We set out to test whether a beaten-down, oversold VIX warns of a coming spike. In 22 years its RSI dropped below 30 only five times, and below 25 never. The reason is the most important thing about volatility.
Here is a tidy contrarian idea. VIX is the most mean-reverting number in markets, so when it gets beaten down and complacent, its RSI should go oversold, and that should be your signal that volatility is too cheap and a spike is due. Buy fear when nobody wants it. We went to backtest it, and hit a wall on the very first step.
The setup barely exists. Over 22 years, about 5,600 trading days, VIX's 14-day RSI closed below the classic oversold line of 30 on exactly five days, one tenth of one percent of the time. Below 25? Never. Below 20? Never. There is nothing to backtest, because the trigger almost never fires, and when it does there are far too few cases to learn anything from. The interesting question is why.
To see how strange this is, put VIX next to a normal asset. The S&P 500's RSI dips into oversold regularly. VIX's barely ever does.
| VIX | S&P 500 | |
|---|---|---|
| Days RSI was oversold (below 30) | 0.1% | 1.7% |
| Days RSI was below 35 | 1.4% | 5.1% |
| Lowest RSI in 22 years | 28 | 16 |
| Median RSI | 48 | 56 |
The S&P spends real time oversold and has driven its RSI as low as 16. VIX's RSI, across the entire 22 years, never once got below 28. Its whole range lives between roughly 34 and 65, with the floor pressed right up against the oversold line and never breaking through. This is not RSI being broken. It is VIX being VIX.
RSI measures the balance between up-moves and down-moves. For it to read oversold, an asset has to fall hard and persistently for a couple of weeks, so the down-days swamp the up-days.
VIX does not fall like that. It does the opposite. Volatility spikes up violently and grinds down slowly. Its declines are gentle leaks, a string of small down-days sprinkled with small up-ticks, so the 14-day balance never tips far enough negative to register as oversold. The only hard, fast moves VIX makes are upward, and those push RSI toward overbought, not oversold. The very thing that makes VIX VIX, fast up and slow down, truncates the bottom of its RSI. Fear does not capitulate. It just slowly drains, and a momentum oscillator has nothing sharp to grab onto on the way down.
You might expect the most complacent setups, the record-low-VIX run-ups before the big spikes, to be where an oversold reading finally shows up. They are not. Before Volmageddon in early 2018, with VIX dropping into single digits (as low as 9), its RSI bottomed at 43, nowhere near oversold. Before the COVID crash it bottomed at 45, and before the August 2024 unwind at 47. The most complacent tape in two decades never printed an oversold RSI, because complacency in VIX is a slow drift, not the sharp fall the indicator is built to detect.
And the handful of times VIX's RSI did dip oversold, it was not from a complacent low at all. Those readings came with VIX around 19, on the way down from a spike, as fear deflated. So even the rare signal points the wrong way: it fires as a scare ends, not as one builds.
We came to test a contrarian signal and found there was almost nothing to test, for a reason that says a lot about VIX.
VIX's RSI closed below 30 on just 0.1% of days, and below 25 never. There is no sample to build a strategy on.
VIX's RSI never once dropped below 28 in 22 years. The S&P's reached 16. The bottom of VIX's range is pinned at the oversold line.
RSI needs a hard, persistent fall to read oversold. VIX spikes up and leaks down, so its declines never swamp its rises. The asymmetry truncates the low tail.
At maximum complacency before 2018, with VIX near 9, the RSI still sat at 43. The most oversold-looking tape imaginable never printed oversold.
Sometimes the honest answer is that the question cannot be asked.
Before chasing the edge in a pattern, check how often it actually happens. "Buy oversold VIX" sounds great and triggers five times in 22 years. A rule that almost never fires is not a strategy, it is a daydream.
Oscillators like RSI are built for things that rise and fall in roughly similar ways. VIX rises and falls in completely different ways, so an indicator that treats up and down as mirror images mismeasures it. The tool does not fit the series.
The contrarian dream is to buy fear at its washed-out low. But fear has no washed-out low. It drains gradually and then explodes, which is exactly why a momentum oscillator cannot time the bottom of it. There is no capitulation in calm.
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…