(MONEY Magazine) – The only sure thing on Wall Street is that there never has been, and never will be, a sure thing on Wall Street. But index funds—those low-cost baskets of all the stocks in a market benchmark—come about as close to a sure thing as you can get. Low in cost, high in tax efficiency, index funds are your best chance of earning nearly 100% of whatever return the market generates. That's why I've consistently recommended them, with almost religious zeal, for nearly a decade.

Mind you, index funds have always had their critics—professional stock pickers, mostly—but for the first time they've come under attack from someone with real mental firepower. Robert Arnott of Research Affiliates in Pasadena is not just a leading money manager, he's also the editor of the prestigious Financial Analysts Journal. Arnott recently sent me his new paper, "Redefining Indexation." In it, he critiques existing index funds and proposes a radically new way to track the market.

Every once in a while something comes along that can shake the faith of even the most fervent believer. I'm not sure that Arnott has really built a better mousetrap. But his proposal underscores a fact about index funds that many investors forget: They may be reliable (in the sense that they always match the market), but they certainly are not risk-free.

Buy high and sell low? Before we explore Arnott's argument, let's remember how index funds work. Say you decided to own the same amount of every stock in Standard & Poor's 500-stock index. You would put 0.2% of your money in each of the 500 companies; out of a $10,000 investment, you'd have $20 in Delta Air Lines, $20 in Microsoft and $20 in each stock in between. You'd then have what's called an equally weighted index fund.

But do you really want to own as much of Delta Air Lines, whose growth prospects seem to be shrinking by the minute, as you do of mighty Microsoft? Probably not—and that's why the typical index fund is "capitalization weighted," allocating its money across stocks based on their total market value. The market thinks Microsoft is worth vastly more than Delta—Microsoft stock has a total value of $305 billion vs. $718 million for the airline. So if you had put $10,000 into an S&P 500 index fund in late October, you'd essentially have been investing $290 in Microsoft and less than $1 in Delta.

Sounds sensible. But Arnott points out that this kind of indexing leaves investors at the mercy of market moods. The more a stock goes up, after all, the more money the index fund ends up having in it. This got particularly bad a few years ago when tech stocks became insanely expensive: The sector went from 29.8% of the index in December 1999 to 33.5% in March 2000. At the market's peak, 44% of the total value of the S&P 500 was in just 25 stocks, including such ticking time bombs as Lucent, Nortel and WorldCom. Instead of "buy low, sell high," indexing had morphed into "buy high, hope to sell higher." People who owned index funds were, in effect, bodysurfing atop a crowd of desperately crazy buyers.

The aftermath was ugly. As tech stocks imploded in the bear market of 2000-02, the S&P 500—and the index funds that seek to replicate it—lost an average of more than 14% a year. Meanwhile, thousands of other stocks with little or no weight in the index were doing just fine. According to portfolio manager Langdon Wheeler of Numeric Investors, the average stock outperformed the S&P 500 by a cumulative 49 percentage points between March 2000 and September 2004.

Considering the alternative To avoid this Achilles' heel of the traditional index fund, Arnott ranks companies by what he calls "fundamental, commonsense measures of size," such as revenue, operating income, book value, total dividends paid and total number of employees. (You can view his paper at rallc.com/articles/indexation.pdf.) The result is an index that focuses on the most important companies on Main Street—not just the biggest and hottest stocks on Wall Street.

What's the difference? At the end of last year, Microsoft was the second-biggest stock in the S&P 500 index, but it was the seventh-biggest company in Arnott's index; Wal-Mart was sixth in the S&P but first in Arnott's index; and General Motors, only the 77th-largest stock in the S&P index, was the second-biggest company in Arnott's. Once a year, Arnott will re-rank the companies, based not on how their stock price has changed but on how the size of their underlying business has changed. That way his index—unlike the S&P 500—won't get driven to extremes by a buying frenzy like the tech mania.

The bonus: Arnott estimates that if his index had existed from 1962 through 2003, it would have beaten the S&P 500 by 1.9 percentage points annually.

I think this is an intriguing idea, and it will be interesting to see how it works in real life. (Arnott's fund is still on the drawing board and will only be available to major institutional clients.) But in the end, he hasn't convinced me that I need to worry much about my plain-vanilla index funds.

For all of indexing's idiosyncrasies, Arnott's version just adds new ones. His emphasis on big companies with lots of physical assets and loads of employees gives his index a tilt toward value stocks, at least compared with the S&P. Who knows whether that stock-picking method will work as well in the next 42 years as it has in the past 42? And figuring out the true size of a company is not an exact science. Laurence Siegel, director of investment policy research at the Ford Foundation, points out that as of last year, eBay had 5,700 employees and $2.2 billion in net revenue, making it small potatoes by Arnott's measure. But the auctioneer had 41 million active users (many of whom function as if they were eBay employees) and auctioned off goods and services with a gross value of $24 billion. Maybe the stock market, which values eBay's shares at around $60 billion, has the right idea. This isn't a little company—it's a giant.

Sure, the prevailing view of the market can be very wrong in the short run, but its collective judgment is tough to beat. Over the past decade, the Vanguard 500 Index fund has beaten competing funds by an average of 1.5 points a year. "All the aberrations," says Vanguard funds founder John Bogle, "get ironed out over the long term."

The important words here are "long term." Arnott has done a service in pointing out that by their very design, index funds can—no, make that will—hit rough patches. If you understand why this happens and know that it's inevitable, you ought to find it easier to hang on through the pain. Nothing's perfect—but for investors who can focus on the long run, close enough will do.