In this article we use a strategy to analyze a variant of the January calendar anomaly in large cap stocks. The results are interesting and suggest this market inefficiency may be present but under specific conditions.
For the backtests in this article I used Norgate data for S&P 500 index that include current and past constituents. I highly recommend this data service (I do not have a referral arrangement with the company.)
General Description of the Strategy
Buy maximum N stocks at the open of the year,
based on a Rank Metric from the previous year and subject to conditions.
Sell the stocks on the open of the first trading day after January 14
The Rank Metric is the total return performance in the previous year. Our objective is to determine the combination of Rank Metric and conditions that produce the best results.
Backtest Details
The backtest starts in 1990 in an effort to minimize the sampling error. This type of backtest is possible when delistings and index constituent rebalancing are taken into account otherwise results may be biased.
The yearly S&P 500 chart below shows that since 1989 (first year for calculating the Rank Metric and any Conditions), there have been 24 up years and 9 down years in the S&P 500. There are 32 years after 1989 for generating the backtest results.
Case 1. Up to N = 5 top/bottom stocks according to Rank Metric. There are no additional conditions.
| CAGR | Total Return | Avg. Yearly return | % Winning Years | |
| Top stocks | 2.1% | 96.3% | 2.6% | 62.5% |
| Bottom stocks | 2.3% | 105.9% | 3.2% | 65.6% |
Case 1 results show there is a small edge when selecting the bottom 5 stocks instead of the top 5 but the difference may be due to sampling error. In addition, the average yearly return and percent of winning years indicate any edge is small and is probably not worth the risks.
Case 2. Up to N = 10 top/bottom stocks according to Rank Metric. There are no additional conditions.
| CAGR | Total Return | Avg. Yearly return | % Winning Years | |
| Top stocks | 1.5% | 58.6% | 1.7% | 50.0%% |
| Bottom stocks | 2.0% | 82.2% | 2.6% | 71.9% |
Case 2 results confirm there is a small bias when selecting the bottom 10 stocks instead of the top 10 but the average yearly return drops as compared to the Case 1. Although the percent winning years increased to about 72%, the results are still not satisfactory. If N is increased to 20, the same pattern as in Case 2 persists at about the same levels for the parameters.
Conclusion: The basic strategy shows a small edge in favor of buying bottom ranked stocks but performance is low and is not worth the added risks.
Below I investigate if there is a way to improve the results. The rules of the modified strategy are also included. This section can be accessed by subscribers with Market Signals or All in One subscriptions.
|
This post is for paid subscribers
Already a subscriber? Sign in |
10% off for blog readers and Twitter followers with coupon NOW10
Charting and backtesting program: Amibroker. Data provider: Norgate Data
If you found this article interesting, you may follow this blog via RSS or Email, or in Twitter.
