TMF: Tell us about your investment philosophy at Legg Mason.

Gay: We begin with the premise that we believe the market is generally efficient in the aggregate. But even within an efficient market, mispricings may occur at the stock level, when stock prices diverge from business value. And that occurs because of sentiment or news flow or short-term events -- some major disruption, whether it's a power failure or a terrorist action. It can cause prices to diverge from the long-term business value. We believe prices move around a lot, but business values don't move around a lot.

The research focus is identifying companies' intrinsic business value and trying to buy the stocks at a substantial discount to our assessment of what that business value is. The largest weights in the portfolio are really in the companies that we have the highest risk adjusted three- to five-year return outlook for. It's not an accident that we allow certain securities to rise or fall within the context of the portfolio, because at all times, the portfolio, we believe, is diversified from a valuation driver standpoint. The largest weights are really in the companies we've got the highest confidence will generate the best returns on a risk-adjusted basis.

TMF: How do you find those companies?

Gay: The stock approach is really a step-by-step approach where we identify sources of ideas. The sources come from lots of different places. We do quantitative screens where we're looking for low absolute accounting metrics, like a low P/E or low price to book, low price to cash flow -- the same type of things you would expect to find from what typical value managers would do. But we're also trying to observe what's going on in the market. Are there any unusual volume spikes in certain names? Are there big volatility moves in certain names? Is there a sudden surge of insider buying or selling? Is there a significant change in the corporate governance of a company?

Bill [Miller] has been, we believe, at the forefront of many managers in the institutional world, to try to push forward the idea that corporate governance is critically important to generating good shareholder returns. And unless you have a long-term orientation and vote your proxies very aggressively on behalf of your shareholders, you can destroy a lot of value fairly quickly. We've been adamant and out in front of many issues. Fortunately, Bill knows (Berkshire Hathaway (NYSE:BRK.A) (NYSE:BRK.B) chairman) Warren Buffett fairly well and knows Chris Davis at the Davis Funds and knows Jack Bogle, the founder of Vanguard. Bill's had numerous meetings over the past couple of years with Davis and Bogle. The corporate governance issue is really critically important. There's a lot of academic research that supports really great returns from companies that have really good long-term governance.

The screen process identifies lots of things that might be unusual and might indicate that there is a disconnect between the company's price and what might be going on from a value basis.

TMF: Assuming you've now found a company that looks interesting, what's next?

Gay: Once we have this universe of ideas that look either statistically significant, that warrant further investigation, or from an information standpoint, warrant some investigation, then it goes into this initial assessment where we ask the question, "Well, it looks cheap. Is it cheap?" That's really where we need to understand the business model. So we pull apart the company's profitability.

The Value Trust itself is a little over a $14 billion portfolio. But as I mentioned, I manage funds that are also indexed to the Value Trust, and we have separate account outfits. So in total, we're managing about $30 billion in one portfolio. Now that's a lot of money. In 34 stocks, that means that we're investing in general $600 million to $1 billion dollars in each individual name, which is quite a lot of money to invest in. They're mostly large-cap companies, obviously, because that's what our mandate is. But we don't want to trade them. We probably couldn't trade them, but that's just not where we think we can add value. Therefore, we really need to understand what the company looks like over a normalized period, not just one year or three years.

It's important to understand whether they're cyclical or not, but also to understand the secular characteristics of the company as well. So we really want to understand what the company's position is in the market, what their position is in their industry, and look at the management and how the management is incentivized. And that gets back to the corporate governance issue, because managements that are paid to beat next quarter's earnings, are probably not going to really care how capital is allocated based on a three- or five-year project. So we really want to make sure as much as possible that we're investing with companies whose interests are aligned in long-term value creation for their shareholders.

The other assessment that takes place at this stage is really to adjust the accounting numbers to reflect the economic reality of the business. What I mean by that is we need to look through the numbers and look at the treatments that the company has adopted, like revenue recognition methods. We need to understand what their pension situation is. Is it fully funded, or is there a significant liability that they're going to have to use their cash to meet. Do they expense options or not? Because if you compare one company that expenses options with another that doesn't in the same industry, a simple P/E comparison is irrelevant because the numbers won't mean anything. We look at normalized profit numbers. We look at the corporate structure. So if it's a really big company like Tyco (NYSE:TYC) that has several different business units, you could simply put a conglomerate multiple on it. That's one way of understanding it. But it's much more important to look at Tyco based on how it does in the fire and safety business, how it does in the electronics business, how it does relative to other health-care businesses -- because it has businesses that operate in very different industries. So it's important to pull apart companies on an unconsolidated basis to really understand what they sum up to when you're trying to value the business.

The third stage, which is really the critical stage for assessing whether a company is an appropriate name in the portfolio, is building the model. The model provides the analyst a framework for understanding the business, both historically as well as going forward. Typically, they begin by populating a model going back 10 years, as if the company existed as it currently exists. For example, for AOL (NYSE:TWX), they would go back and say, "If AOL-Time Warner were one company, what did those businesses look like over the previous 10 years?" Even if that was not the case, that they were one company obviously, just so the analyst understands this is the set of financials we have to work with today. This is where they came from. It will give them a picture of how capital was allocated through those businesses, what the margins were. Ten years gives you a more normalized picture, so if you went through a recession or a recovery or a big boom or more aggressive or mild spending periods, it will give you a better picture of what the company looked like.

On a going forward basis, the analyst then will build the model out 10 years. And again, this gives them a much more normalized picture of what the company is likely to do for the next 10 years. Obviously, the further out you go in time, the higher the degree of uncertainty. No one could predict, whether it's AT&T (NYSE:T) or a Coca-Cola (NYSE:KO) or an IBM (NYSE:IBM) or an Amazon (NASDAQ:AMZN), what the returns are likely to be in three years, five years, 10 years. The further out you go in time for any of those businesses, the higher the degree of uncertainty. So the analysts spend a great deal of time running through scenarios and assigning probabilities to those scenarios. We believe that increases our margin of safety, that within the analytical process, we've captured what is likely to take place over the next 10, 15 years. It allows the analyst to understand the relationships between growth and returns and normalized returns. And it also has an intense focus on free cash flow -- when the company will become free cash flow positive, what the returns on capital allocation are likely to be.

We've found over time that there is a very high correlation between companies that have returns above their cost of capital and can sustain those returns on invested capital above the cost of capital, and a rise in the market value of the businesses. There's a much higher correlation between that than simply identifying whether a company is trading in a low P/E and hoping it goes back to a market multiple. Whenever we look at the academic evidence for it, it's almost like a scatter graph, where there's no correlation between simply buying a company on that one low P/E factor and hoping that determines whether it's going to go up in the future. Sometimes, businesses change fundamentally, and they deserve to be low P/E stocks. Other times, companies that are extremely high P/E stocks are actually very cheap.

When we're going through this stage of building the model, there's also a heavy focus on liquidity analysis. Especially when you're dealing with companies that haven't quite reached a level of cash flow generation, you really need to understand how they're allocating capital and what the returns are and when they're going to get to cash flow positive. There's a heavy intensity on what just the profit picture is of the business -- long-term profitability -- and what the competitive advantages of the business are and what the sustainability of those competitive advantages are.

We will do lots of different valuation analyses on the business: liquidation analysis, sum of the parts analysis, private market value analysis, LBO analysis. We will typically do five to seven different valuation cuts at a business, even a simple peer group value analysis, looking at it relative to everybody else in its industry. And even what we call cross-sectional analysis, where we look at it for relationships for other companies with similar accounting characteristics, to say, "What does a market pay for a business that generates a certain level of returns with a certain level of growth with a confidence level that will be sustained for the next five years?" [Note: Bill Nygren talked about a similar process.]

Basically, what we're trying to come up with is a central tendency of value. So each of these values that come up after the valuation work is done gives us a range of an estimated intrinsic value for the business. The tighter the clustering, the higher the confidence. For example, if we think Tyco under certain scenarios is worth 20, but under other scenarios is worth 100, we have lower confidence than if we think under the range of scenarios it's worth 25 to 40. So the tighter the clustering for any of these analyses, the likelihood that the weighting within the portfolio will be higher.

Read Matt Logan's complete interview with Mary Chris Gay:

Are you on the hunt for value stocks? Join Philip Durell in his search for deeply undervalued companies in Motley Fool Inside Value. Each month his newsletter covers two companies that represent real value for investors. Start yourfree 30-day trialnow.

Fool contributor Matt Loganowns shares in Berkshire Hathaway but none of the other companies mentioned. The Motley Fool has adisclosure policy.