CHICAGO (CNN/Money) - Back in the bubble days, Wall Street analysts demonstrated considerable creativity in making the numbers of unprofitable dot.coms smell sweet.

Analysts were happy to ignore innumerable "one-time" charges, even if they mysteriously seemed to recur every quarter, and anything else that made the numbers look bad.

Among the most creative of the bubble era analysts was Merrill Lynch's Henry Blodget, who in one particularly inventive moment suggested that investors should reward Yahoo for, among other things, the tax breaks the company got when employees exercised stock options.

(Never mind that a cushy tax windfall isn't exactly the same thing as earnings from a company's core business. Never mind that employees would continue to exercise their options only if Yahoo's stock price kept going up and up.)

These days, Merrill would rather we forgot all about Blodget's inventive "analysis" -- not to mention the similarly creative advice that some are alleging Merrill gave to a certain company called Enron.

There's no clearer evidence of Merrill's attempts to move beyond the bad old days than its initiative on quality of earnings. Merrill identifies several key metrics by which investors can attempt to judge earnings quality and thus attain what Merrill calls "a 360 [degree] view of reality."

All cynicism aside, the initiative is a worthwhile one -- and, it almost goes without saying, a 180 degree turn from Blodget-style analytics.

Merrill's earnings-quality checklist is a strange and somewhat wonky one. Here's a quick overview:

• Return on Capital: Merrill calculates this by adding together pretax income and interest expense and dividing by total assets minus current liabilities; the higher the returns the better.
• Cash Realization Ratio: This is cash from operations divided by net income, which Merrill says you can use to see how much of a company's net income is not puffed up by "non-cash sources ... that may be biased by aggressive, management valuation judgements."
• Productive Asset Reinvestment Ratio: Capital expenditures divided by depreciation expense -- essentially looking to see that the company is making adequate investments in its own future. (Cutting CapEx can help earnings in the short run but hurt them in the long run -- Merrill wants earnings to be "repeatable.")
• Tax Rates: Companies that have lost money in the past get tax breaks on future earnings -- Merrill advises checking to make sure earnings weren't pumped up by an unsustainably low tax rate.
• S&P Ratings: Merrill looks at the S&P's stock and credit ratings for each company in an attempt to judge whether the company's earnings look to be sustainable (heavy debt and big dividends could put future earnings at risk).

Merrill notes that quality of earnings alone won't determine whether or not a stock is a good bet -- there is, after all, the question of valuation -- and some of the companies that boast high marks for earnings quality may not rate as highly with Merrill's analysts.

But Merrill's quality of earnings analysis of the stocks in the S&P 500 is nonetheless instructive. Of the 28 stocks that score high marks in all of Merrill's quality of earnings categories, only one (Intel) is a "real" tech stock; electronic payment processor Concord EFS also makes the list, as do local phone companies Bellsouth, Verizon, SBC and Alltel. Food distributor Sysco makes the list (Cisco does not).

Indeed, while consumer and (oddly enough) energy stocks score the highest marks overall for earnings quality; techs score the worst. Looking at the 25 top-scoring companies that Merrill analysts also rate a "buy" or a "strong buy," the only techy name on the list is Concord EFS.

Picking stocks seemed a lot easier when all that seemed to matter was whether or not a company beat the numbers by a penny or two every quarter. Now it turns out these numbers have to be based on something real? Who knew!

Sign up to receive the Tech Investor column by e-mail.

Plus, see more tech commentary and get the latest tech news.