SWEETBREADS AREN'T PASTRIES

. I learned that one bite too late at a French restaurant years back. Turns out it's a dish made from the thymus gland of a juvenile cow or sheep. I tell you that to set the tone for a truly unappetizing recipe I'm about to share. It's how Wall Street cooks up its price targets. You know: Shares of Next Big Thing, now $35, are forecast to hit $50 within a year, and hence we rate it as a "buy." That sort of thing.

I have two motivations. I want you to avoid the financial equivalent of my unsavory surprise by ignoring price targets altogether. Also, I want to introduce a new idea on stock picking that uses Wall Street's price-target math, only backward, to show which companies seem most likely to live up to investor expectations.

There are two basic ways to calculate a fair price for a business. "Relative valuation" involves comparing a company with others in the same line of work, often using a simple price ratio: price/earnings, price/sales, price/book value and so on. It's a reliable tactic, although one that's perhaps best used in a screen for potential bargains rather than in assigning an exact price to a specific company. Mostly, though, Wall Street prefers "absolute valuation," which puts a current price on money that a company is expected to make in the future. It's a mathy process, and so it earns the analyst who performs it the same mystified reverence afforded the faith healer or fashion designer. But really, it's guesswork plus a string of calculations, which equals guesswork.

Suppose I offer to sell you a coupon good for $100 cash, payable by me a year from now. How much would you bid? I hope you didn't say $100. You should be compensated for tying up your money for a year. At the very least you should discount your price for the interest you would have gotten from a bank certificate of deposit or the like. Let's say that puts your bid at $95. But hold on. You also have to adjust for the possibility that I'll skip town, go broke or offer a confused shrug and pretend to speak only Portuguese a year from now when you try to present the coupon. You decide to pay only $93.

Congratulations: You're a business card away from being a stock analyst. Only, companies don't tell you how much they'll pay you or when. You have to guess. And not just for a year. You need a good 10 years of forecasts to come up with the price you should pay today. (The further out the projected payments go, the less they're worth today, and so the money a company makes after year 10 doesn't matter nearly as much as the money it will make before it.) It used to be popular to base today's price on forecasted dividends, money that at least we know will be of value to stockholders if it shows up. Today companies are so stingy with dividends that analysts instead forecast the amount of spare cash companies will generate, regardless of how they will use it. That's called discounted cash-flow analysis. Start by projecting a decade of sales, margins, equipment investments and taxes. (Forget for a moment that a majority of trained forecasters won't get this quarter's earnings estimates right.) Then use math to turn the anticipated stream of cash into a suitable current price.

For more SmartMoney Magazine features, turn to the current issue.

Finally, apply a discount according to the company's risk. How do you measure that? You don't. You guess, based on either the jitteriness of its stock chart or on a hodgepodge of financial characteristics. Be careful; a slight change in your risk measure produces a dramatic change in the price you end up with.

Numerous long-term studies suggest neither price targets nor the recommendations they lead to hold much predictive power. But discounted cash-flow analysis can be put to more reliable use if, rather than starting with an unknown (tomorrow's cash flow), we start with a known today's market price. By working backward we can figure out the amount of money a company would have to make next year, the year after that and so on to keep it on track for being worth what investors are paying today. That, essentially, was the message of a 2001 paper and subsequent book titled "Expectations Investing," authored by Northwestern professor turned consultant Alfred Rappaport and Legg Mason strategist Michael Mauboussin. A stock's price, the two argue, is an informational gift from the market. Rather than ignore it, as Wall Street's seers do, embrace it and learn what it says about the expectations of other investors.

Enter Julian Koski and Armen Arus, a pair of former securities analysts who five years ago set out to turn this backward valuation math into an advisory service called Transparent Value. Their methodology, dubbed Required Business Performance, assigns each of 2,000 or so companies under coverage a probability that it will hit the sales numbers that are baked into its price. It's not nearly a purely quantitative process; a staff of 90 analysts must judge each company's recent financial trajectory to figure out whether next year's required numbers seem likely. Earlier this year Required Business Performance was licensed by Dow Jones as the basis of a stock index. It starts with a Wilshire index of 750 large companies and then overweights the top 30 percent or so that have high probabilities of success, while underweighting the ones with low probabilities. The index has been out for only a month, so the percentage point by which it has beaten the benchmark doesn't mean much. But a back-tested version based on the Dow Jones Industrial Average returned 72 percent over seven years ended May 2007 more than double the Dow's return, besting it in 22 of the 28 quarters.

Transparent Value provided me with an assortment of high- and low-probability stocks, which I've listed for you. I recommend you treat the probabilities themselves as a rough gauge rather than an exact reading. A few of the stocks have probabilities of 100 percent. Not even a fart joke at a grade-school assembly is that assured of success. JPMorgan seems a better bet than U.S. Bancorp, judging by the nearly 80 percentage points separating their probabilities. A breakdown of the analysis shows JPMorgan can justify its price by delivering just 4.4 percent sales growth this year, a sum that's in keeping with company forecasts and that seems modest compared with last year's 15.1 percent growth. U.S. Bancorp, meanwhile, must find a way to follow up last year's negligible growth with a 5.8 percent increase this year.

Bet on Best Buy, but avoid the Gap. Not only is Best Buy cheaper relative to this year's profit forecast, but it has also done a better job of beating estimates in recent quarters. Gap, in fact, ranks dead last among the 21 clothing sellers in Transparent Value's database.

Western Digital shares have more than doubled in price since September 2005, when I pointed out in an online column how cheap the hard-drive maker looked next to trendy flash-memory makers at the time. But I turned cool on Western's stock in a March follow-up, noting it had recently experienced a perfect storm of profits, with demand for big laptop drives surging as competitors failed to bring theirs to market, and that such performance might be difficult to repeat. Transparent Value couldn't disagree more, calculating that sales this year need only grow a third as fast as last year, according to the stock price, but will likely grow more. It's less than confident, though, on mouse, headset and joystick specialist Logitech, whose price implies a quickening of sales growth this year.

Warship and combat-electronics maker Northrop Grumman has soundly beaten the broad market over just about any stretch within the past decade. Wall Street envisions 5 percent sales growth this year. The stock's price implies 6 percent, according to Transparent Value, and the company has just a 37 percent chance of hitting that. Faring better might be conglomerate United Technologies. It does a dash of weapons business through its Sikorsky unit, best known for its Black Hawk attack helicopter. But most of the company's wares are less intimidating: elevators, air conditioners, fire alarms and airplane engines. The stock price implies sales growth of just 6.6 percent this year, less than half last year's increase and well below Wall Street's projections.