To iterate the conclusion: all starting dates and ending dates in any performance study are random, and therefore the results are random.
With this premise, the performance gain shown at every date point along each curve is equally probable. But--and here is the striking phenomenon--when you span 26 weeks' worth of individual, equally probable, coin-toss events into six-months groups every week, they emerge into an undulating pattern where hills succeed valleys and vice versa, each following the other!
Since this is so, and the rolling pattern recurs, when you start to perceive your account in a valley, you can next expect a subsequent return to the next coming hilltop--without fail.
But the most important thing is that, depending on your system, each rolling wave cumulates in the landscape view to a higher level of spread above the market, or a lower level if you've got the wrong system.
