Chapter 9

If you are going to invest your money according to a system, you should be very clear about two things. First, does it work only in theory or also in real life. Second, does it rely on pure luck or on solid, repeatable effects. This chapter is the X-ray image of the fair value strategy. It is where academic work meets my own backtests, live tests and the data that powers the Fairvalue Calculator.

Since the 1970s, empirical studies have repeatedly shown that cheaply valued stocks tend to deliver higher returns than expensive ones. In the early 1980s, Sanjoy Basu analysed NYSE stocks to see whether companies with low price earnings ratios performed better than the market. His conclusion was clear. Low P/E stocks earned persistently higher risk adjusted returns than high P/E stocks, even after controlling for company size (Basu 1983, Journal of Financial Economics).

In the early 1990s, Fama and French showed that two very simple variables explain a large part of the differences in stock returns: company size and the ratio of book value to market value. Small, cheap companies with low price book ratios generated significantly higher long term average returns than large, highly priced growth stocks (Fama and French 1992, The Journal of Finance). This work firmly established the so called value effect in academic finance.

Further studies such as “Contrarian Investment, Extrapolation, and Risk” by Lakonishok, Shleifer and Vishny show that investors tend to extrapolate good news too far into the future and overreact to bad news. These overreactions create exactly the kind of mispricing that value strategies exploit. You buy where mood and price have become too negative, even though the fundamentals are still sound (Lakonishok, Shleifer, Vishny 1994, Journal of Finance).

De Bondt and Thaler argue in a similar way. Their overreaction studies show that extreme loser stocks often deliver strong rebound returns in the years that follow (De Bondt and Thaler 1985, Journal of Finance).

For a fundamental fair value strategy, the work of Joseph Piotroski is particularly interesting. He asked whether you can separate the fundamentally healthy companies from the rest inside an already cheap universe of low price book stocks. His answer was yes. With a simple nine point scoring model, the Piotroski F Score, based on classic income statement and balance sheet metrics, he was able to improve returns of value stocks by about 7 to 8 percentage points per year over the period from 1976 to 1996 (Piotroski 2000, Journal of Accounting Research).

Frankel and Lee looked at what happens when you model the intrinsic value of a company and then systematically buy stocks whose prices significantly lag that intrinsic value. Their result is clear. High value to price ratios predict significantly higher future returns for several years. In other words, when estimated intrinsic value lies well above the price, the price tends to catch up over the following years in statistical terms (Frankel and Lee 1998, Journal of Accounting and Economics).

From these and many other studies you can extract a common core. There is a value factor. Cheap stocks outperform the market over long periods. Quality matters. Strong balance sheets, solid profitability and conservative investment policies improve the outcome. Markets overreact in both directions and tend to revert towards intrinsic value over time. These three pillars, value, quality and mean reversion, are the foundation of the Fairvalue Calculator method.

Academic research often remains abstract. I wanted to know what a concrete, everyday fair value strategy can actually deliver in practice. That is why I have built my own backtests, live tests and samples over the years, now based on more than 50,000 stocks that are monitored continuously in the Fairvalue Calculator database. The valuation side includes models such as P/E, price sales, price book, EV/EBITDA, discounted cash flow, the Graham number, Buffett style fair value, Peter Lynch models, Fama French approaches and other established methods. On the quality side, metrics such as return on equity, EBIT margin, Piotroski F Score, debt ratio, debt payback period and the Altman Z Score are included. The tools of the Fairvalue Calculator are essentially a practical translation of the same factors that research has identified as return drivers.

One central element of my analyses is a study of 100 stocks that have been tracked in the tool since 2015 and that traded clearly below their calculated fair value at the time of purchase. If you look only at the evolution over time, you see a very clear pattern. The median time for the price to move back towards the calculated fair value is about five years. This confirms the core idea of mean reversion. Markets swing in both directions, but intrinsic value acts like a centre of gravity that prices tend to return to over time. For investors this means that a fair value approach almost never works in turbo mode overnight. Its strength unfolds over a multi year horizon.

The typical paths of this reversion can be described in four broad patterns. In the classic case, the price rises and slowly catches up with intrinsic value. In the second case, the price seems to move sideways in the chart while fundamentals deteriorate and fair value drifts down. In the third case, fundamentals improve and fair value rises while the price reacts only moderately. In the fourth, mixed case, both price and fair value move towards each other. In all four versions the message is the same. Price and intrinsic value are not static, but they tend to move together over time.

Things become even more concrete in a simple but telling sample of 32 stocks selected in 2005 from the United States and Germany. All 32 were clearly undervalued according to the Fairvalue Calculator at that time. From 2005 to 2010 these stocks achieved an average annual return of about 12 percent, while the market, measured roughly as the average of DAX and Dow Jones during a period dominated by the financial crisis, managed about 3 percent per year. That implies an outperformance of about 9 percentage points per year. Over the period from 2005 to 2019 the outperformance was still around 7 percentage points per year, and from 2005 to 2022 around 8 percentage points. If you assume a long term average market return of about 9 percent, you arrive at a realistic, though not guaranteed, expectation of roughly 17 percent per year for fair value stocks.

Of these 32 stocks, 26 reached their calculated fair value within the observation period. The remaining 6 moved significantly closer without fully closing the gap. On average it took about five and a half years until fair value was reached or largely priced in. This fits remarkably well with the 100 stock study and with the results of Frankel and Lee, who see a holding period of three to five years as sensible for fair value strategies (Frankel and Lee 1998).

In an update to August 2025, the same 32 stocks were evaluated again. The result is both impressive and sobering. On the one hand, the 32 fair value stocks gained an average of about 846 percent, while the overall market rose by about 162 percent over the same period. The median return for the sample was about plus 180 percent, the lower quartile about minus 29 percent and the upper quartile about plus 460 percent. Some names, such as Goldman Sachs or CTS Eventim, became real high flyers. Others, such as Leoni or Salzgitter, remained problem cases. The message is clear. The strategy can push the average return significantly higher, but it always remains a game of probabilities. Without diversification and patience it does not work.

To rule out the possibility that these 32 stocks were just a lucky accident, I used the Fairvalue Calculator database to run automated tests across thousands of stocks. Today more than 50,000 securities worldwide are tracked with their fair values and their relative performance versus the market. In these automated backtests the same pattern appears again and again. Undervalued stocks, defined by a clear discount to fair value and filtered by quality criteria, deliver a noticeable long term excess return compared with the broad market. The exact height of this excess return depends on region, sector, selection filters and rebalancing rules, but the direction of the effect is robust.

The role of the investment horizon is particularly interesting. If you group the data by holding period, a clear picture emerges. In the first five years after purchase, the advantage of the fair value strategy is at its strongest. In the backtests, the average additional return in this phase is around 9 percentage points per year compared with the market. After that, the advantage gradually flattens. From roughly ten years onwards the extra return becomes smaller, but it remains positive. The intuitive explanation is straightforward. A large part of the catch up process happens during the first years. Once a stock has reached or exceeded its intrinsic value, it behaves more like the market again. Investors who sell too early miss the phase where the positive surprises tend to be largest.

The academic literature on value investing and the results from the Fairvalue Calculator line up surprisingly well. Studies such as those by Piotroski, Frankel and Lee or Fama and French provide the theoretical and empirical basis. Valuation discounts, fundamental data and market overreactions are linked to later returns. The in house backtests, live tests, samples and survival curves show that these effects are not limited to academic datasets. They also appear in a practical, rules based system that has lived through all crises and sideways phases since 2005.

What does this mean for realistic expectations of a fair value strategy in everyday life. Historically, stock markets have delivered average annual returns somewhere between about 7 and 10 percent, depending on index and period. The combination of undervaluation and quality adds, according to the mentioned studies and my own analyses, around 7 to 9 percentage points per year on top, at least over the first five years after purchase. This places a disciplined, well diversified fair value strategy in a realistic corridor of about 15 to 18 percent per year as a long term expectation. This is not a promise, not a guarantee and certainly not a one way street upwards. It is a statistical expectation based on data from both academic research and the practical history of the Fairvalue Calculator.

Equally important as the return side are the risks and limitations. Value signals are cyclical and can underperform for years. In certain phases, for example during extreme growth bubbles, you will look worse than the market for a time. Transaction costs, taxes and errors in fundamental data can dampen results further. For this reason the Fairvalue Calculator deliberately uses simple, economically sensible rules, avoids look ahead bias, tests strategies across regions and time periods and emphasises diversification across sectors, size segments and countries. That is also why the tool includes a portfolio manager that tracks not only fair values, but also weightings, sector allocation and risk indicators.

Taken together, the picture is clear. The fair value method is not an esoteric gut feeling. It stands on three legs. Decades of scientific evidence for value and quality, in house long term studies and backtests based on tens of thousands of stocks, and a rules based, transparent implementation tool. You can never and should never exclude the possibility that the future will differ from the past. But if you base your decisions on data, fundamentals and statistical effects, you are not betting on hope. You are betting on probabilities. That is exactly what this method is about.