Two kinds of Fundamental Analysis

The term "fundamental analysis" gets thrown by around professional and amateur investors in ways that create unnecessary misunderstandings and debates.  I have noticed that there are (at least) two different methods or approaches that users of the term may be referring to.  I will call them "modeling" and "screening."

"Modeling" better captures  what traditional fundamental analysis  involves.  It is time-consuming, difficult, and requires substantical experience.  It usually refers to analysis of a company, but could also be applied to an industry or an economy.  The phrase "take it apart and put it back together" describes what goes on with this type of fundamental analysis.  The analyst takes the three of four financial statements (usually just the income statement, balance sheet, and cash flows, but sometimes also the statement of shareholders' equity) and studies them.  By examining historical trends in the company's numbers and comparing them to other indicators like economic and industry data, he looks for a few key relationships that capture most of the variability in the data.  Can revenues be explained as the product of a few volume and price series?  Are there a few cost items that can explain much of the movement in expenses?  Is there a measure of market size that can be used to track a company's revenues as the product of market share and market size?  Can a company's capacity be measured and related to its capital spending and its production?  After analyzing these questoins, the analyst endeavours to create a model, a highly simplified version of reality that condenses the available reported data and estimates of some values that may not be reported into a spreadsheet that shows the relationships between inputs and outputs.  The model can then be used to project financial results into the future.  As each quarterly financial report is released, the analyst compares the acutal results with the predicted results and looks for clues about what has been happening and adjusts the model to incorporate the new information or understanding.  A typical senior, experienced "sell-side" analyst may cover 10 to 20 companies with this type of analysis, doing it full-time.  It is not something that the typical retail investor has the time or the training to do.

For the typical retail investor, fundamental analysis takes the form of what could be called "screening".  Companies can be ranked on sales growth, gross margin, EPS growth, dividend yield, return on equity, or any number of similar variables.  The consensus of analyst-predicted EPS estimates can be used.  This type of fundamental analysis allows the user to process data on thousands of companies in contrast to the dozen or so companies that the "modeling" analyst is limited to.  Sometimes the screening will be done is a series of steps, or phases, initially screening a universe of companies to select a smaller number for further study, and then applying other criteria to the names that "pass" the previous screen.  The phrase "a mile wide and an inch deep" comes to mind.

To illustrate the differences between the two approaches consider the P/E ratio.  For the modeler, the P/E ratio is the price divided by the projected EPS (current year or next four quarters).  The P/E is then compared to the P/E ratios of other companies and a judgment may be made about whether the P/E is higher or lower than it "should" be.  It is a measure of the market's expectations about future growth.

For the screener, the P/E ratio is a measure of value.  A high P/E means the stock might be overvalued, and a low P/E means it might be undervalued.  The screener does not much care about whether the E is past or projected. He wants to use an E that is defined the same way for all the companies.  The screener is just trying to reduce thousands of companies down to a more manageable number.