Also, a lot of that 35 percent is driven by what analysts call "tail" events—historical anomalies that skew the results. Strip out the high P/Es of the dot-come bubble and the low P/Es of the recessed early '80s, says Philips, and the 35 percent falls to 15 or 20 percent. Reduce the period from 10 years to one, and it disappears altogether.
Philips also warns against rules of thumb which say, for example, that after a lost decade of historical underperformance, the pendulum is due to swing back toward outperformance. "Mean reversion is a very dangerous game to play," he says. "Historical patterns look like they can be tracked and profited from. We tend to be a little cynical with respect to things like that. If it were that easy, a lot more people would be a lot more successful."
[In Pictures: The S&P's 10 Worst Trading Days]
Still, with bond yields at record lows and bond prices exposed to interest-rate and inflation risk, inevitably investors face the Henny Youngman question: Equities are risky compared to what?
"If you're a long-term investor—and here we agree with the takeaways of the Fidelity report—your best bet for realizing positive real returns—on average, over time—is going to be equities," says Philips. "They do a great job of mitigating the impact of inflation over longer periods and they should outperform more conservative investments—your money-market, your fixed-income investment—because they are riskier assets."
Even over the past decade, says Fox, investors using dollar-cost averaging have made money in stocks. Not much, but some.
"The lost decade is going into the rear-view mirror, you have made money in the stock market over last 10 years, and I think you'll have similar returns going forward," says Fox. "I don't think we're going back to the 1980s and '90s, where you made 17 or 18 percent a year. But there's nothing wrong with 8 percent a year for a decade. And compared to the alternatives, I think it's going to prove pretty good."