A trader argued that a falling VIX beta is no reason to be long vol, just a sign of an entrenched calm regime you pay carry to fight. Across 36 years, the data backs him up.
A volatility trader made a sharp argument on X, and it is worth testing because it cuts against a common reflex. The reflex: VIX beta is falling, vol has stopped reacting to selloffs, so the spring is coiled, get long vol. His counter: a low or falling VIX beta is not, by itself, a reason to be long vol. It just tells you the regime has stopped reacting, and you will bleed carry waiting for a vol bid that is not coming. We put it to every trading day since 1990. He is mostly right.
First, the term. VIX beta is how much the VIX moves, in points, for every 1% the S&P falls. A high beta means vol is twitchy: a small dip sends the VIX jumping. A low beta means vol has gone quiet and shrugs the same dip off. The trader's point is that a low reading describes a calm market, not a loaded one.
We ran a rolling 42-day VIX beta on every day since 1990 and sorted the days into five buckets, from the most reactive vol regimes to the least. Then we looked at what the VIX actually did next. The less vol was reacting, the less likely it was to spike.
| VIX-beta regime | Avg VIX | VIX, next 40 days (median) | Odds of a +5pt spike (40d) |
|---|---|---|---|
| Most reactive (twitchy vol) | 19.3 | -0.5 pts | 54% |
| 2 | 21.4 | -1.0 pts | 47% |
| 3 | 19.6 | -0.3 pts | 37% |
| 4 | 19.8 | -0.3 pts | 39% |
| Least reactive (low VIX beta) | 17.1 | +0.2 pts | 28% |
In the least-reactive regime, the one the trader is describing, the VIX over the next forty days went essentially nowhere, a median of two-tenths of a point, and a 5-point spike showed up just 28% of the time. That is the lowest of any bucket, against 54% when vol was at its most reactive. Buying vol because beta is low is buying the setup least likely to pay. Low beta also lines up with a low VIX (an average of 17 here), so it is, in large part, just another way of saying the market is calm.
This is the heart of his argument, and it holds. VIX beta is one of the stickiest readings in the vol market: its level today correlates 0.75 with its level a month later. If you are in the least-reactive regime now, you are still in it twenty days later 61% of the time, against 20% by chance. Calm begets calm.
It shows up in the selloffs themselves. Across the 1,117 days the S&P fell at least 1% since 1990, vol got bid almost every time, a median jump of 1.83 VIX points. It got offered, meaning the VIX fell or sat flat into the drop, on just 5% of them. And after one of those rare offered selloffs, the next pullback was offered three times as often as usual. Non-reactivity clusters. A market that has stopped flinching tends to keep not flinching.
Here is the kicker the trader leads with, and the data confirms it. In the calm, low-beta regime the VIX curve sits in steeper contango: the 3-month VIX runs about 2.1 points above spot, versus 1.6 in the reactive regimes. A long-vol position bleeds roll-down every day, and it bleeds more precisely when beta is low. So the regime is not just failing to bid vol, it is charging you rent to wait, and the two-tenths-of-a-point drift you might collect does not come close to covering it.
One line goes a step too far. He says that if vol gets offered into a 1% selloff, the next pullback will not bid vol either. The clustering is real, an offered selloff makes the next one three times more likely to be offered, but three times a 5% base rate is still only 15%. The next pullback bids vol roughly 85% of the time regardless. Non-reactivity is stickier than random; it is not the near-certainty that phrasing implies.
A low VIX beta does not warn that a spike is coming. It tells you the market has stopped flinching, and that it tends to keep not flinching. That is a reason to not be long vol, not a reason to be. Just remember that entrenched is a base rate, not a promise: a 28% spike probability is not zero, and calm regimes break violently when they finally break. Low beta gives you the odds, not the date.
Thirty-six years of daily VIX and S&P 500, testing whether a low or falling VIX beta is a reason to be long volatility.
In the least-reactive VIX-beta regime, a 5-point vol spike showed up just 28% of the time over the next 40 days, against 54% when vol was twitchy. Low beta is the worst long-vol setup.
VIX beta today correlates 0.75 with its level a month out. Least-reactive now means still least-reactive in 20 days 61% of the time, versus 20% by chance.
In the calm regime the 3-month VIX sits about 2.1 points above spot, versus 1.6 when reactive. Long vol bleeds more roll-down exactly when beta is low.
An offered selloff triples the odds the next is offered too (15% vs 5%), but vol still gets bid on the next pullback about 85% of the time. Clustering, not certainty.
Why a quiet vol gauge is not a coiled spring.
Low VIX beta describes a market that has stopped reacting, and reacting and about-to-react are not the same thing. The data says it tends to keep not reacting, not snap back.
Even a setup with a tiny forward edge loses if you pay more to hold it than it returns. In calm regimes the roll-down dwarfs the drift, so long vol bleeds while you wait for a bid that rarely comes.
Persistence tells you calm usually continues, not that it always will. Low beta sizes the odds; it does not call the turn, and the turn, when it comes, is violent.
We test popular signals the honest way: every instance counted, every result measured against a plain baseline. See what else held up, and what did not.
Browse Reality Check
Comments
Loading comments…