Trading Nation

A new method for predicting volatility ‘tsunamis’

A new method for predicting VIX spikes

Is volatility about to spike? It's a question frequently asked by anxious investors in times of relative calm.

Now there's a new answer.

In a recent paper entitled "Forecasting a Volatility Tsunami" that was awarded the 2017 Charles H. Dow Award from the Market Technicians Association, Andrew Thrasher argues that it's not low volatility that presages a volatility spike, but a consistent level of volatility.

Thrasher, a portfolio manager at the Indianapolis-based Financial Enhancement Group, explained in a interview Monday on CNBC's "Trading Nation" that while "it's commonly reported that whenever the volatility index gets to a really low level like we've seen this year, it will often mean revert and spike higher ... I thought that seemed way too easy."

"So I began to look at the market and look for times of commonality in the VIX before these spikes, and what I found is that when the volatility index contracts, when it has a very narrow range, that is often an environment that predicts large spikes in the 'fear index,'" he added.

Whether the CBOE volatility index is set to rise can be an important question for traders and investors focused on short term investments. The VIX uses the prices of S&P 500 options to predict the magnitude of market moves over the next 30 sessions. Since options are more commonly used to hedge against downside than to speculate on upside, the VIX tends to spike as the market is dropping. Avoiding stocks when the VIX is about to spike, then, could prove highly beneficial.

When it comes to predicting these spikes, Thrasher's specific finding is that when the VIX is at a low level, its average peak over the next three weeks is a 24.9 percent rise, while its average trough is a 9.3 percent fall. Meanwhile, when the VIX is trading in a narrow range, its average peak represents a 34.3 percent rise, while its average trough is a 8.7 percent fall. That is to say, the VIX tends to spike higher and fall less after it has been trading in a narrow range than after it has been trading at a low level.

Adding to this analysis, Thrasher also examines the VVIX, which is the volatility index of the volatility index (and so measures the expected magnitude of future moves in the VIX). He notes in his paper that "it appears the combination of the VIX and VVIX are timelier in their production of a signal before spikes within the Volatility Index."

Some caveats may be in order. The sheer volume of market data frequently makes it possible to find spurious signals purely by accident. This is particularly true when second-order data, such as dispersion levels, are added. In general, financial theory would suggest that decisions made purely on public data about price moves should not reliably lead to outperformance.

And ironically, Thrasher's indicator actually threw off a false signal just this year.

Over the past three months, the VIX saw "very narrow dispersion," which was "actually one of the longest periods of seeing such a low trading range in several years," Thrasher wrote to CNBC. "However the recent rise in the VIX in April has ended the streak and we are now no longer seeing that environment that precedes spikes."

Still, those who are anxious about a potential "tsunami" might indeed do well to heed Thrasher's advice and pay more attention to the volatility of volatility rather than the VIX's absolute level.