Seasonality and The Variance Premium
The idea of “sell in May and go away” isn’t new. It is pretty much the opposite of new. It is an anomaly that has persisted for 322 years (the UK market since 1693).
The effect is persistent, large and significant, statistically.
Since 1950, the summer (May to October) annualized returns for the S&P 500 have been 1.4% while those for the winter have been 7.0%. This effect has been studied across many countries and it holds almost everywhere.
The effect also holds for the variance premium, although we don't have nearly the same amount of data. Since 1990, the difference between the VIX and the subsequent 30-day realized volatility has averaged 4.5 points in winter (43% of the realized volatility level) and 3.7 points in summer (41%). Both a t-test and a Kolmogorov-Smirnov test give the point difference a 99% confidence level (the % difference is a little less significant).
While this is smaller than the effect in equity indices it still has direct trading applications. Both of the CBOE volatility strategy indices BFLY and CNDR do better in the winter. BFLY has an annualized return of 7% in the winter and 4.7% in the summer, while CNDR returns 9% in the winter and 4.7% in the summer. Neither of these results are statistically significant but they would be if we had the same amount of data as we do for the S&P 500.
This result shouldn't be entirely surprising. The link between positive equity returns and a high variance premium is well known. But it is still nice to confirm that the raw effect can be translated into money. This isn't always the case.