Curve Contango Index · spot → 9M+4.69/mofront-weighted roll yield · contango
Front · M1 → M2+5.93%17.90 → 18.96
Mid · M4 → M7+6.24%
Spot → M1 basis+1.75+10.9%
Curve OI · vol360.5kvol 162.2k
S&P 500 · Downside-Risk Index
0 · tested calm100 · max measured risk
NORMAL · P(−2% week) 11% · n=865
state: no flag (not all-calm)
walk-forward validated (2000–2026): predicted odds matched realized within ~1.5pp. Drawdown odds, not a sell signal — mean forward return stayed positive in every state.
VIX · Path Index
0 · VIX likely lower100 · VIX likely higher
BALANCED · P(VIX up next wk) 50.2% · n=618
state: steep contango
vol direction (unlike price direction) tested with real signal — this is the measured frequency
Tested PlaybookWhat this isEvery card is a signal TESTED on long history (cash VIX/SPX 1990–2026 (36y, 1,905 wks) · VIX3M 2009+ · VX futures 2013–2026 (13.1y, 685 wks) · cross-asset: VVIX 2006+, SKEW 1990+, MOVE 2002+, VXTLT 2004+, TLT/USO/GLD/FXE ETFs 15–24y) with the same rigor standards as the COT study: tradeable Fri→Fri windows, costs, era splits, bootstrap CIs, independent re-computation.The tiers are honestROBUST survived everything. WEAK = right direction, CI includes zero — context only.NO-EDGE = a popular belief the data does not support (shown so you don't trade it).What it is notHistorical odds, not predictions; conditions, not trade tickets. No sizing, no personalized advice. The robust findings are about RISK (when violent weeks cluster), not direction — direction was not predictable in any tested state.
tested 2026-07-04 · 14 setups · click a card for the evidence
Mixed conditionsNo elevated-risk flag is on, but the all-calm state isn’t met either.
What to do right now
Run normal-to-cautious size — no tested risk flag is on, but the all-clear isn't met either
Prefer the MILD contango zone (0.92–1.00) — the tested payer
Don't scale up because contango steepened — the steepest zone netted ≈ 0 after costs
DO NOT position for a crash off high SKEW — 36 years: no predictive power
No LONG/SHORT call on any underlying — in 36 years of testing, no state predicted direction. Anyone selling you one off these indicators is selling noise. The tested edge is in RISK and STRUCTURE, above.
Friday-close signals → next-week (Fri→Fri) outcomes; futures P&L on a 30-day constant-maturity VX position built from official settlements; 0.5%/week cost on acted weeks; by-week bootstrap CIs; pre-registered thresholds (no tuning on reported data); era splits pre-2020 vs 2020+; independently re-computed by two adversarial verification agents. Historical frequencies — not investment advice, no forward guarantee; every short-vol cell's tail (worst weeks −25% to −52%) is part of the finding, not a footnote.
Key Readings
the inputs behind the dials & playbook · hover the ⓘ
Roll yield · VIX / VIX3MWhat it is30-day VIX ÷ 3-month VIX3M.How to readBelow 1 = near-term vol cheaper than longer-dated (contango); above 1 = near-term bid (backwardation). <0.92 steep · >1.0 inverted.Used forThe term-structure carry signal; crossing above 1.0 is a classic regime-flip flag.TestedOur 13-year test (see Tested Playbook): crossing 1.0 is a ROBUST violent-week flag (4× odds), but "steeper contango = better short-vol harvest" is NOT supported — the steepest zone netted ≈0.0.848steep contango · 40th %ile<0.92 steep contango · >1.0 inverted
Variance risk premiumWhat it isVIX − RV30 — implied vol minus the realized vol that actually occurred.How to readPositive = options price more vol than has happened (a premium to harvest); near zero / negative = the cushion is gone.Used forContext on how much cushion vol sellers currently have. A thin VRP (low percentile) means you're paid little for the risk.TestedOften called the core edge in selling vol — but in our 13-year test the premium's LEVEL was not a 1-week timing signal (rich-VRP short-vol netted +0.17%/wk, dead — see Tested Playbook).−1.17VIX − RV30 · 1st %ilepositive = implied richer than realized
VIX mean-reversionWhat it isWhere today's VIX sits in its past-year range.How to readVol mean-reverts: >90th is historically stretched (tends to fall back), <10th is depressed.Tested⚠ "Cheap vol → buy protection" failed our 13-year test: long vol in the depressed state bled −1.8%/wk (t=−3.5). Cheap kept getting cheaper — see Tested Playbook.28th1-year percentile>90 stretched · <10 depressed
Front roll carryWhat it isThe front VX future's mechanical decay toward spot, in $/day per contract, if the curve holds.How to readPositive = a short front-month position collects this each day (a long one bleeds it).Used forTurns "contango" from a description into a concrete daily carry figure ($1000 per vol point per VX).+88$/day+0.088 pt/day · per VXmechanical front-month decay at this curve
Raw readings from official data. What each one is worth is on its ⓘ and in the Tested Playbook — several popular uses of these numbers tested as myths.
Term Structure Detail
as of 2026-07-02 ·
Cash Term Structure · 1 Day → 1 YearWhat this isCBOE's constant-maturity VIX family — the S&P 500's implied-vol curve from 1 day (VIX1D) out to 1 year (VIX1Y), from options prices, no futures involved.How to readUpward slope = normal (long-dated risk costs more). The FRONT inverting (VIX1D/VIX9D above VIX) = imminent-event pricing; the whole curve inverting is the tested violent-week flag (see Tested Playbook).NotesVIX1D is jumpy by construction (it collapses after each event) — read its shape, not its level.
implied vol by horizon · overlays 1W / 1M ago
VIX1D1-day13.22+0.20 (+1.5%)
VIX9D9-day12.37−0.77 (−5.9%)
VIX30-day16.15−0.44 (−2.7%)
VIX3M3-month19.04−0.12 (−0.6%)
VIX6M6-month21.50−0.13 (−0.6%)
VIX1Y1-year23.16−0.03 (−0.1%)
Today
How to use this curveSlope up (normal): longer-dated protection costs more — the market prices no imminent event. The steeper the front (VIX1D→VIX), the calmer the next few sessions are priced.Front kink up (VIX1D or VIX9D above VIX): the market is paying up for THIS week — an event is priced (CPI, FOMC, earnings). Tested: VIX9D>VIX ran 2.9× violent-week odds. Defined-risk only into it.Whole curve inverted (VIX above VIX3M): the tested 4× violent-week alarm — see the Playbook card. Compare with the 1W/1M overlays to see whether the curve is normalizing (front falling back) or building (front rising toward inversion).Level vs shape: the LEVEL says how much fear; the SHAPE says when it's priced for. A low-but-front-kinked curve is more dangerous than a high-but-upward one.