Skip to main content
Numbers across the site — scoreboard cells, volatility badges, pulse observations — carry two compact statistical qualifiers. Both answer the same underlying question: is this reading unusual, or normal for this instrument?

​
σ — the z-score

How far the move sits from its own average, in standard deviations. −3.3σ means 3.3 standard deviations below what this particular series typically does over this particular horizon — measured against its own history, never a generic benchmark. That’s the point: a 2% day is nothing for a crypto basket and a genuine event for Utilities; σ makes them comparable. Rough calibration: |σ| < 1 is ordinary (don’t dramatize it), 1–2 is notable, ≥ 2 is genuinely unusual — that’s the threshold where cells get flagged and pulse observations fire.

​
p — the percentile

Where today’s reading ranks against that series’ own history. p97 = higher than 97% of comparable past readings; p0 = the most extreme low on record for that window. Same convention as latency metrics (p50, p99). The percentile rides alongside the z-score on purpose: returns are fat-tailed, so a σ figure alone can exaggerate or undersell — the percentile keeps the z honest. A −3.3σ · p0 cell says both “statistically extreme” and “literally never been worse in this sample.”

​
Where you’ll see them

  • Factors scoreboard — every cell: −3.3σ · p0
  • Volatility shelf — p62 badges ranking VIX/MOVE/SKEW against their trailing year
  • Breadth Regime — RSP/SPY at p47 of its 2-year range (>p85 EXPANSIVE, <p15 HOLLOW)
  • Market Pulse / Daily Recap — generated commentary quotes the same notation (“VIX +1.5pt but only 1Y p30 — a twitch off a calm base, not panic”)
One nuance: percentile windows differ by surface — the vol shelf ranks against the trailing 1 year, Breadth Regime against 2 years, factor cells against each series’ own full history. Hover any badge for the exact definition.