Some analysts claim that each decade has certain years which outperform others. For example, some believe that years which end in "5"¯ outperform years which end in "1"¯. Is this true? Are there years in the decade which outperform others?
To find out, I ran a test on Dow Jones data where the investor would be absent one of the ten years of the decade since 1900. Since we use a decimal system it is easy to specify a particular element in the cycle, as it can be marked by the last digit. Is it best to be out of the market in years ending in "0"¯? Or years ending in "1"¯? We can continue this query for all of the ten possible years in the decade.
Some important limitations should be noted. Most importantly, I should emphasize the fact that we are looking at a small number of cases . As there is only just over 100 years of data for the Dow Jones, we only get ten cases for each year of the decade. This is a very small universe and this is one of the most compelling reasons to view the results with some degree of caution. Nonetheless, as we will see, the results are quite a bit better for some of the years than was the case with the Presidential Strategy.
|
Figure 1
Decennial Strategy vs. Buy and Hold
| Out Year |
Decennial Strategy |
Buy and Hold |
| 1 |
$2,737,258 |
$1,478,555 |
| 2 |
924,692 |
1,263,396 |
| 3 |
776,884 |
1,527,939 |
| 4 |
836,511 |
1,574,847 |
| 5 |
92,827 |
1,592,164 |
| 6 |
586,630 |
945,617 |
| 7 |
2,352,178 |
1,169,094 |
| 8 |
276,351 |
1,348,049 |
| 9 |
758,348 |
1,667,035 |
| 0 |
3,951,919 |
1,562,623 |
|
|
The mechanics of the test are as follows. For each target year, the Dow Jones index was sold at the first market day's closing price. (This would of course be a basket of Dow stocks in the pre-ETF era) The Dow was then bought at the price of the first market day of the year following the target year. For example, if we are looking at the years ending in "1ā" the first transaction following the initial purchase of 147 "shares"¯ in 1900 would be the sale on January 2, 1901. The investor would have sold the initial shares for 70.44 on January 2, 1901 for $10,339. The proceeds from this sale would have been applied to a new block of 161 shares the following year on January 2. The shares would have been held until January 1911 at which point they would be sold, then the proceeds from that sale would be applied to a new batch in January 1912. The process would repeat every decade until 2002 (i.e. the last purchase of 272 shares on January 2 of that year).
Certain simplifications were made for the test. For example, I used fractional shares rather than whole number shares. Commissions, slippage and dividends were not taken into account. Because of the infrequency of trading using this strategy most of these would have a negligible effect. Dividends are the only possible exception. Still we are only looking at the differential between the dividend yield and short term interest rate (assuming the proceeds from the stocks were so invested) for one year per decade so even this doesnŹ¼t represent a material difference.
The results of this test are shown in Figure 1. On the left hand column you will see the particular year of the test. This is the out-market period during which the stocks were not held. The next column shows the ending account value using this strategy. In the right hand column I have noted the Buy-and-Hold ending account value at the last purchase using the strategy. These do vary somewhat because the final point selected was different depending on which year the final purchase was made for the strategy.
