What Quant Value Models Can and Can't Do for You

By Justin Carbonneau (@jjcarbonneau) –

Two weeks ago I wrote an article, Differences in Value, where I outlined the various criteria (i.e. price-to-sales, price-to-book, and enterprise value-to-EBIT) used in the value models on Validea. As the piece got passed around Twitter, there was a very good comment by Tren Griffin, an investor and thoughtful writer.

Griffin tweeted the following (see below – included in the tweet was a screenshot of a page from Berkshire’s Hathaway’s 1992 letter where Buffett discusses value, growth and the concept of value investing and buying stocks below their intrinsic value based on future cash flows (and not based on metrics like the P/E) – I encourage you to read that and/or the full letter here):

What Griffin is saying here is that there is a night and day difference between being a “value” investor of the Graham or Buffett mold vs. a systematic (or factor-based) value investor and relying on models that use value metrics or ratios, similar to the ones on Validea.

A few years ago Griffin wrote a very good piece — Ben Graham’s Value Investing ≠ Fama/French’s Factor Investing — explaining why buying stocks based on measures like book-to-market (the inverse of price-to-book) is nothing like the Ben Graham / intrinsic value way of investing.

His comment and his previous piece got me thinking about how important it is to understand what quantitative value models can and can’t do. Since we run multiple quantitative models based on investors like Graham, Buffett and 20+ others, I have a bias towards using these models in stock selection. However, Griffin brings up excellent points and it’s important to flesh out the differences between quantitative value approaches and those that go much deeper into the world of true value investing (aka. Graham and Buffett).

Let’s start with the “CAN’T Do” first:

  • Most models aren’t a replacement for deep fundamental analysis: I was reminded of this after watching two excellent podcast episodes, both on Tobias Carlisle’s Acquirer’s Podcast. One with Joe Calandro, author of Applied Value Investing, and the other with Tim Travis of the value shop, T&T Capital Management.
    • In Calandro’s book he walks the reader through various M&A and valuation case studies, including examples of Eddie Lampert’s Sears acquisition and Berkshire’s purchases of GEICO and Gen Re. Calandro outlines a modern day Graham/Dodd value continuum in which a company’s assets, earnings, franchise value and potential future value are computed. This requires a very detailed look at the balance sheet, income statement and an assessment of any sustainable competitive advantage a firm may have. Many of the items on the financial statements need to be adjusted and assumptions need to be made about future profitability and growth. Calandro’s framework comes out of Columbia Business School’s value investing class, led by Brue Greenwald, the Robert Heilbrunn Professor of Finance and Asset Management and one of the preeminent teachers of value investing.
    • Another good example is Tim Travis’ talk about his investment and value thesis on Assured Guaranty. This is a good example of how deep value investors go to understand the business, the profit drivers, the risks, valuation and where the market may be getting it wrong. When you dive into a company this deeply, there is a level of conviction that develops over time. The conviction level manifests itself in the portfolios of successful value investors. I wrote about Buffett’s level of conviction with Berkshire’s portfolio here, “How Active is Buffett’s Portfolio?
  • They can’t judge the character of management or culture: Quant models can’t assess the character of management or value of a company’s culture, experiences, stories and narratives that shape a company over time. No quant model I know of could have looked at Steve Jobs and predicted the ultimate success of Apple, just like no quant model could have analyzed Herb Kelleher of SouthWest Airlines and the culture that developed there. And yet, we know from history, stock price performance, and the details behind the CEOs of great companies that the environments they helped shape and the qualities they possessed are important drivers of success for companies. Buffett has shared a few yardsticks on ways to evaluate management, but these are far from quantifiable in any type of model.
  • They can’t predict the future of an industry or a business: Systematic models, at least the ones we run here at Validea, can’t predict the future of an industry or the future of a business. The decline in energy stocks in 2015/2016, the pain in retail, fall of General Electric, the underperformance of value stocks, a result of sectors like energy and financials lagging (see Lawrence Hamtil’s piece, “How Inflation Makes the ‘Value’ Factor a Sector Bet“), are not things that can be calculated and easily input into a model.
  • They’re not a guarantee for market beating performance: Even the best, most researched quant models are not a guarantee for market beating returns. Yes, selecting stocks by a value “factor” has been shown in academic tests to produce outperformance over an index weighted by market capitalization, but it’s not a sure thing. Timing and luck play a huge role in whether or not an investor has the potential to achieve those returns. Griffin brings up another point, and it’s worth pointing out, that many quantitative models that use value metrics are benefiting from the “halo” of elite value investors like Graham and Buffett, even though the two methods are in fact extremely different and largely independent of each other.

Now let’s look at a few things quant value models CAN help with.

  • They can help you source investment ideas: Even though the number of stocks in the U.S. has been on the decline, there are roughly 3,000 securities that are investable by our calculations (we use a series of liquidity requirements that keeps our models out of microcap securities). There is no way investors can look under every rock, so screens and filtering mechanisms can help. For instance, let’s say you want to find companies trading a certain valuation (or better yet multiple value metrics like Jim O’Shaughnessy’s VC2 model), or with certain earnings and sales growth rates, or levels of debt, returns on equity or dozens of other fundamental criteria that suits you as an investor. Quantitative screening can help you take a list of hundreds of stocks and narrow it down to a manageable starting point.
  • They can be used to help implement discipline and consistency: Implementing and following a quantitative investment model in the context of a portfolio isn’t easy, but if an investor truly embraces a systematic approach there is no doubt it introduces a level of discipline, consistency and repeatability into the investment process. That isn’t the same as saying the model will go on to outperform the market, but it is saying that the quantitative model has the ability to avoid many of the emotional pitfalls and biases we have and either know or don’t know about ourselves. I’m not going to go through all the behavioral biases here, but one very specific area that detracts from returns is a sell discipline (Barry Ritholtz pointed this out in a piece for Bloomberg earlier this year, “Stock-Pickers Don’t Know How to Sell“). Rather than selling based on a gut instinct or feeling, a quantitative model sells if something in the fundamentals or valuation has changed or if there is a better scoring opportunity in the investable universe.
  • Quantitative models can combine investment fundamental investment criteria and don’t need to key in on one factor only: Griffin’s article focuses on one type of value factor, book-to-market, which is the leading factor tested in academic studies. He’s right to focus on this since when people say “value has outperformed over the long term”, they are often talking about the cheapest stocks based on the book-to-market factor. However, there is no rule of thumb that quantitative value models have to start and stop with one factor. Validea’s models all take a number of criteria together – many are combining value with other variables from the balance sheet, income statement and try to analyze a stock as much as possible in the way a human would. In the Validea system, we go even further in implementing logic that makes adjustments in the calculations like an analyst might and layers in quality overlays like my partner Jack discussed in this piece, “The Double Edged Sword of Avoiding Value Traps“.

While the lines between quantitative value investing and real value investing continue to blur, it is important to understand the differences between the two. Both have their advantages and disadvantages, and which one works best is very dependent on issues specific to the investor who is implementing the strategy. For those with the skill to deeply analyze companies and the ability to control their emotions, traditional value investing can work very well. For those who cannot do those things, a disciplined quantitative system could be a better solution. As is the case with many things in investing, the answer to this question is more gray than black and white.

Photo: Copyright: 123rf.com / ismagilov

Justin J. Carbonneau is Partner at Validea Capital Management and Validea.com. You can follow Justin on Twitter @jjcarbonneau.