Yale economist Robert Shiller has been right far more often than most in his profession, and he commands far more respect. Not only did he warn of stock market peaks in 2000 and housing troubles in 2004, but he did so largely on the basis of his own metrics and data. He draws different conclusions from most because he thinks differently from most.

CAPE -- the cyclically adjusted P/E ratio -- is one of his metrics. A standard P/E ratio can be misleading at the tops and bottoms of business cycles, as one year's earnings may be unsustainably high or low. CAPE gets around this by averaging together 10 years of earnings and adjusting for inflation.

What it shows right now troubles some market watchers. Since 1871, CAPE has averaged about 16. Right now, it's at 22. Many have used CAPE to argue that the Dow Jones (^DJI 0.46%) and S&P 500 (^GSPC 1.03%) are overvalued -- by quite a lot, in fact.

But there are two sides to every story. Last week I asked Charles Schwab chief investment strategist Liz Ann Sonders what she thought of the CAPE argument. "Here are a couple of problems with that metric," she said. Hear her out (transcript follows):

Sonders: Look, I'm not generally a critic of people's valuation metrics, and I understand the rationale for the CAPE valuation, and I know Dr. Shiller, and I have great respect for him, but here are a couple of problems with that metric. I think there's a lot of validity to looking over a long historical timeframe to attempt to even out the economic cycles, but one of the problems is in general, economic cycles have gotten shorter. And right now, depending on whether you measure them from trough to trough or peak to peak, to take basically an average of those two varieties of measurement, you're talking about no more than about a six-year span. So looking at a 10-year span goes beyond the average span of an economic cycle this time.

The other problem is that we've had, effectively, at least in the 10 years that ended in 2012, two -- let's hope -- unprecedented burst bubbles, inclusive of the one from an earnings perspective that was worse, in the 2000 and 2003 timeframe, which was a 60-year record decline in earnings.

There's other problems, too. One of the nice things about the Shiller CAPE is it goes all the way back to the 19th century. The problem is, accounting standards then versus accounting standards now -- it's an apples-to-oranges comparison. There's a static measurement in terms of CPI, but CPI gets changed regularly.

And then as far as back to the accounting-standards thing, he looks purely at reported earnings. So what has happened, particularly in the last cycle and with the housing crisis, is what we know, at least in terms of the banks, a lot of those hits were non-recurring, yet they're getting put in the numbers for a 10-year span. So I think there are enough problems with the construction of that, not to mention the fact that in the '90s bull market, you went to above the mean in 1991 -- a mere nine years shy of the peak in the market -- and you've been above the mean for much of this bull market as well. So I understand to get a sense of long-term valuation, why using it, but if anybody's using it as an entrance and exit timing tool, I think it's fairly dangerous.

Then, if you're a raving bull, you can support your case by looking purely at forward P/E, which the market still looks quite cheap right now. I think valuation to some degree is in the eye of the beholder, but because CAPE has gotten so popular, I think it's important to at least point out some of the problems with that measure.