(MONEY Magazine) – Did you know I can run a mile in under four minutes? I haven't actually done it yet. But I ran 100 yards in 13.1 seconds the other day. If I hold that pace, I figure I can do a quarter mile in 57.6 seconds, a half in 1:55.3 and a mile in 3:50.6, easily breaking four minutes. Now if I just can whittle one teensy second off my 100-yard-dash time, I can lower my projected mile to 3:33, making me the fastest miler in the world! Not bad for an aging boomer who considers his daily walk to the train station a grueling exercise regimen.

You don't have to be a track coach to see the flaw in this fantasy. The chances of maintaining my 100-yard-dash pace for an entire mile are about zero. Funny thing is, though, millions of us base our planning for important goals like retirement on financial projections that are just as flimsy. When the market cranks out 15% or higher returns as it did during the '90s, we blithely extrapolate those gains into the future. And when stock prices drop and the economy slides into recession, we err on the side of pessimism. Either way, we end up with a false sense of how likely we are to achieve our goals.

This month, we're going to take a look at the art--not science, mind you--of forecasting. Of course, no amount of number crunching will allow you, or anyone, to predict stock and bond returns with 100% accuracy. But by approaching forecasting as a disciplined process rather than a guessing game you can avoid some common mistakes and improve your chances of meeting your goals.

The pitfalls of history. The most common forecasting faux pas is extrapolating past returns into the future. The problem with this approach is that you can come up with wildly different projections, depending on which slice of history you select. Let's say that you're thinking of investing $5,000 a year in stocks to fund your toddler's college education. How much money will you have available for tuition, books, pizza parties and the like in 2017? If you think you'll get the same annualized return over the next 15 years as the last 15--about 13.5%--then you would figure on a college fund of roughly $239,000.

But what if stocks return an annualized 7.7%--as they did from 1969 through 1983? You might be looking at a college fund of just under $143,000. Uh-oh. And if we get 15 years like those ending in 1974, when stocks gained just 4.3% annually, that fund would barely break $100,000. Bye-bye Ivy League.

Averages can mislead. But these examples raise an even more subtle flaw beyond over- or underestimating returns. The flaw is that we assume we'll earn the annualized return year after year. In fact, that's how most retirement and college-planning calculators work. This doesn't reflect reality, however. Stocks and bonds deliver great gains some years, lousy ones in others. And when you're adding money to or pulling money from a portfolio, it's the up-and-down sequence of returns that determines the value of your portfolio. So those precise figures that calculators spit out are really a fiction.

To get an idea of how the order of returns can affect a portfolio's value, let's assume that in 1970 you started investing $10,000 a year for retirement in a mix of 80% stocks and 20% bonds. Over the next 30 years, such a portfolio would have earned an annualized 12.9%. But as the chart on page 53 shows, the size of your account depends on how you earn those returns. If you had earned 12.9% every year--as most calculators would assume--your balance would have totaled $3.1 million. In reality, you would have ended up with a lot more--$4.5 million--because big returns came at the end of the period, when you had more money invested. If you had gotten the same annualized return but the gains had come in the reverse order--1999's first, then 1998's and so on--your portfolio's value would have totaled just $2.5 million, because the big returns arrived at the beginning, when you had less invested. In short, even if you get the average return right, you can end up with more or less than you expect.

How then can one plan in the face of so much uncertainty? Here's the four-step approach I recommend.

Start with realistic assumptions. For reasonable expectations about future stock gains, look first to the two main building blocks of equity returns--earnings and dividends. After all, says Roger Ibbotson, Yale finance professor and chairman of Chicago research firm Ibbotson Associates, "the returns we ultimately get depend on what corporations can deliver." In the short run, earnings jump around a lot. Last year, profits were down 16%, while this year analysts expect them to rebound by about 15%. But over the past 50 years or so earnings have grown at a pace of about 7% a year. Add the recent dividend yield of about 1.5%, and you're talking about a long-term expected return of roughly 8.5%.

To gauge long-term bond returns, you can take the yield to maturity on 10- to 20-year bonds, which is what you'll earn if you buy and hold such bonds until they're repaid. With long-term Treasuries recently yielding 5% to 5.5% and investment-grade corporates yielding 6% to 6.5%, a range of 5% to 7% is a reasonable estimate for returns.

But these figures are just a baseline. If your portfolio includes a dollop of small stocks, you might boost your expected return a bit. Similarly, you might make adjustments depending on your outlook for profit growth. Ibbotson expects earnings to grow at a slightly higher rate than in the past, in part because companies now pay out less of their profits in dividends, leaving more capital for future growth. For this and other reasons, he forecasts a long-term return for stocks of about 9.4%.

The big wild card for stocks, though, is what Vanguard founder John Bogle refers to as the "speculative" component of returns, which is reflected in price/earnings ratios. During the '90s, investors were so eager to own stocks that they bid P/Es into the stratosphere, generating spectacular returns. If you believe P/E ratios are headed back to nosebleed territory, then you might want to add a percentage point or two (or even more) on top of that 8.5% baseline. If a jump in P/Es strikes you as wishful thinking in light of the beating investors have taken the past two years, you'll leave the baseline as is, or adjust it downward. Bogle believes P/Es are more likely to go down than up, so he thinks stocks will deliver annualized returns at the low end of a range of 6% to 9% over the next decade or so.

Think probabilities, not certainties. Whatever return you arrive at, the odds of its being right on target are minuscule. So you've got to think in probabilities, or ranges of possible outcomes.

Say you're 45 and have $140,000 invested in a mix of 70% stocks and 30% bonds. You contribute 6% of your $70,000 salary to your 401(k), your employer kicks in 3% and you want to retire in 20 years with a $60,000 inflation-adjusted annual income that will last until you're 95. What are your odds of pulling this off?

By plugging these and other details into the Retirement Planner tool on MONEY's website (www.money.com), you'll find you have a 66% chance of accumulating the $990,000 or so you would need at retirement to generate the income you want. You'll also find you've got a 90% chance of having at least $540,000--and a 10% chance of ending up with $1.8 million or more. You can also see how altering assumptions such as your savings rate, retirement date or asset mix changes the odds of your reaching your goal.

A few other sites can also give you a sense of the range of probabilities involved in forecasts. Financial Engines (www.financialengines.com; $39.95 per quarter), Morningstar ClearFuture (www.clearfuture.com; $7.95 per quarter) and Fidelity's PortfolioPlanner (www.fidelity.com; available to Fidelity customers only) also use complex forecasting techniques to calculate your odds of accumulating enough money for goals such as retirement. T. Rowe Price's Retirement Income Calculator (www.troweprice.com), by contrast, can help retired investors determine whether they'll run out of money before they run out of time. And one site, financeware.com (www.financeware.com; $19.95 a month), lets you make projections that combine both aspects of retirement planning--accumulating as well as drawing down your assets.

Diversify to increase your odds of success. Spreading your money among a variety of different types of stocks, as well as bonds, not only lowers the volatility of your portfolio; it also narrows the range of values your portfolio might hit in the future. True, that means you'll have less upside potential. But you'll also reduce the chance of ending up far short if your return expectations are off-base.

Monitor your progress. It's crucial that you conduct periodic reality checks to see whether you're gaining on your goals or backsliding. If your retirement savings aren't growing as quickly as you'd forecasted, perhaps a fifty-fifty mix of stocks and bonds isn't going to deliver the returns you need. You might need to increase your stock exposure. Or boost your savings. Or do both.

But if you don't realize until you're about to retire that your nest egg is much smaller than you'd anticipated, you won't be able to do much about it, except maybe latch onto a rich relative with a kind heart. So plan early, check your progress often and make adjustments as needed. Speaking of adjustments, you'll have to excuse me. I've got to shave a second off my 100-yard-dash time.

Senior editor Walter Updegrave is the author of Investing for the Financially Challenged (Warner Books). You can reach him at investing101@moneymail.com.