My email inbox over the past two years has often resembled a cavalcade of sorrow and regret: Physicians, engineers, teachers and ad salespeople have bared their souls to admit repeated and unexpected failure as investors -- buying tech stocks near highs in 1999 or 2000 and holding on as the market blasted them to bits.

I have always found these confessions intensely curious because they explode most assumptions about the link between speculation and emotionalism. These were not foolish people; they were logical, tenacious, high achievers who planned their long-term trades carefully. How could they fail so abjectly at the task of investing, which seems on the surface to reward the rational?

The answer is that the market does indeed reward reason, but in ever-changing ways both hidden and extremely complex. Like the locked-down rules of convention or science found in law and physics texts, the rules of stock investing begin from a steady and sensible base. But the rules can also mutate quickly, paradoxically and capriciously. Nimble, experienced traders who observe and act upon these rule changes with ease extract huge sums of money from the market in short order. Investors who instead play strictly by a fixed set of rules -- however useful and immutable they might once have seemed -- are often their chastened counterparts, the losers of that money.

My quest over the past few years to define a few sets of rules for market behavior led me to create the stock screens and mechanical portfolio-building techniques that have gone under the rubric "SuperModels" in this column. Many have worked very well; yet I've been frustrated by those ever-changing cycles that, with increasing speed, turn once-useful insights into useless trivia.

Recognizing the limits of my ability to capture the nature of change in the market, a year ago I embarked on a new mission: To find a crack team of independent stock researchers that could define a dynamic group of rules that would bend fluidly with the market's extreme turbulence. Rather than codifying a fixed set of familiar precepts about the link between stock prices and earnings growth, I wanted to see if we could use the blazing new power of computers and software to find convention-shattering relationships buried below the surface of market movements. In short, I wanted to help create a new sort of stock-assessment system that embraced volatility, rather than ran from it, and that took a fresh approach to the standard one-size-fits-all market wisdoms.

Presenting StockScouter

The result of these efforts is before you now in StockScouter, a rating system that launched Friday in the new version of

CNBC

on MSN Money that we believe will help individual investors quickly assess a stock's potential for outperforming the broad market. Working closely with Camelback Research Alliance, an autonomous research firm in Arizona at the cutting edge of financial engineering for institutional money managers, we have identified statistically predictive traits that affect the performance of successful U.S. securities and developed a systematic way to help you discover, research, hold and sell them.

StockScouter, like similar systems that cost Wall Street pros hundreds of thousands of dollars per year, depends on advanced mathematics, an innovative mix of measurements and historical testing to attempt to forecast the short- and long-term outlooks for all U.S. companies that have traded on one of the three major exchanges for at least the past six months. As of July 2001, that represented a universe of about 6,200 stocks. (We may be adding foreign stocks soon.)

In rating the outlook for stocks from strong to poor on a 10-point scale, StockScouter does not make subjective judgments. Instead, it compares the fundamental and technical qualities of individual companies and their stocks with benchmarks that have proven statistically predictive of stock performance in the past. It then assigns an expected six-month return to each stock based on this statistical profile and then balances that return against expected volatility. This ratio of expected return to expected volatility, or risk, yields a stock's final overall rating.

The balance of expected reward with expected risk is a key concept that sets StockScouter apart from other stock-rating systems, such as ones published by Value Line and Standard & Poor's. Stocks with expected high future returns would see their ratings reduced if the volatility of those returns is also expected to be high. Thus an ideal stock in our system is expected to move briskly and directly to a higher price, rather than simply briskly.

StockScouter is not perfect; no predictive rating system could be. But its models have been thoroughly tested to the highest academic and professional standards using more than a decade of historical data -- a period of time rich with variation. And we believe that, in conjunction with the other stock research tools available on MSN Money, it will help individuals make more thoughtful investment choices by revealing the way successful professional money managers accumulate, sort, evaluate and act upon financial information.

How StockScouter Works

StockScouter rates each stock in its coverage universe from 1 to 10, with 10 being best. Ratings are scored on a bell curve. This means that there are fewer stocks with a rating of 10 than with a rating of 9 and fewer stocks with a rating of 9 than a rating of 8; likewise, there are more stocks with a rating of 2 than with a rating of 1. Stocks thus bunch up in the middle, which is why stocks rated 4 to 7 are expected to perform more or less in line with the market. Ratings are recalculated and updated in our database daily to reflect the most current technical, fundamental, ownership and valuation data available.

The ratings derive from the Camelback research team's analysis of four key factors that reflect either a company's fundamental quality or investors' ardor for its shares. The system rates each factor's performance on a five-point scale from A to F, just as you'd see on a report card. These grades appear on StockScouter's Details pages along with a list of the subfactors that make up each factor. Subfactors do not get grades of their own; rather, the system tells you whether they add a positive, negative or neutral bias to the factor grade. The factor grades are also displayed for quick reference on StockScouter's Summary page for each stock, and a pro/con list also summarizes the subfactors' findings. Here are the four factors and some of their key subfactors.

Fundamentals

. This factor assesses a company's past earnings growth, its estimated future earnings growth and its capacity to beat brokerage analysts' consensus estimate. To receive a high grade, a company generally must grow reasonably fast, beat analysts' growth estimates and be expected by veteran analysts to grow earnings in the future. However, a powerful earnings surprise or big boost in estimates by an experienced analyst can also boost the grade of a stock with otherwise seemingly lackluster fundamentals. (Click

TheStreet Recommends

here to see the top-rated stocks with top-rated fundamentals.)

Ownership

. This factor assesses whether a stock is under accumulation by executives and board members. To receive a high grade, a stock generally must be under accumulation by high-ranking executives or board members in significant quantities. (Click

here to see the top-rated stocks by ownership grade. Caution: Insider trading information can sometimes be a week old.)

Valuation

. This factor assesses whether a stock's price is high or low relative to its current level of sales, earnings and expected earnings growth. It's counterintuitive, but our research shows that big companies' shares should be slightly more expensive than peers to receive a high grade. Meanwhile, as you'd expect, small companies' shares must be much cheaper than peers to receive a high grade. (Click

here to see the top-rated stocks with top-rated price action.)

Technical

. This factor assesses whether a stock's price trend is positive or negative. To receive a high grade, a stock's price must either be rising over the past six months or must have risen on more days than it has declined over the past six months, or both. However, a stock with a very low price relative to its 10-week trend can receive a high grade if the system believes it has plunged to a level from which it is likely to rebound. (Click

here to see the top-rated stocks by technical grade.)

Computing 'Core' Ratings

StockScouter does not simply average subfactors to yield a factor grade. Instead, subfactors are added together in different weights that vary with a stock's size and sector. Likewise, factor grades are not simply added or averaged to yield a final overall rating. They are also weighted in a variety of ways according to a proprietary methodology developed by the statisticians and financial engineers at Camelback.

After all of the factors and their constituent subfactors are measured and weighed, StockScouter awards each stock a core rating. The highest core rating would generally be straight A's; next would be three A's and a B, and so on. That core rating is then balanced against the standard deviation, or volatility, of the stock's price change over the past 12 months to yield the final overall rating. It's convenient to think of the overall rating as the ratio between expected return and expected risk. A stock with high expected return and high expected risk will generally receive a lower rating than a stock with modest expected return and very low expected risk.

A portfolio managed with guidance from the StockScouter system would attempt to stay invested in stocks rated 8 to 10 with high factor grades, and avoid stocks rated 1 to 3 with low factor grades. More specifically, the Camelback team has developed a variety of portfolio strategies that I will share with you over the next few months and beyond. They are not always strictly straightforward, but a little complexity is the price to pay for the system's potential power.

Come back next week, when I will explain my plan to use StockScouter to build a portfolio for the year 2010, six months at a time. Then on Aug. 8, I will explain the powerful

Market Trends Feature of our system to show you how to build a portfolio optimized for the short time frame of one month.

At the time of publication, Jon Markman did not own or control shares in any of the equities mentioned in this column.