Summary This begins a series of articles communicating my investment philosophy, strategy, and process. I’ve always used stock screens. They’re basically a necessity and more people use them than those just using explicit screening tools. What you screen for matters, and I like to use EV/EBIT for reasons explained below. Screening and using rules like EV/EBIT are fundamental to adding a passive, systematic layer to my investment process, which I feel complements the deep research and intuition. As a new investment manager, I’d like to begin communicating my investment philosophy and strategy in a coherent way. A series of articles will be part of the process of communicating my process. In this, the first article, I will focus on how I source ideas and why I do it this way. Screening I begin with a stock screen. A stock screen is a query of the universe of public securities. There are tens of thousands of public companies globally. There’s too many for one person to do detailed research over any reasonable period (say 1 market cycle or 5-10 years). Unfortunately, I’m one person, so screening it is. There are some gifted investors who manage to be more than one person and who avoid screens. Warren Buffett famously went through every company in the Moody’s stock manuals from A-Z in his partnership days. Others who avoid explicitly using screens and don’t go A-Z rely on intuition to find companies worth analyzing. For example, perhaps an investor reads every new idea published on Value Investors Club or Seeking Alpha, two popular investment research sites I use later in my process. I’d argue this is still screening, just using non-financial criteria. These investors are screening for securities based on the criteria that they are covered on VIC or SA. Criteria are the heart of stock screening and I think they’re a necessity. Evolution What criteria do I use? Since I began investing several years ago, I’ve tried many different criteria. I began with pre-made screens like the Piotroski F-Score, Magic Formula, Ben Graham’s Net-Nets, and others. Gradually I moved toward making my own screens with tools like finviz.com . After a few years, I moved in a different direction. I’d done a lot of reading about the underperformance of most active managers and behavioral psychology. I’d begun playing poker and thinking about the future more in terms of odds and possibilities than certain outcomes. These and other factors led me to use stock screening for more than just finding stocks with metrics that I intuitively like. I began wanting the screens to return a basket of stocks that, based on extensive historical data, would outperform on average. I became interested in systematic strategies. Helpful books on this path were: Joel Greenblatt’s The Little Book that Beats the Market , which I’ve read several times Tobias Carlisle and Wesley Grey’s Quantitative Value James O’Shaughnessy’s What Works on Wall Street Investing on the long side is not zero sum. Stocks in the US have gone up just under 10% nominally and just under 7% really since the 1800s. But as an active investor, I am implicitly not content with market performance. I am trying to achieve what most active investors covet: long-term, sustained market outperformance. Alpha. When you think of market performance as “zero,” the market is zero sum. From there, it is a good idea to frame the question not as “How do I invest?” but instead “How do I sustainably outperform?” Base Rates I think a big part of the answer is by selecting stocks from baskets that outperform. Surely among the unmanageable tens of thousands of stocks out there, there are many baskets of a few hundred, selected based on various criteria, that have historically outperformed. Indeed there are. The academics are all over this. Here I will highlight and contrast just two though. Momentum One is momentum. This is buying stocks that have increased recently and either selling them when they begin to decline (“trend following”) or just holding them for a designated period like one year. This takes many forms because there are many definitions of “increased recently.” Has it increased in the last minute, hour, day, week, month, 50 days, 200 days, year, or 5 years? In general, I’ve gathered that over periods of measurement less than a year, momentum predicts outperformance. Once you extend it further to 5 years, this actually reverses. Momentum then underperforms and stocks that have performed the worst over the prior 5 years (“dumpster diving”) outperform. Here is some data from What Works on Wall Street to support the claim that momentum has predictive value. I won’t elaborate on the details of the tests but they did seem substantive and compelling: Strategy (from universe of All Stocks) Geometric Average Return 1951-2003 All Stocks 13.00% 50 Best 1 year price performance (“1YPP”) 12.61% 50 Worst 1YPP 4.06% Strategy (from universe of Large Stocks) Geometric Average Return 1951-2003 Large Stocks 11.71% 50 Best 1YPP 14.73% 50 Worst 1YPP 9.11% Strategy (from universe of All Stocks) Geometric Average Return 1955-2003 All Stocks 12.55% 50 Best 5YPP 6.89% 50 Worst 5YPP 16.77% Strategy (from universe of Large Stocks) Geometric Average Return 1955-2003 Large Stocks 11.18% 50 Best 5YPP 8.11% 50 Worst 5YPP 14.16% Valuation Metrics – EV/EBIT Another is valuation metrics. A valuation metric is a metric designed to measure the value of a company relative to something else. Valuation metrics are a price tag. They are what you pay over what you get. I label this general category “valuation metrics” because the one metric I am most interested in is not the only valuation metric that predicts market outperformance. Most valuation metrics have significant predictive value. Low PE and Low PB were identified as having predictive value several decades ago and still have substantial predictive value (read: they still work). But there is one that works better than the rest and that is the Enterprise Value to Earnings before interest and taxes multiple or EV/EBIT. First, what is Enterprise Value? Enterprise value is the true economic price of an entire company. It is the company’s market capitalization (share price x number of shares), with adjustments for the cash, debt, and other obligations the company has. Second, what is Earnings before interest and taxes? This is the company’s bottom line, its net income, with interest and tax costs added back. This is done to make performance comparable. A company’s capital structure (the amount of debt and cash it has) changes and this can also be a point of difference between companies. If we want to compare the operating performance of a company with that of another company or its own performance in a prior year, we get rid of the interest and tax to isolate for what we’re trying to measure. Put simply, EBIT is a purer measure of the profitability of most companies’ operations than any other number on the income statement. Together, EV and EBIT create a very powerful metric because they are both very sound measures of what they independently seek to capture: price and profit. As I mentioned, EV/EBIT is a quite powerful metric. I’ve done the following backtest in Bloomberg: US stocks Excluding utilities and financials Market cap > $20mm Equal weight (about 200 holdings at any one time) 1 year holding period Annual rebalancing Lowest 10% of the market on EV/EBIT From 1995-2015 (furthest back I could go with the test) This strategy generated annual returns of 21.68% versus 9.43% for the S&P 500. There are some other predictors in there like the inclusion of micro-caps, which historically outperform, and equal weighting, which outperforms, but there’s nothing wrong with that given that I don’t size based on market cap in my accounts and am able to invest in micro-caps. The biggest issue with this test is the limited sample size of only 20 years, but that’s all I could get with the data I had. In Quantitative Value, Carlisle and Grey subject EV/EBIT to many things that are “proper” for academic studies, but unnecessary and really detract from performance, and yet EV/EBIT still performs really well. They also did the lowest 10% of the market on EV/EBIT and excluded utilities, financials, REITs, and ADRs, but they also: Excluded any company from the universe with a market cap less the 40th percentile on the NYSE which translated in the study to less than $1.4B in 2011 dollars Market cap weighted instead of equal weighted Nevertheless, they found that the strategy returned 14.55% annually over 48 years from 1964-2011, beating the S&P 500 by 5.03% per year. Further, the top decile (highest 10% of population on EV/EBIT) underperformed the market by 2.43%/yr, so there is a spread of about 7.5% in annual performance between the top and bottom deciles. The most meaningful takeaways there are that the predictive ability still holds up with rigorous testing and over many market cycles (almost 50 years is a good-sized sample). Finally, EV/EBIT is one of the metrics used in the Magic Formula. The Magic Formula takes the 3500 largest stocks by market cap in the US and assigns a number rank to each based first on return on invested capital, a profitability metric, and then on EV/EBIT. So each stock has two rankings. These rankings are added together. The 30 stocks with the highest (smallest number) combined rank are equal weighted and rebalanced annually. According to Greenblatt, this strategy did like 30% annual returns over almost 20 years ending around when the book was first published in. Note that it’s been a while since I last read the book so those numbers may not be precise, but the bottom line is that the results were really good. Some issues here are the limited sample size in terms of years and the size of the basket, but the results are still compelling. Studies trying to replicate Magic Formula have found that the inclusion of ROIC actually detracts from its performance. In other words, EV/EBIT’s predictive ability is driving more than 100% of the performance. But Why? So EV/EBIT and momentum both perform well. But why? I don’t think it is enough only to have historical predictive value. It also makes sense. The test I use is “would I look for this if I were analyzing any one stock or business for prospective purchase?” If it doesn’t make sense but looks good and we go with it, we assume a major risk: data mining. One study attempting to illustrate data mining found that 99% of S&P 500 movement over 12 years was predicted by butter production in Bangladesh. Correlation does not equal causation. Past predictive value does not equal future predictive value. Source: Forbes And this is why I really like EV/EBIT. It makes sense. If I were looking at an individual company, a low EV/EBIT would look very appealing to me. In fact, I often value stocks, in part, using this metric. It also makes sense that buying things at lower prices is a good strategy. Momentum does not make as much sense to me. Why buy things now when it’s gotten so much more expensive? Except for certain luxury items, the appeal of most products decreases as the price increases. Conclusion So this is a big part of my process. I screen based on EV/EBIT, generate a list of a few hundred companies, and go through them one by one. There is intuition involved, but I’d say the list generation process is pretty systematic. Both are important and I like where my strategy is positioned. There are elements of both deep analysis and disciplined rules in my process and I think that’s a good place to be. I don’t know if I’ll always be using EV/EBIT and I doubt it will always be my primary focus, but I think the more important point is to have a defined process that makes sense, and, for me, to stay positioned at the crossroads of active and passive investing, rules and intuition as the lines between these seeming dichotomies blur in the future.