At any rate, if the market trends upward and trading is close to random (regardless of what the trader believes) then being out of the market occasionally will always be bad statistically. If there are vast numbers of traders all actually acting randomly then some will out preform others and some tiny number might out preform the market. In fact, if you have the data, you can plot the performance of every trader against the market average and actually measure how far from random the average trader is.
Maybe price to earnings ratio "should" relate to stock price. But in actuality stock price is determined by what people are willing to pay for it obviously. To the degree that people are buying and selling _not_ based on P/E, say for example they are doing so somewhat randomly, then the P/E won't reflect accurately the stock price: it won't be predictive.
These hulking paragraphs is what the demo is basically saying.