How to Read the Volatility Regime
Markets do not move with one constant level of risk. They spend long stretches in calm conditions, pass through choppy transitions and occasionally fall into short, violent stress episodes. The Vol Regime page labels every S&P 500 session since 2007 as Low, Mid or High Vol, so you can see which of those states the market is in today and how it has behaved in each one.
1. What a volatility regime is
Volatility clusters. A large daily move tends to be followed by more large moves, and a quiet day by more quiet days. This is one of the most reliable patterns in financial markets, and it means the recent past says a lot about the near future level of risk, even though it says little about direction.
A volatility regime is a persistent state in which volatility, the behavior of daily returns and the pricing of options all look alike. Instead of reading a single number such as the VIX on its own, a regime model asks a broader question: which of a few typical market states does today most resemble?
2. How the regimes are estimated
The regimes come from a proprietary unsupervised machine-learning model developed by QuantDaddy. Its exact specification and parameters are not published, but the approach can be described in four steps:
- Inputs. Each session is described by indicators derived from historical S&P 500 implied volatility surface data, together with index price history. The surface matters because option prices show how much risk the market is paying to protect against, not only how much it has already experienced.
- Learning the states. The model is not told in advance what calm or stressed markets look like. It finds three recurring states in the data on its own, and they are then ordered from lowest to highest volatility and named Low, Mid and High Vol.
- Walk-forward, no look-ahead. The model is trained on earlier history and rolled forward one session at a time from 2007, with periodic re-training as new data arrives. Each day is classified only with information that was available on that day, so the history on the page is out of sample: it shows the labels the model would have produced at the time, not labels fitted with hindsight.
- Daily output. For every session the model gives a probability for each of the three regimes. The regime shown is the one with the highest probability.
Why "unsupervised"? There is no official list of calm and stressed days to learn from. Letting the model discover the states avoids imposing arbitrary thresholds such as "VIX above 20 means stress", which would ignore how the whole volatility surface behaves.
3. The three regimes in numbers
The Regime Statistics panel summarizes each regime over the full out-of-sample history. As of the October 28, 2026 update (January 2007 to October 2026, 5,020 sessions), the figures were:
| Low Vol | Mid Vol | High Vol | |
|---|---|---|---|
| Share of sessions | 55.5% | 35.9% | 8.6% |
| Average length of a spell | 6.8 sessions | 3.3 sessions | 2.5 sessions |
| Average VIX | 15.1 | 22.8 | 36.7 |
| Average realized volatility | 8.2% | 19.2% | 42.8% |
| S&P 500 annualized return while in the regime | +24.7% | −26.0% | +48.1% |
| S&P 500 annualized volatility while in the regime | 7.2% | 21.0% | 47.5% |
Three things stand out:
- Volatility rises in clear steps. Realized volatility roughly doubles from Low to Mid Vol and doubles again into High Vol. The regimes really are different worlds for anyone who holds or prices options.
- Calm is the normal state. More than half of all sessions are Low Vol, and Low Vol spells last about twice as long as Mid Vol spells. High Vol is rare and short.
- Returns are not what you might expect. The index lost ground on average during Mid Vol sessions and gained during High Vol sessions. Stress episodes contain both the worst sell-offs and the sharpest rebounds, and the rebounds often happen while volatility is still extreme. With fewer than 9% of sessions in High Vol, these averages are dominated by a handful of very large days. They describe what happened, not a reason to buy into stress.
The statistics are recalculated with every update, so the live page may show slightly different values from the table above.
What happened after a regime began
Averages over whole regimes mix very different days. A more practical question is what the index did after a regime started. Since 2007 there have been 1,123 regime starts: 410 into Low Vol, 542 into Mid Vol and 171 into High Vol. For each start, the chart below measures the S&P 500's price return from the close of that first session over the next 5 sessions (about a week) and the next 20 sessions (about a month).
Two things stand out:
- The typical outcome barely depends on the regime. The median 5-session return was +0.5% after Low Vol starts, +0.5% after Mid Vol starts and +0.4% after High Vol starts, and the index was higher after five sessions about 60% of the time in every regime. Over 20 sessions the medians were +1.4%, +1.6% and +0.7%.
- The range of outcomes depends on it a lot. Over 20 sessions the middle half of outcomes spanned about 4 points after Low Vol starts (−0.9% to +3.2%), 5 points after Mid Vol starts (−1.4% to +3.5%) and 8 points after High Vol starts (−3.6% to +4.4%). The bad tail grew too: the 5th percentile was −7.3%, −8.3% and −11.5%.
In other words, the regime tells you how wide the distribution of the next few weeks is likely to be, not which way it leans. That is what a volatility measure should do, and it is why the regime is more useful for sizing risk and pricing options than for picking direction.
4. Reading the probabilities
The header shows three probabilities that add up to 100%. They are more informative than the label alone:
- One probability near 100%. The session clearly resembles one state. On October 28, 2026, for example, the model gave Low Vol 97%, Mid Vol 3% and High Vol 0%, the fourth Low Vol session in a row.
- Two probabilities close to each other. The market sits near the boundary between two states. The label can flip back and forth from day to day, and that flickering is itself a sign of a transition period.
- A rising probability before the label changes. A regime change usually shows up first as a growing share for the next state in the Regime Probabilities chart, a few sessions before the most likely regime switches.
The page is updated once a day after the US close. Before that update has run, the latest session shows a preliminary estimate, which is replaced by the final value after the daily run.
5. Transitions: how regimes change
The transition matrix shows how often a session in one regime was followed by a session in each regime the next day. Read each row from the regime you are in today:
| Today ↓ / Tomorrow → | Low Vol | Mid Vol | High Vol |
|---|---|---|---|
| Low Vol | 85.3% | 14.1% | 0.6% |
| Mid Vol | 21.5% | 69.9% | 8.6% |
| High Vol | 5.3% | 34.2% | 60.5% |
The diagonal is the chance of staying put, and it explains the spell lengths above: a regime that persists with probability p lasts on average about 1 ÷ (1 − p) sessions, which gives roughly 7 sessions for Low Vol, 3 for Mid Vol and 2.5 for High Vol.
The off-diagonal entries show the usual path. A jump straight from Low to High Vol is rare (0.6% of Low Vol sessions); stress usually builds through Mid Vol first. On the way out, High Vol hands over mostly to Mid Vol rather than straight back to calm. Treat these as historical frequencies, not odds for tomorrow: in a fast crash the market can move from Low to High Vol within a session or two.
6. Using the regime with other tools
The regime is most useful as a backdrop for the other charts on this site:
- With GEX. Dealer gamma describes how hedging flows may dampen or amplify moves; the regime describes how volatile the market already is. Reading both tells you whether a negative gamma reading arrives in an otherwise calm market or on top of existing stress. See How to Read Gamma Exposure.
- With the VIX term structure. Over the year to October 2026, the front of the VIX futures curve was in contango on almost every Low Vol session (the second month typically about 7% above the first), while about a third of Mid Vol sessions had an inverted front. High Vol usually comes with a flat or inverted curve. When the two disagree, for example a Low Vol label with a curve that has started to flatten, it is worth watching the probabilities closely. See How to Read the VIX Term Structure.
- With the volatility surface. Skew and the term structure of implied volatility can start to change before realized volatility does. See How to Read the Implied Volatility Surface.
- With Vol Sentiment. The regime describes US equities only. Vol Sentiment shows whether calm or stress is shared across currencies, rates, credit and commodities.
7. Reading the Vol Regime page
- Header shows the regime badge, the three probabilities, how many sessions the current regime has lasted, the latest VIX and the model's short-window realized volatility, and the date of the latest session.
- S&P 500 by Regime colors each session of the index green (Low Vol), amber (Mid Vol) or red (High Vol), so calm uptrends, choppy transitions and stress episodes are visible at a glance. Use the Range buttons to change the period.
- Regime Probabilities stacks the daily probabilities. This is where early signs of a transition appear.
- Regime Map plots VIX against realized volatility, one dot per session, colored by regime. The three regimes occupy different parts of the map, and the ringed dot is the latest session. A dot sitting between clusters is a borderline day.
- Regime Statistics shows the table and the transition matrix described above, calculated over the full out-of-sample history.
8. Limitations you should know
- It describes, it does not forecast. The regime tells you the current volatility state, not the direction of the next move. Persistence makes today's regime a reasonable guide to tomorrow's, but transitions into High Vol can be abrupt.
- Three states are a simplification. Real markets move along a continuum. Sessions near a boundary can be labeled differently by a small change in inputs.
- Re-training shifts boundaries slightly. The model is re-trained periodically as new data arrives, which can move the borderline between two regimes. Past labels are kept as they were produced at the time.
- One market only. The model covers the S&P 500. Stress that starts elsewhere, in rates or credit for example, may not show up until it reaches US equities.
- Statistics come from one history. Average returns by regime depend heavily on a few extreme episodes such as 2008 and 2020 and would look different over another period.
Used with those caveats in mind, the regime is best treated as a label for the market's current risk environment, a quick way to put every other number on this site in context.
Continue with How to Read Cross-Asset Vol Sentiment or How to Read Gamma Exposure (GEX).