# Volatility Regime Shifts

> This eleven-page research report, including its colophon, compares annualized VIX-implied volatility with close-based S&P 500 realized risk over subsequent 30-calendar-day windows. It examines selected historical episodes, sample-thinning checks, and matched data through September 4, 2026, and cites four selected sources. Overlapping origins are dependent; the volatility gap is neither a trading return nor a variance risk premium.

Published: 2026-10-11
Publisher: BlackRidge (https://blckridge.com/)
Canonical: https://blckridge.com/research/volatility-regime-shifts-20261006/
PDF: https://blckridge.com/research/volatility-regime-shifts-20261006/volatility-regime-shifts-20261006-en.pdf

---

# Volatility Regime Shifts

This study compares risk implied by S&P 500 options with the index’s subsequent realized movement across matched forward horizons. It tests daily primary Cboe market series alongside selected historical episodes.

## High VIX often exceeds later realized risk, while realized risk stays elevated

Among 736 overlapping origin days with VIX ≥30, subsequent 30-day realized risk was below the starting VIX on 85.9% of days. Mean realized risk was 31.88% annualized in this sample, versus 15.30% annualized across all 9,233 complete origins. That comparison provides a quantitative baseline for high risk, not a monthly price move. [01] [02] [03]

Share of 736 overlapping VIX ≥30 origin days when subsequent 30-day realized risk was below the starting annualized VIX reading. [01] [02] [03]

Mean VIX across the high-VIX sample was 38.18% annualized; mean subsequent realized risk was 31.88%. Their 6.30 percentage-point difference is calculated across all 736 origins, including days when realized risk exceeded VIX. Realized risk remained high in absolute terms. [01] [02] [03]

Realized risk exceeded the starting VIX on 14.1% of origin days. Across all 736 origins, starting VIX and subsequent realized risk had a correlation of 0.6086. The windows overlap. These descriptive frequencies are not independent trials or forecast probabilities. [01] [02] [03]

## Realized risk surged after a low pre-selloff VIX

On 19 February 2020, VIX stood at 14.38%. Realized risk over the following 30 calendar days reached 88.13% annualized. The gap compares option-implied pricing with subsequent daily moves. It does not explain the episode or predict what another high-VIX period will deliver. [01] [02] [03]

The 88.13% figure is the annualized realized risk measured from daily closing prices over the 30 calendar days after 19 February 2020. It is calculated as 100 × √[(365/30) × Σ(daily log return)²] in that window. [02] [03]

Daily chart for January to May 2020, with a shared origin-date axis comparing VIX with realized risk over the following and prior 30 calendar days, all annualized percentages. Sources: Cboe VIX closes [01] , S&P 500 closes [02] , and VIX methodology [03] .

- Origin date
- VIX (%)
- Next 30 days (% annualized)
- Prior 30 days (% annualized)
- 2020-02-19
- 14.38
- 88.13
- 12.79
- 2020-03-16
- 82.69
- 64.03
- 82.29

On 16 March 2020, VIX reached 82.69%, after realized risk of 82.29% in the prior window. Realized risk over the following 30 calendar days was 64.03%. Higher option-implied risk coincided with substantial subsequent movement, though less than the already elevated preceding window. These dates anchor the comparison. They do not identify a macroeconomic cause. [01] [02] [03]

Realized risk is 100 × √[(365/30) × Σ(daily log return)²] in the window. Squaring captures movement in either direction, unlike net price change. Each complete horizon contains 15 to 22 daily returns. The calculation uses daily closes and differs from a sample standard deviation scaled by the square root of 252. Daily closes cannot recover intraday variation. If the horizon ends on a weekend, the calculation uses the last available close within the window. It does not interpolate the unobserved interval. [02] [03]

## Higher starting VIX accompanies higher subsequent realized risk

Across six ordered starting-VIX bins, mean subsequent realized risk in the highest bin is 4.45 times the lowest-bin mean. Higher implied readings accompany higher future risk in the full sample. The daily windows overlap, so these historical frequencies are not forecast probabilities. [01] [02] [03]

Mean future realized risk in the highest starting-VIX bin divided by the mean in the lowest bin.

The two mean series show bin means across six ordered starting-VIX bins, not a time trajectory or continuous fitted prediction. The last table column is the within-bin share with future realized risk above initial VIX. Realized risk = 100*sqrt((365/30)*sum(log(P_i/P_(i-1))^2)), using daily closes in (t,t+30 days], without demeaning. Origins: Jan 2, 1990-Sep 4, 2026. Sources: [01] [02] [03] .

- Starting VIX bin
- Windows
- Mean future realized risk (%)
- Share with future risk above initial VIX (%)
- <15
- 2958
- 9.94
- 13.3
- 15–20
- 2851
- 13.16
- 12.7
- 20–25
- 1824
- 17.63
- 15.7
- 25–30
- 21.60
- 15.3
- 30–40
- 27.01
- 11.4
- ≥40
- 44.25
- 21.2

The bins are below 15, 15-20, 20-25, 25-30, 30-40, and 40 or higher. Mean subsequent realized risk rises from 9.9% in the first bin to 44.3% in the last. Each annualized estimate uses daily S&P 500 closes over the next 30 calendar days, with 15-22 returns per complete window. [01] [02] [03]

The full-sample Pearson correlation is 0.72. Higher VIX readings accompany higher subsequent realized risk across daily origins, but that association does not calibrate the next window. The difference between starting implied volatility and later realized risk is measured in volatility points. It is not an ex ante variance risk premium or an option-trading return. [01] [02] [03] [04]

## The implied-realized gap stayed positive across three historical blocks

Across all three historical blocks, the average VIX reading exceeded subsequent 30-day realized volatility. The gap narrowed in the middle block, then widened in the latest block, but did not return to its earlier level. These are descriptive averages, not evidence of a newly identified regime. [01] [02] [03]

Average annualized implied volatility minus subsequent realized volatility, in volatility points, for the latest historical block.

Annual origin-day means, 1990 to 2026. 2026 ends Sep 4, with realized windows through Oct 4. Realized risk: 100×√((365/30)Σ[ln(Pᵢ/Pᵢ₋₁)]²), daily closes in the next 30 calendar days, no demeaning. VIX is annualized percent. Sources: Cboe [01] [02] [03] .

- Interval
- Origin days
- Mean VIX (%)
- Mean subsequent realized volatility (%)
- Mean gap (volatility points)
- 1990-2006
- 4283
- 19.05
- 14.22
- 4.83
- 2007-2019
- 3272
- 19.26
- 15.88
- 3.39
- 2020-2026 partial (through Sep 4)
- 1678
- 20.75
- 16.89
- 3.86

The mean gap was 4.83 volatility points in 1990-2006, 3.39 in 2007-2019, and 3.86 in 2020-2026 through September 4. The latest block therefore shows a positive average difference, with the gap still below the first block's. Mean VIX rose from 19.1% in the first block to 20.8% in the latest; mean subsequent realized volatility also rose, from 14.2% to 16.9%. [01] [02] [03]

Higher VIX readings were associated with higher subsequent realized risk. The correlation between VIX and subsequent realized risk was positive in every block, though weaker in the latest. Most origin days share part of the forward window, so counts describe observations, not independent trials or forecast odds. Block averages cannot establish a regime break, statistical significance, or a return to the earlier gap. [01] [02] [04]

## VIX exceeded subsequent realized risk in the August 2024 window

One selected episode shows a 22.22-point gap between the VIX reading and subsequent close-based realized risk. The historical record also contains the opposite outcome, when realized risk rose well above the initial VIX level. These are descriptive windows, not event estimates or forecast probabilities. [01] [02] [03]

Difference in volatility points between option-implied VIX and close-based realized risk over the next 30 calendar days.

Daily S&P 500 cumulative price change (%) from the 1 July 2024 close through 4 September 2024, plotted separately from annualized risk measures. Price index closes [02] .

- Origin date
- VIX (%)
- Subsequent realized risk, annualized (%)
- Origin-to-end net index change (%)
- 2008-09-15
- 31.70
- 81.00
- -23.88
- 2020-02-19
- 14.38
- 88.13
- -31.93
- 2022-06-13
- 34.02
- 22.04
- 1.39
- 2024-08-05
- 38.57
- 16.35
- 6.43

At the 5 August 2024 close, VIX stood at 38.57%; the following 30-calendar-day window produced annualized realized risk of 16.35%, a 22.22-point gap. The table reports a +6.43% net endpoint price return for the same episode. Daily squared returns capture movement along the path; the final index level records only net direction and distance. [01] [02] [03]

The 2022 anchor shows 22.04% annualized realized risk alongside a +1.39% net endpoint price return. Annualized realized risk and cumulative endpoint price return have different units and cannot be compared directly. [01] [02] [03] The 2008 episode is the counterpoint: on 15 September, VIX at 31.70% was followed by 81.00% realized risk. In the selected 2020 window, VIX was 14.38% and subsequent realized risk 88.13%. The dates mark historical anchors, not identified causes; daily-origin windows overlap, so their frequencies are not independent trials. This gap is descriptive, not an ex-ante variance risk premium or an option-trading return. [01] [02] [03] [04]

## The below-VIX frequency remains high after sample thinning

Across daily 30-day windows, subsequent realized volatility finished below the VIX-implied measure in 86.2% of cases. The share was 85.0% after thinning to non-overlapping windows, but that check does not make market episodes independent or settle how much the starting anchor matters. [01] [02] [03]

This high-VIX non-overlapping subset is small, and thinning depends on the chosen starting anchor.

- Sample
- Windows (n)
- Realized below VIX (%)
- Mean realized volatility, annualized (%)
- All windows, daily origins
- 9233
- 86.2
- 15.30
- All windows, non-overlapping
- 85.0
- 15.27
- High VIX (30 or higher), daily origins
- 85.9
- 31.88
- High VIX (30 or higher), non-overlapping
- 91.2
- 32.63

VIX is an option-price measure of expected volatility over the next 30 calendar days, annualized. Realized volatility here comes from squared daily log returns within that horizon, also annualized; it is not the net price move. Inputs are Cboe VIX and S&P 500 daily closes. [01] [02] [03]

Higher VIX readings are associated with higher subsequent realized risk: mean realized volatility was 31.88% annualized in high-VIX windows, versus 15.30% across all windows. Still, realized volatility fell below the VIX reading in 85.9% of high-VIX daily windows and 91.2% of high-VIX non-overlapping windows. The latter rests on 34 windows, spanning 1990-08-06 to 2025-04-07; greedy thinning starts 1990-01-02. Overlap and clustered high readings remain, and the result may depend on that anchor. These are descriptive sample frequencies, not forecast odds. [01] [02] [03]

## Recent VIX averaged above subsequent realized risk

Mean VIX was 18.77% in the recent sample; realized risk over the following 30 days averaged 14.55%. Higher VIX generally preceded higher realized risk. The average difference summarizes this sample, not a forecast for any particular date. [01] [02] [03]

Mean percentage-point difference between annualized VIX and close-based realized risk over the following 30-calendar-day windows, Jan 2025 to Sep 2026.

The chart plots three monthly mean series from Jan 2025 to Sep 2026: originating VIX and following 30-day realized risk, both annualized percentages, plus their gap in percentage points. September is partial. Cboe daily series and method [01] [02] [03] .

- Origin
- VIX, annualized (%)
- Following 30-day realized risk, annualized (%)
- Window end
- 2026-09-04
- 14.53
- 9.80
- 2026-10-04
- 2026-10-05
- 15.52
- Outcome pending
- ∅

The 420 daily origins overlap. Each day's VIX is compared with realized risk over the next 30 calendar days, so nearby observations share much of the same price path. The mean gap was 4.22 percentage points; the median was 5.73. September 2026 is partial in the monthly chart. Frequencies describe this sample, not independent trials or forecast odds. [01] [02] [03]

The relationship was positive, with a correlation of 0.46. Higher VIX tended to precede higher realized risk, although realized risk fell below VIX in 89.0% of these windows. This gap is not an ex ante variance risk premium or an option-trading return. Zhou treats variance risk premia as a separate measure. [01] [02] [04]

## When VIX Is High, Realized Risk Can Still Exceed It

VIX offers an imperfect signal of the month ahead: higher readings are associated with higher subsequent realized risk. In 208 overlapping 30-day windows beginning at VIX ≥40, annualized realized volatility exceeded VIX in 21.2%. This is a descriptive share of overlapping observations, not an independent-event probability or a forecast. [01] [02] [03]

21.2% of 208 overlapping 30-day windows at VIX ≥40. The share is descriptive; overlapping windows are not independent observations.

- VIX threshold (annualized implied volatility, %)
- Complete origins (count)
- Future realized volatility above VIX (%)
- Mean future realized volatility (annualized, %)
- ≥20
- 3424
- 15.3
- 21.69
- ≥25
- 1600
- 14.8
- 26.33
- ≥30
- 14.1
- 31.88
- ≥40
- 21.2
- 44.25

At VIX thresholds 20, 25, 30, and 40, realized volatility exceeded VIX in 15.3%, 14.8%, 14.1%, and 21.2% of origins. Rates varied non-monotonically. At ≥40, mean realized volatility was 44.25% (n=208). Among 736 VIX ≥30 origins, 104 showed realized volatility above VIX. The largest realized-minus-VIX gap there was 62.23 volatility points (VIX 31.99, realized 94.22, 4 March–3 April 2020). No structural-regime forecast follows. [01] [02] [03]

Risk-neutral pricing cannot establish causation. This percentage-point risk difference is not a return, trade profit, or an ex ante variance risk premium (VIX squared versus expected physical variance). Zhou provides context, not calculation data. Trade assessment requires contract terms, rebalancing, intraday data, margin, liquidity, and fees. [01] [03] [04]

## Evidence base: 9,233 complete windows

Matched daily data cover 9,233 origins from January 1990 through September 4, 2026, pairing Cboe VIX closes with S&P 500 price-index closes to compare implied risk with subsequent realized risk over 30 days [01] [02] .

The Cboe CSV identifies daily VIX observations and contains DATE, OPEN, HIGH, LOW, and CLOSE fields. CLOSE is the listed closing-value field.

The Cboe CSV identifies S&P 500 index observations and contains DATE and SPX fields. DATE is the date field, and SPX contains index values. The header has no CLOSE field.

Cboe’s methodology document is titled “Cboe Volatility Index Methodology” and lists Introduction, Index Objective, Supporting Documentation, and Index Construction.

Hao Zhou’s working paper examines variance risk premia, asset-predictability puzzles, and macroeconomic uncertainty. It offers conceptual context for research on the relationship between option-implied risk measures and economic uncertainty. The paper is cited for its subject and framing, not as a source for this report’s calculations, market observations, or Federal Reserve Board policy.

### Method

Realized risk for each origin t is 100 × √((365/30) × Σ[ln(Pᵢ/Pᵢ₋₁)]²), using daily S&P 500 closes in (t, t+30 calendar days] without demeaning; complete windows contain 15–22 returns [02] . VIX is the annualized option-implied measure for a 30-calendar-day horizon [01] [03] . Weekend horizons use the last available close; moves until the next close are omitted. Origins overlap, so frequencies are descriptive, not independent trials. The provider’s current history is not vintage data; closes omit intraday paths and transaction costs. This volatility gap is not a variance risk premium, which compares implied variance with expected physical variance [04] .

## Keep up with the research.

Choose optional research updates, or learn how the account structure works.

## Independent research, introductions clearly defined.

BlackRidge is an independent research bureau introducing private investors to quantitative traders through multiple strategy providers. We publish quantitative research. Investors access strategies through PAMM accounts at the broker. We are not a fund or broker and never hold client money.

Your funds stay at the broker. The account is opened in your own name. BlackRidge never receives, holds, or has withdrawal rights over client capital.

No fee on your profits. A one-time access payment of 6.7% of the agreed trading level, paid to the selected strategy provider. No management fee, no performance fee, no profit share, in any year.

You fund the risk, not the exposure. Allocations are notionally funded: you agree a trading level and fund the margin and drawdown allowance behind it. Trading losses can exceed the deposit without applicable negative balance protection; the separate 6.7% access payment is nonrefundable.

blckridge.com/research : published research. Notional funding and PAMM accounts explained in full, and answers to the questions this raises.

We use AI models to gather and aggregate source material and to help prepare each report. Read it with its cited sources, sample, methods and limitations, and send corrections through research methodology and corrections .

The client pays the selected strategy provider a one-time access payment of 6.7% of the agreed trading level. BlackRidge receives an introduction fee from that provider and does not receive broker compensation. About BlackRidge .

Minimum funded capital $25,000. The strategy, its live records, the broker, the selected strategy provider and the full commercial terms are presented on an introductory call and confirmed in writing before any payment or deposit.

This report is published for information only. It is not investment advice and does not take account of your circumstances. Trading leveraged instruments including CFDs carries substantial risk and is not suitable for all investors; you may lose the capital you fund and, depending on your broker's terms, may owe more than your deposit. Findings in this report may combine third-party evidence, historical calculations and illustrative scenarios; they are not verified live trading results of any strategy provider. Past performance does not indicate future results.

## Sources

- [01] [Cboe VIX Daily Price History](https://cdn.cboe.com/api/global/us_indices/daily_prices/VIX_History.csv). The Cboe CSV identifies daily VIX observations and contains DATE, OPEN, HIGH, LOW, and CLOSE fields. CLOSE is the listed closing-value field.
- [02] [Cboe S&P 500 Daily Price History](https://cdn.cboe.com/api/global/us_indices/daily_prices/SPX_History.csv). The Cboe CSV identifies S&P 500 index observations and contains DATE and SPX fields. DATE is the date field, and SPX contains index values. The header has no CLOSE field.
- [03] [Cboe Volatility Index Methodology (Readable text)](https://r.jina.ai/https://cdn.cboe.com/api/global/us_indices/governance/Volatility_Index_Methodology_Cboe_Volatility_Index.pdf). Also: [Original PDF](https://cdn.cboe.com/api/global/us_indices/governance/Volatility_Index_Methodology_Cboe_Volatility_Index.pdf). Cboe’s methodology document is titled “Cboe Volatility Index Methodology” and lists Introduction, Index Objective, Supporting Documentation, and Index Construction.
- [04] [Hao Zhou, Variance Risk Premia, Asset Predictability Puzzles, and Macroeconomic Uncertainty (FEDS 2010-14)](https://www.federalreserve.gov/pubs/feds/2010/201014/index.html). Hao Zhou’s working paper examines variance risk premia, asset-predictability puzzles, and macroeconomic uncertainty. It offers conceptual context for research on the relationship between option-implied risk measures and economic uncertainty. The paper is cited for its subject and framing, not as a source for this report’s calculations, market observations, or Federal Reserve Board policy.

---

## Citation context

In 736 overlapping origins with VIX at or above 30, subsequent 30-day realized risk was below starting VIX on 85.9% of days. Mean realized risk in this high-VIX sample was 31.88% annualized, compared with a mean starting VIX of 38.18%. These describe option-implied volatility and later realized risk, not returns.

The report matches daily Cboe VIX closes with S&P 500 index closes across 9,233 complete origins from January 2, 1990, through September 4, 2026. It measures subsequent realized risk from squared daily log returns over the next 30 calendar days and annualizes the result.

Daily forward windows overlap, so observations are dependent and reported frequencies are not independent trials or forecast probabilities. Provider history is not point-in-time vintage data; daily closes omit intraday movement. The difference between implied volatility and later realized risk is not a trading return or an ex ante variance risk premium.

Primary input: [Cboe VIX daily history](https://cdn.cboe.com/api/global/us_indices/daily_prices/VIX_History.csv).

Stable permalink: [https://blckridge.com/research/volatility-regime-shifts-20261006/#citation-context](https://blckridge.com/research/volatility-regime-shifts-20261006/#citation-context).

## Read next

- [Historical Covariance Forecasts Underestimate Stress Risk by a Factor of 1.6](/research/covariance-estimation-error-20260929/). It complements the separate covariance-estimation-error report, which tests historical covariance forecasts during market stress, while this report compares option-implied volatility with subsequent realized index risk.