The philosophy that we have is unchanging. We explain what we do by separating strategy, process and philosophy because we believe that there are core truths to investing that do not change. That's what we call our investment philosophy. Our philosophy isn't about our daily work or how we organize our departments or how our research process works-instead our philosophy is about what we believe. "We hold these truths to be self evident...." If there were a Legg Mason cheer, it would be made up of these philosophical points. If we were Walmart and we had a song to sing in the morning, this would be it.

The first one reads, "we believe that the value of a company is the present value of the future free cash flows." This is the textbook definition of value. It's not something we made up; it's something that everyone learns in business school. It's very well-documented in all of the theoretical literature. We don't believe, however, that value is a low P/E ratio or some other symptom that you see in the market.

We also think that very broadly, over the long term, the market is efficient. We're not saying that we believe that the market is inefficient and that because we're so smart we're going to find all of those inefficiencies. We are saying that over the long term the market will reflect a company's value - the present value of its future free cash flows.

How is it that we can outperform this very efficient market? Well, we also believe that at the extremes there are mispricings. Stock prices diverge from business values due to behavioral factors. All of the human interaction in the market, all of the different agents trying to make some money, creates some inefficiencies because of the way human nature is and the way people react to news. If you think about the market and a general distribution of values across the market, most of the companies are priced about where they should be. There is some noise around that central area where you see 10-15% size moves around the value of the business. But we're not particularly interested in that area. We're interested in mispricings way out on the ends of the curve that happen because people over-react to news or are engaging in short-term sentiments or are taking the latest data point and overrating it with respect to the long-term story or value of the business. These factors cause stock prices to diverge temporarily from what we think their intrinsic business value is. But because the market is efficient over the long term, those mispricings will return to what we believe the true business value is.

We also believe that the market is adaptive. We are not rules based. We don't think that we have to do something the same way all the time. Strategies that worked twenty years ago may not work any more. There are no simple rules for investing. They will change because other people are trying to make money on those rules as well. So we have to be adaptive in our choice of information sources, and in the way we think about our strategy.

These points comprise our philosophy. How do we implement a process that capitalizes on these tenets?

We have a step-by-step approach to finding mispriced securities. It is not a formulaic endeavor. The four-step process, illustrated below, is an attempt to present in simplified form what is actually a very complicated process.

Where do we get our ideas? Like most money managers, we do screens. You may read articles about how some managers start with the universe of 5,000 stocks, then screen for certain factors to get the list down to 100 stocks, and then conduct rigorous research to identify the 12 or so that they want to invest in. If it were that easy, that would be great, but it's not. Nevertheless, there are elements of that process that are useful. So like most money managers, we start with quantitative screening. We look for low absolute accounting metrics because sometimes that will tell us that something is mispriced. Every week we run screens on different accounting metrics like low P/E ratio, low price to book, and other similar accounting metrics. Then we ask the question, "this is where the stock is priced, but should it be there?" And in most cases the answer is, "yes, the stock has a low P/E because it deserves to." But occasionally something will show up on that screen that doesn't belong there.

We also look at relative valuation discrepancies. Many businesses are at high P/E's because they deserve to be. They have high growth characteristics or very good economics. Or they may not have P/E's. So we need to look at other measures. For example when we bought Nextel last year, they didn't have any earnings or free cash flow. But we found it attractive because compared to other subscriber-based businesses it looked like Nextel was very mispriced. It was trading for about $1,000 per subscriber and the other subscribers were trading for about $2,000 or more.

We also look at unusual market volume, price moves or volatility. Recall that human behavior is one of the reasons that stocks are mispriced. There could be overreactions to news, for example. We look for very large downward moves that occur over a day or a week. We look at everything that's at a new 52-week low. We look at everything that has very unusual market volume. We also look for insider buying and selling because insiders often know their companies better than anyone else. In particular we look for insider buying because as Peter Lynch says, "there is only one reason to buy and many reasons to sell."

We also examine qualitative factors. Our team spends a lot of time in the field looking at news flow, information from conferences, magazine articles. We're looking for business model transitions. The market often misprices a company when it has changed itself for the better. One of the companies in the Special Investment Trust is DeVry. We became interested in them after they announced a change in the way they were allocating their capital-from large university centers to a focus on on-line education. This indicated a more efficient capital allocation model. Another example is corporate governance changes, or when companies announce changes in the way they're allocating options. These types of actions indicate that management is doing something to change the business and the market may not recognize it yet. We're looking at the impact of news events (we like to joke that we look at anything negative that's on the front page of the New York Times three days in a row).

We also look at weak links. There is some work on social networks done by Duncan Watts that addresses the way people interact-how information transfers between people. For example, one study looked at how people get jobs and discovered that most of us don't get new jobs from a friend, but from a friend of a friend. In order to create a diverse set of ideas and opportunities, we consciously expose ourselves to alternative sources of information through vehicles like the Santa Fe Institute, industry conferences, other investors, reading magazines that are not investment based, and so on. This is how ideas bubble up through the system.

Once an idea bubbles up and it looks like it has potential, we put it through a stress test. We ask of the stock price, "this looks like it's cheap, but is it?" Often the stock will look attractive but after a little bit of work it proves otherwise and we discard it. There are two factors we use for the stress test. First we want to understand the story about the company. We want to understand its business model and the economics that make it look like its depressed. Second, we look at the accounting and make any adjustments that we need to. It may look like it has a low P/E, but once we adjust for unconsolidated assets or future liabilities or phantom liabilities for example, we may discover that interest in the stock is unjustified.

If the stock gets through this level of analysis we take it to our next phase. We build a model. We have a twenty-seven-page Excel model that crashes our server every time we run it. But it's not the kind of model that you enter data into and then it spits out the answer. Instead it's a tool we use to understand the business and therefore try to understand what the business should be worth. It includes the income statement, balance sheet and statement of cash flows all tied together. We build the historical part of the model and this helps us understand what the relationships have been between growth and capital. We want to know why capital has gone up or down in one period or another and why free cash flows were at a certain level. So by trying to understand the variable costs and capital needs and other economic characteristics of this business we prepare ourselves to make a forecast. We end up with a tool that helps us talk to the company about what they think their long term profitability should be in relation to their historical performance, what their competitive advantages and position in the industry are, what the structure of their industry is, what regulatory issues concern them, and what their model of corporate governance is.

Once built, the model allows us to make various forecasts, think through different scenarios, consider probable outcomes, and then based on the scenarios, determine a future value or worth. The model also gives us a tool to look at what cash flows will be and relate this back to liquidity issues, debt service, debt rollover, whatever is required to get through a period of liquidity crisis.

Once we've built the model, created a number of scenarios and selected the most likely scenario, we can determine what we think the company is worth. To do this we start with accounting methods. We now have a much better understanding of all of the accounting issues around the company and we are now in a position to ask ourselves what the right P/E should be for this company. We also look at the company not only with respect to its peers and to its history (the most common measures), but also in its relation to the market as a whole. This is important for big transitions. For example, when Value Trust held Dell, the company was a traditional value stock. We bought it at five or six times earnings and it grew to ten-eleven times earnings. Most value investors then started selling it because that's what PC companies sold for. At the time we couldn't understand why a company with 50% top line growth and 75% return on invested capital should trade for only eleven times earnings. As we looked across the market it became clear that Dell could justify a much higher P/E, and it did. This is why we focus on the position of the stock within the market as a whole. We also go through a variety of other valuation exercises.

To recap so far, we create the model. The model generates different scenarios. Those scenarios can be valued on a discounted cash flow basis. That is one approach that we use. But as any investment banker knows, a discounted cash flow can be made to say anything you want it to. So we have to triangulate that against other techniques to see if it makes sense. So we use a variety of other techniques. We do a liquidation analysis. For example, if you sold off all of the hospitals of a company, how much would they be worth? We'll look at a sum of the parts analysis. With Time Warner, there are six different businesses with six different economics and they are all evaluated on a different set of metrics. We'll look at private market value if that's appropriate. All of these tools help us make the valuation more robust. We fold all of this information back into the discounted cash flow calculation via various scenarios and we put probabilities on the different scenario outcomes.

Next we derive our central tendency of value. The terminology is important. We don't have target prices. It's important to differentiate central tendency of value from target prices. The central tendency of value is adaptive. It can take in new types of information over time and feed it into the scenarios to change the probability of each outcome, which in turn will change the value. Our framework is designed to adapt to the evidence. We also incorporate a cross-sectional analysis. If we find that under LBO, DCF or P/E analysis the value looks to be, for example, $50, then we have a high degree of confidence that $50 is the value of the company. If these different analyses yield disparate results-even if the central tendency of value averages $50-we'll have less confidence in that value than if the results were more tightly grouped. This approach also plays in how we construct our portfolios. Instead of a single point estimate of what a company is worth, we have at our disposal a framework that allows us to think about what makes that value happen-the risks and drivers associated with it. For example, we may have a high degree of confidence in the value for a certain company and understand the drivers extremely well. But these risks and drivers may be very different from the rest of our portfolio and this will influence how much of the company we want in the portfolio and also how we will manage the diversification of these drivers and risks at the portfolio level.

That's an introduction to our philosophy and our process, and we apply them across all of our portfolios.