Blog - Investing Notes

November 11, 2008 - Update: VIX and Future Stock Market Returns

Experts and pundits often cite a very high (low) Chicago Board Options Exchange (CBOE) Volatility Index (VIX) as an indication of investor panic (complacency), and therefore of a pending U.S. stock market advance (decline). Is this conventional wisdom regarding VIX as a market reversal sentiment indicator accurate? To check, we relate the level of VIX to future stock market returns (using the S&P 500 index to represent the overall stock market) over horizons of 10, 21, 63 and 252 trading days. Because VIX tends to "stick" in high and low regimes over extended periods, we also relate the level of VIX relative to its 63-trading day moving average to future stock market returns. Using daily closes for VIX and for the S&P 500 index over the period January 1990 through October 2008, we find that:

The following chart compares the 63-day (three-month) moving average of VIX and the level of the S&P 500 index over the entire sample period. It appears that there is sometimes an inverse relationship between the two series, and sometimes not. It is not obvious that the level of the VIX moving average predicts stock returns.

For greater precision, we apply some statistical tests.

Over the entire sample period, the Pearson correlation between VIX and S&P 500 index returns over the next 10 (21 / 63 / 252) trading days is 0.02 (0.04 / 0.09 / -0.03). These very low correlations suggest no relationship between the level of VIX and future stock market returns.

Because of the apparent regime-like variation of VIX, we also relate extreme values of VIX relative to its 63-day moving average to future stock returns. The following chart summarizes average S&P 500 index returns over the 10, 21, 63 and 252 trading days after:

• All days in the sample.
• Days when VIX is less than 80% of its 63-trading day moving average (ranging from 221 instances for 252-day returns to 242 instances for 10-day returns).
• Days when VIX is greater than 125% of this moving average (ranging from 295 instances for 252-day returns to 335 instances for 10-day returns).
• Days when VIX is greater than 140% of this moving average (ranging from 114 instances for 252-day returns to 141 instances for 10-day returns).

The chart suggests that a relatively low VIX is slightly bullish across all return horizons. It also suggest that a relatively high VIX is somewhat bearish over the near term and bullish for the intermediate and long terms.

However, because the instances of relatively extreme VIX are clustered, these results tend to overemphasize contributions to average returns in certain periods and do not translate to any realistic systematic trading strategy. To eliminate clustering effects, we winnow signals such that return measurement intervals do not overlap and would therefore be tradeable.

The final chart summarizes, for the winnowed (tradeable) subsamples, average S&P 500 index returns over the 10, 21, 63 and 252 trading days after:

• All days in the sample.
• Days when VIX is less than 80% of its 63-trading day moving average (ranging from 13 instances for 252-day returns to 51 instances for 10-day returns).
• Days when VIX is greater than 125% of this moving average (ranging from 13 instances for 252-day returns to 74 instances for 10-day returns).
• Days when VIX is greater than 140% of this moving average (ranging from 9 instances for 252-day returns to 29 instances for 10-day returns).

Results are mixed. Over the short and intermediate terms, stock market returns after extreme VIX conditions are mostly close to the norm. Over the long term, returns after all the relatively extreme VIX conditions underperform the norm. However, subsample sizes are small, especially for long-term returns, so these conclusions are not reliable. Just a few new instances could substantially alter average returns.

Moreover, there are not enough trading opportunities to make a strategy based solely on the selected thresholds for VIX signals interesting.

In summary, evidence form simple tests does not support a belief that a high (low) VIX or a relatively high (low) VIX reliably predicts future stock market returns or supports a systematic trading strategy as a standalone signal.

For related research, see Blog Synthesis: Sentimental Journey and Blog Synthesis: Volatility Effects.

To discuss this research, go to the Sentiment Indicators Forum or the Volatility Effects Forum.