Kaeppel’s Sector Seasonality Strategy

A reader suggested looking at the strategy described in “Kaeppel’s Corner: Sector Seasonality” and updated in “Kaeppel’s Corner: Get Me Back, Clarence”. The steps of this calendar-based sector strategy are:

1. Buy Fidelity Select Technology (FSPTX) at the October close.
2. Switch from FSPTX to Fidelity Select Energy (FSENX) at the January close.
3. Switch from FSENX to cash at the May close.
4. Switch from cash to Fidelity Select Gold (FSAGX) at the August close.
5. Switch from FSAGX to cash at the September close.
6. Repeat by switching from cash to FSPTX at the October close.

Does this strategy materially and persistently outperform? To investigate, we compare results for three alternative strategies: (1) Kaeppel’s Sector Seasonality strategy (Sector Seasonality); (2) buy and hold Vanguard 500 Index Investor (VFINX) as an investable broad index benchmark (VFINX); and, (3) a simplified seasonal strategy using only VFINX from the October close through the May close and cash otherwise (VFINX /Cash). Using monthly dividend-adjusted closing levels for FSPTX, FSENX, FSAGX, the 13-week Treasury bill (T-bill) yield as the return on cash and VFINX over the period January 1987 through April 2012 (25+ years), we find that:

Calculations assume that:

• VFINX returns for February-March 1987 are the same as those for the S&P 500 Index (March 1987 is the earliest available data for VFINX).
• The return on cash for a month is one twelfth the prior month T-bill yield.
• Given the nature of the trades, ignore trading frictions.
• Ignore the tax implications of trading.

The following chart compares the cumulative values of $10,000 initial investments at the end of January 1987 for all three strategies over the entire sample period. The Sector Seasonality strategy clearly outperforms, with a burst in 1998-1999 that includes a 55% jump in the gold fund during September 1998.

For another perspective, we look at monthly return statistics.

The next chart depicts the average monthly returns and standard deviations of monthly returns for the three strategies over the entire sample period. The Sector Seasonality strategy generates a substantially higher average monthly return than the benchmark alternatives, but with a considerably higher variability in monthly returns. However, the cumulative profile above suggest that monthly upsides, rather than downsides, drive the high standard deviation.

How do the monthly parts contribute to the whole for the Sector Seasonality strategy?

The next chart shows the average return by calendar month for the Sector Seasonality strategy and for VFINX over the entire sample period. The Sector Strategy beats VFINX on average in 10 of 12 months, with outperformance most pronounced for September. Four very strong years (1998, 1999, 2005 and 2007) drive the exceptional performance of the gold sector fund in September.

Is the average outperformance of the Sector Seasonality strategy consistent over the sample period?

The next two charts summarize average monthly returns for the three strategies defined above by subperiod (February 1987 – December 1994, January 1995 – December 2002, January 2003 – December 2011 and May 2010 – April 2012) and by calendar year.

The first chart shows that the Sector Seasonality strategy outperforms the other two strategies in the three longer subperiods, with outperformance most pronounced in the middle subperiod. However, the strategy underperforms the past two years.

The second chart shows that 1998 and 1999 drive the elevated outperformance of the middle subperiod (due to strong performances during those two years by the technology sector fund and the gold sector fund). The Sector Seasonality strategy beats buying and holding VFINX based on average monthly returns in 16 of 25 years (64%) and VFINX / Cash in 15 of 25 years (60%).

Is the outperformance of the Sector Seasonality strategy weakening overall?

The final chart plots the monthly performance of the Sector Seasonality strategy relative to that of VFINX over the entire sample period, with a best-fit linear trend line. Overall, the Sector Seasonality strategy beats VFINX in 167 of 303 months (55% of the time). The trend line is approximately flat, offering no evidence that the Sector Seasonality strategy is weakening. This persistence argues against both data snooping bias as an explanation of past performance and market adaptation.

In summary, evidence from simple tests indicates that Kaeppel’s Sector Seasonality strategy may offer persistent average market outperformance.

Cautions regarding findings include:

• The sample period is short for analysis of seasonality, especially for subperiods.
• There be still be material data snooping bias in strategy development, with the out-of-sample subperiod not long enough to expose it.