By BRETT ARENDS

As investors look into their crystal balls and try to decide which stocks to own in 2013, they could always invest in the ones most highly recommended by Wall Street pros.

Or investors could just try flipping a coin.

On Wall Street, unlike in high school, popularity is no sure guide to success.

The stocks that are most popular with financial analysts and investment managers often prove to be poor investments. And those that are the least popular often turn out surprisingly well.

"The consensus is often wrong," says Lawrence Glazer, portfolio manager at Mayflower Advisors, an investment advisory in Boston with $1 billion under management. "There's no better contrary indicator than when a stock is loved by Wall Street and Wall Street research."

Every year around this time, I look at the stocks that started the year with the highest ratings among analysts who followed them—the most "buy" recommendations and the fewest "sells"—and see how those picks did.

I also look at the stocks that started the year with the lowest ratings among analysts.

The information comes from Thomson Reuters, TRI -0.59%which monitors recommendations across the market.

In theory, at least according to the Wall Street marketing machine, the stocks with the highest ratings should outperform the overall market by a country mile. The stocks with the lowest ratings should do very poorly. This is, after all, what analysts are paid for.

In 2012, let it be said, Wall Street analysts had a good year. If you had invested an equal amount in their 10 favorite stocks a year ago and reinvested the dividends, you would have made an overall return of 21%. For example, power-tool maker Snap-On SNA -0.70%gained 58%, while pumps and valves company Flowserve FLS -0.82%jumped 48%. Specialty chemicals company FMC FMC -0.76%was rated a "buy" by nine analysts and a "sell" by just one. It rose 36%.

The return from the 10 most popular stocks beat the 14% overall return by the Standard & Poor's 500-stock index.

Yet the stocks that began the year least popular among analysts didn't do as badly as one might expect. As a group, they produced an average investment return of 10%. Several were spectacular performers: Movie streaming company Netflix NFLX -0.72%gained 31%. Investment company Federated Investors FII -0.70%—which had just two "buy" recommendations among 15 analysts—leapt 49%.

There is no clear pattern to how analyst picks perform. In two of the past five years, the least popular stocks—those which began each year with the lowest rating among analysts—handily outperformed the most popular stocks. In two other years, it was a virtual dead heat.

Overall, buying the 10 most popular stocks each year would have brought you total losses of 9% over the past five years. Owning an index fund that simply tracked the S&P 500 instead would have been much better, earning a 10% profit. And buying the 10 least popular each year would have been better still, earning 16%.

According to FactSet FDS +0.60%data, the worst-performing stock in the S&P 500 this year has been educational company Apollo Group, APOL +0.44%owner of the online University of Phoenix. The stock has tumbled 63%.

Yet of the 20 analysts who followed the company a year ago, 12 recommended the stock as a "buy," while eight gave Apollo a neutral "hold" rating. Not one recommended selling the stock.

Similarly, among 61 analysts there were no "sell" ratings a year ago for Cliffs Natural Resources, CLF -1.20%Allegheny Technologies ATI -1.35%or Newfield Exploration NFX -1.33%. Investors had losses of at least 30% on each of those three stocks.

People who live outside the bubble of Wall Street often find this upside down. They are used to a world where the best products and services get the best reviews. No one would go to a restaurant panned by customers, or a movie that earned a dismal 0% at review website RottenTomatoes.com.

Yet on Wall Street, investors step through the looking glass. The first, quite often, shall be last, and the last shall be first.

In a 2010 paper that ran in the Accounting Review, which is published by the American Accounting Association, business professors Michael Drake of Brigham Young University and Lynn Rees and Edward Swanson of Texas A&M University examined the investment performance of the most and least popular stocks on Wall Street over the period from 1994 to 2006. Their conclusion: "Overall, following the consensus analyst recommendation does not generate positive abnormal returns for investors, and in some periods can actually generate significant negative returns."

The reasons for this are embedded in the system. Restaurants and movies aren't changed by their reviews. Yet on Wall Street, positive recommendations for a stock tend to drive up the price. That makes it a more expensive—and less attractive—investment.

Where there are many positive reviews, the price often is driven too high in relation to the risks and likely returns.

On the other hand, where there are few positive reviews, the price often has fallen too low.

There are other problems.

Ten years ago, the Sarbanes-Oxley Act cracked down on the most obvious conflicts of interest, such as investment banks effectively being paid by companies to churn out positive recommendations. Yet deeper and less obvious conflicts remain. Analysts who go against the herd—recommending the bargain-basement stock of a company in crisis, for example, or advising clients to sell a highflier—face career risks if the analysis turns out to be wrong. They may also face a backlash from any company for which they issue a "sell" recommendation.

As a result, just 5% of all analysts' recommendations across the S&P 500 are "sells," according to Thomson Reuters data.

Analyst research still be very valuable. A good research report can often show an enormous amount of insight about a company's fundamentals, risks, industry challenges and opportunities. The least valuable part of the note is the actual recommendation. There, as so often, investors are on their own.

Write to Brett Arends at brett.arends@wsj.com