structural identity, with a mild risk-on lean. Both can be true at the same time in the same market universe; the stress score for each market is self-contained.

1. Features: the Big-6 — Hurst exponent, annualized volatility, vol-of-vol, kurtosis, skewness, sample entropy. (Same set used in the DNA radar and clusters pages.)
2. Normalization: each feature is z-scored against this market's own lifetime distribution. That means a stress reading of 2.0 on NG and a reading of 2.0 on ES both mean "two standard deviations from this market's own normal" — directly comparable across markets.
3. Distance: Euclidean norm of the 6-vector. Zero means "at structural identity," higher means "further."
4. Signed stress: z-vector dotted with the risk-off sign weights (Ann vol +1, VoV +1, Kurtosis +1, Skewness -1). Direction-neutral features (Hurst, sample entropy) get zero weight. Positive output = risk-off lean, negative = risk-on lean.
5. Bucketing: the distance's percentile rank vs the market's own full history of daily distances determines the label (Normal / Elevated / Stressed / Extreme).

Last rebuild: 2026-11-07T03:03:32+00:00. Source pipeline: build_stress_distance.py. Companion artifact: personality_stress_history_.parquet under output/stress_history/ (full daily distance + signed series).

• All markets — tile grid grouped by archetype.
• Personality clusters — structural map + today’s state map across all live markets. The stress score IS the distance between those two snapshots.
• Feature glossary — plain-English definitions for the Big-6 features.