Is Shiller P/E Worth It?

The Shiller P/E has been in the news lately (1, 2, 3) as various critics and proponents say say that the high level (~24 vs 17.5 average in the last century) is either indicative of danger or not applicable in this case. To review: the Shiller P/E is the 10 year inflation adjusted trailing P/E of the S&P 500. Also called the cyclically adjusted price-to-earnings ratio, the ratio is meant to smooth the volatile earnings of the business cycle and generate a genuine measure of value for the market.

After reading AQR's paper extolling the use of Shiller P/E (10 yr trailing P/E) while acknowledging its limitations, I am tempted to ask the question - Shiller P/E even worth it as a tool?

(Source: http://www.multpl.com/shiller-pe/)

There is little question that the indicator is elevated. Apart from the 90s technology craze and the 20s pre-Depression bubble the indicator has never been this high. But what does this actually mean for markets going forward? Before even questioning the assumptions behind Shiller P/E, what is its track record taken at face value? Again, AQR (and many others) have done some analysis:

As highlighted - we are in the 2nd to highest decile. This implies a measly 0.9% mean annual return going forward. But what about the volatility? The best 10yr return was 8.3% annually vs -4.4% worst. Not exactly compelling for a buy, but is it worth it to sell (short)? Selling now and being wrong means possibly missing/losing 8.3% annually for 10 yrs. Does that present an attractive risk reward?

On the other hand, buying at the lower two deciles is far more compelling. The worst case is a 4.8% return for the lowest return & 10.3% average. I'll take that all day.

Bottom line - I don't think Shiller P/E is useful for calling tops, even using the proponents' data. It is, however, possibly useful for bottoms. The market may correct severely tomorrow, but don't think that using this indicator one would have sold anywhere near the top.