Does P/E matter?

Does PE matter? People say that PE (Price to earnings ratio) is a primary factor when choosing a stock to outperform, but lets put it to the test Mythbuster style… The first, most obvious idea is to backtest “low” PE stocks:

Wow! That looks encouraging…. but how does it compare to the baseline. The strategy we used was simply PE < 20, and a subset of the market that has stocks with reasonable market caps, reasonable volumes, and a few employees. We now plot the baseline, which is just the tradable screen.

That’s rather surprising — the “null market screen” managed to outperform over the long haul… What’s in this “null” screen anyway?

Let’s look at those both at the same time. We do this by making the backtest we just ran into a “value” and adding it to the line as a plot (the % meaning plot that value on the same axis as the backtest itself). This lets us compare the two backtests side by side over a variety of time intervals, using the “backtest by time” functionality. This takes the performance we saw, and lets us see it in a histogram of performance, period by period. We’ll do this to see which years the two screens outperformed the S&P 500, vs underperforming them.

So, it looks like we outperformed each year (except 2011, and 2014)! The tradable stocks did just about the same, except that it underperformed more in 2011 and 2014. Maybe we’re just boosting the volatility, and generating excess returns just from the positive direction of the market. If so, we should underperform during the bad times, and outperform during the good ones…

Like that. Notice that the outperformance during the good times is just about the same as the underperformance during the bad times. Based just on that, I’d be reluctant to invest in the screen, as one could just go for more volatile stocks in your portfolio if you want more action… For an example of a screen that is not just beta boosting, I present you with the following results

Notice that although we still suffer from excess volatility, the good outweighs the bad. I’d take this screen (which takes estimated earnings yield and slices it 44 different ways), over the other screen any day. We’ll see this screen (or a variant thereof) next time…