The question
A ten-year result is one number. A few great years can make it, with many dull ones hidden inside. If a rule only worked in one part of the decade, the ten-year figure would not show it.
So the launch bar, the test a strategy must pass before anyone is asked to follow it, has a rule about this. A strategy has to beat the S&P 500 in the first five years and in the last five, each tested by itself.
How we tested it
Each strategy got two fresh tests. The first starts with nothing on and ends on . The second starts with nothing on and ends on .
Both put in every trading day, under the same rules as the ten-year test. The fresh start matters. The second test gets no head start from stocks bought in the first.
The yearly return is the steady yearly rate that would turn those daily deposits into the final amount.
The result
What it means
A strategy that finishes ahead in both halves is less likely to be a one-off. Its ten-year result was not carried by one good stretch.
A pass by a hair shows much less. Where a strategy and the index finish within a fraction of a point of each other, the fair reading is that the rule added nothing in those five years.
Finishing ahead is also not the same as being ahead along the way. A strategy can win a five-year test and still spend most of it behind the index. That is the subject of what a bad stretch looks like.
What it does not prove
- Two halves are only two tries. The index itself made a year in the first and in the second. Neither test shows how these rules do through five poor years for the market.
- Five years is short. One or two unusual years can decide a five-year test either way.
- Nobody invested this money. It is a test on past prices, and live results are likely to be lower.