The Surface Beneath the Fear Gauge

Take every session since 2011 on which the VIX closed between 14 and 16, a range any desk would describe as quiet. There are 709 of them, about 18 per cent of the sample. Across that set, the spread between three-month and nine-day SPX implied volatility ranges from minus 4.6 to plus 9.1 index points, and the Cboe SKEW index ranges from 112 to 183. Knowing the VIX to within two points leaves both of those almost entirely undetermined.

Figure 1 is that result. It is a claim about information rather than about the benchmark's construction, which is careful and does exactly what its methodology says.

Section 01What the gauge is, exactly

The VIX aggregates weighted prices from a broad strip of out-of-the-money SPX puts and calls around two maturities and interpolates to a constant 30-day horizon. It is computed without first choosing an option-pricing model, and it is an annualised measure derived from risk-neutral expected variance rather than a physical forecast or an at-the-money implied volatility.

Its limitation is not that it ignores the wings. It uses them. The limitation is that aggregation is not invertible: two surfaces with different skew, different term structure, or different tail prices can integrate to the same scalar. That is a property of any summary statistic, and it would be unremarkable except that the residual variation turns out to be large.

Section 02How much the scalar leaves open

The measurement is straightforward. Cboe publishes VIX9D, VIX, VIX3M and SKEW as separate official indices computed from the same option strips. Conditioning on the VIX in narrow buckets and asking how much the others still vary gives a direct answer.

Across two-point VIX buckets, the median within-bucket standard deviation of the three-month minus nine-day spread is 3.26 index points, against an unconditional standard deviation of 3.72. Fixing the VIX to within two points removes about 12 per cent of the variation in the term structure. For SKEW the numbers are 10.80 against 11.92, a reduction of about 9 per cent. Ninety per cent of the shape survives knowing the level.

The term structure is not a small residual either. Three-month implied volatility sits below nine-day on 13.4 per cent of sessions in the sample. An inverted curve and an upward-sloping one are different states for anyone holding calendar exposure, and both are compatible with a VIX of 15.

Section 03Why the level can mislead

Index volatility is a derived quantity, built from constituent volatilities and the correlations between them, so a low index reading can mean genuinely subdued single names or lively single names that are not moving together. The second state is more fragile than the first, because a correlation spike converts it into a large index move without any individual stock doing anything new. Neither the VIX nor this article measures that decomposition, which needs single-name option data.

for each 2-point bucket b of VIX:
    sd_b( VIX3M - VIX9D )        within-bucket dispersion of the term structure
    sd_b( SKEW )                 within-bucket dispersion of the strike shape

report median over buckets, against the unconditional sd

measured, daily closes, Jan 2011 to Sep 2026:

    VIX3M - VIX9D    within-bucket sd  3.26    unconditional  3.72    -12%
    SKEW             within-bucket sd 10.80    unconditional 11.92     -9%

    VIX in [14, 16]:  n = 709 sessions   (18.0% of the sample)
        VIX3M - VIX9D   ranges  -4.6  to  +9.1
        SKEW            ranges   112  to   183

index_var  ~  sum_i w_i^2 var_i  +  sum_{i!=j} w_i w_j corr_ij sqrt(var_i var_j)

Spring 2025 made the point in fast motion. After the 2 April tariff announcement the VIX rose from 21.51 to a peak of 52.33 on 8 April and was back at 24.70 by the end of the month. Read only as a level, the gauge said panic and then said calm. Read as a surface, the same weeks were a violent repricing of skew and short-dated volatility followed by an equally violent normalisation. A book flat to the headline number throughout could still have been badly exposed to the skew that steepened or the term structure that inverted.

The level is the part that reaches the headline. Nine-tenths of the shape is in the part that does not.

The benchmark is most useful when its compression is treated explicitly. VIX answers a precise question about 30-day SPX variance under option prices, and it answers it well. Skew asks where tail protection is expensive. Term structure asks when uncertainty is concentrated. Dispersion asks whether index variance comes from constituent volatility or from correlation. The measurement here is only that the first answer constrains the second and third far less than their common treatment as one topic would suggest. Issue 015 takes the maturity axis further.

This research is analysis and commentary for general information. It is not investment advice, an offer, or a solicitation, and it contains no price forecasts. Statistics in the text are the author's own calculations from the official index values cited.