This article opens a new series dedicated to the major ratios of quantitative investing, tested on real data. After exploring the factors that work in Zurich and Paris, it was time to expand the geographic scope and tackle new ratios. Welcome to The Ratio Wars.
What Warren Buffett does without naming it
Warren Buffett never mentions "FCF Yield" in his annual letters to Berkshire Hathaway shareholders. And yet, if you look at how he has selected companies for 60 years — businesses that generate mountains of cash relative to what they cost — you are describing exactly this ratio.
In this article, I do not merely explain the ratio — I test it on 22 years of real European data, with realistic simulation parameters: brokerage fees, slippage, annual rebalancing. The results speak for themselves.
What is FCF Yield — and why it beats the P/E ratio
The formula
There are two variants of FCF Yield. The simplified version divides free cash flow by market capitalisation:
FCF Yield (simple) = TTM FCF / Market Capitalisation
The more rigorous version — the one I use in this backtest — divides by Enterprise Value:
FCF Yield (EV) = TTM FCF / Enterprise Value
TTM FCF refers to free cash flow over the trailing twelve months. EV comprises market capitalisation plus net debt — which significantly changes the analysis for leveraged companies, as we will see.
Why FCF Yield beats the P/E ratio
The P/E ratio (Price-to-Earnings) is the most widely used ratio among retail investors. It also has three structural flaws that FCF Yield corrects:
- Net earnings are manipulable; free cash flow far less so. Accrual accounting allows management to smooth, anticipate or defer charges. Cash is either in the bank account or it is not. A high net profit combined with weak FCF is often a sign of poor earnings quality.
- The P/E ratio ignores debt. A company trading at a P/E of 10 while carrying 5× its EBITDA in debt is not cheap — it is expensive on a total enterprise value basis. FCF/EV neutralises this bias by comparing cash generated to total value (equity + debt).
- FCF incorporates maintenance capital expenditure. Net earnings can look flattering for a company that under-invests in its productive assets. FCF (= operating cash flow minus capex) reflects what the company generates after maintaining its productive capacity.
This ratio is a close cousin of EBITDA/EV, which I analysed in the previous article in this series. The main difference: FCF is a real cash flow after capex; EBITDA is a proxy for operating profitability before investment.
Backtest methodology
Universe and period
The backtest covers the period from 1 January 2004 to 6 March 2026 — 22 years including one major crisis (2008-2009) and several corrections (2011, 2018, 2020, 2022 and 2025). The universe covers European market securities, with three minimum liquidity filters: price above €0.50 (eliminates penny stocks), listed on a main market and not delisted, and median daily volume over 126 days above €10'000. If a company's FCF is negative or unavailable, the stock is automatically ranked in the bottom decile.
Simulation parameters
| Parameter | Value used | Rationale |
|---|---|---|
| Number of stocks | 40 | Replicable for retail investors, sufficient diversification |
| Rebalancing | Annual (52 weeks) | Optimal after real costs (see box below) |
| Commissions | 0.15% per trade | European low-cost broker (IB, Trade Republic, etc.) |
| Slippage | Variable by liquidity | More accurate than a uniform fixed rate |
| Transaction price | (High + Low + 2×Close) / 4 | Realistic execution price estimate without optimism |
| Weighting | Equal-weighted | Simple to replicate, no size bias |
| Ranking | Within-sector | Eliminates structural biases between sectors |
| Benchmark | MSCI Europe (EUR) | Consistent with the geographic universe |
⚙️ Why annual rebalancing, not monthly?
Screen backtests (without costs) show that the FCF Yield signal strengthens with rebalancing frequency — up to weekly. That is a positive sign: the signal is persistent and confirms itself in the short term. But once real costs are integrated (commissions + slippage), the dynamic reverses. Annual rebalancing emerges as the best compromise between signal efficiency and transaction costs — and by far the most practical frequency for a retail investor.
Why within-sector ranking?
An important methodological detail: FCF Yield ranking is calculated within each sector, not across the entire universe. The reason is straightforward: the FCF Yield of a regulated utility is not comparable to that of a software publisher. Ranking all companies together would systematically overweight asset-light sectors and underweight capital-intensive ones — regardless of their relative valuation within their own sector. Backtests confirm this: within-sector ranking consistently produces better results than universe-wide ranking.
Backtest results
FCF/EV: the numbers
| Metric | FCF/EV Model | Europe Benchmark |
|---|---|---|
| Annualised CAGR | 11.61% | 7.30% |
| Total return | 1'043% | 377% |
| Sharpe Ratio | 0.66 | 0.49 |
| Sortino Ratio | 0.88 | 0.65 |
| Max Drawdown | -58.78% | -58.42% |
| Annualised Std Dev | 16.98% | 13.96% |
| Beta | 0.96 | — |
| Annualised Alpha | +4.63% | — |
The headline result: +4.31 percentage points of annual CAGR above the benchmark over 22 years, with realistic transaction costs included. One euro invested in the FCF/EV model in 2004 is worth €11.43 today. The same euro in the benchmark is worth €4.77.
These results are consistent with the analyses presented in Les Déterminants de la Richesse, where I systematically test major valuation ratios on real data. FCF/EV ranks among the best single valuation ratios — those comparing a single fundamental criterion to a valuation element — ahead of both P/E and Price-to-Book.
Detailed annual performance
| Year | Model | Benchmark | Excess |
|---|---|---|---|
| 2004* | 9.37% | 13.05% | -3.68% |
| 2005 | 37.34% | 25.65% | +11.69% |
| 2006 | 37.84% | 20.61% | +17.23% |
| 2007 | 7.62% | 3.84% | +3.78% |
| 2008 | -49.92% | -42.71% | -7.21% |
| 2009 | +74.65% | +29.33% | +45.32% |
| 2010 | 15.81% | 10.00% | +5.82% |
| 2011 | -10.62% | -7.96% | -2.66% |
| 2012 | 18.43% | 17.58% | +0.84% |
| 2013 | 39.85% | 20.27% | +19.58% |
| 2014 | 1.77% | 6.51% | -4.75% |
| 2015 | 35.98% | 8.32% | +27.66% |
| 2016 | 14.47% | 2.52% | +11.95% |
| 2017 | 18.93% | 10.04% | +8.89% |
| 2018 | -23.58% | -9.62% | -13.97% |
| 2019 | 18.38% | 25.57% | -7.19% |
| 2020 | 4.88% | -3.22% | +8.10% |
| 2021 | 33.42% | 25.73% | +7.69% |
| 2022 | -15.74% | -9.15% | -6.58% |
| 2023 | 16.97% | 15.72% | +1.24% |
| 2024 | 11.72% | 8.97% | +2.75% |
| 2025 | 32.54% | 19.89% | +12.65% |
| 2026** | -1.71% | 2.40% | -4.12% |
| * From 01/01/2004 — ** As of 06/03/2026 | |||
FCF/EV vs FCF/MktCap: why debt changes everything
I tested both variants under strictly identical conditions. The comparison is revealing:
| Metric | FCF / EV | FCF / MktCap | Benchmark |
|---|---|---|---|
| CAGR | 11.61% | 8.34% | 7.30% |
| Sharpe | 0.66 | 0.43 | 0.49 |
| Max Drawdown | -58.78% | -66.27% | -58.42% |
| Beta | 0.96 | 1.15 | — |
| Annualised Alpha | +4.63% | +0.75% | — |
The verdict is unambiguous. But the most instructive figure is not the CAGR — it is the Sharpe ratio: 0.66 for FCF/EV versus 0.43 for FCF/MktCap, which falls below the benchmark (0.49). In other words, FCF/MktCap generates outperformance in absolute return terms, but with a worse risk-adjusted return than simply buying the index. The alpha of the simple variant is near zero (+0.75% annualised) and its beta of 1.15 reveals the cause: by ignoring debt, this ratio selects leveraged companies that appear cheap on price but are not cheap on total enterprise value. This leverage amplifies crises (MDD -66.27% vs -58.42% for the market) and does not translate into a genuine signal.
FCF/EV, with a beta of 0.96 and alpha of +4.63%, delivers real risk-adjusted outperformance. The difference between the two variants illustrates why capital structure cannot be ignored in a valuation ratio.
The limits of FCF Yield
Years of underperformance
Over 22 years, the model underperformed the benchmark in 7 years: 2004, 2008, 2011, 2014, 2018, 2019 and early 2026. The worst were 2008 and 2018. In 2008, the model fell harder than the benchmark (-49.92% vs -42.71%). The explanation is simple: 2008 was catastrophic for financial stocks — banks, insurers, credit companies — which represented a large portion of cheap value stocks on the FCF Yield screen. These sectors were hit far harder by the subprime crisis than fundamentals would have justified. The 2009 rebound (+74.65%, or +45 points above the benchmark) confirms that the underlying selection was sound.
Sectors with structurally low FCF
FCF Yield requires some nuance by sector. For REITs and property companies, AFFO is a more relevant indicator than FCF — and tests confirm that excluding them marginally improves results. For banks, the free cash flow concept is not directly applicable — but excluding them slightly degrades performance. Within-sector ranking proves to be a good safeguard: it manages these structural differences without requiring manual exclusion of entire sectors.
Look-ahead bias and point-in-time data
The simulation uses Portfolio123's "Point-in-Time Preliminary" method: the financial data used are those available at the time of rebalancing, not revised data applied retroactively. This is a critical point of rigour — one of the most common biases in financial backtests consists of using data that were not yet published at the time of the decision.
Recent 3-year performance
Over the past 3 years, the model shows a Sharpe of 1.33 versus 1.19 for the benchmark, with a MDD of -11.80% versus -16.34%. The FCF/EV signal has not eroded — it has strengthened. This was not guaranteed in an environment of elevated rates and a market dominated by technology growth stocks, a sector with limited European representation.
How to replicate this strategy
- A screener with FCF and EV for Europe. This is the main practical obstacle: free consumer tools (Yahoo Finance) do not provide FCF Yield on EV for European stocks. Specialist tools like Portfolio123 cover this need but are paid. That is a genuine barrier to replication. I provide a regularly updated list of the 40 current positions below.
- 40 positions, annual rebalancing. One working day per year to rebalance. The model's annual turnover is approximately 80% — in practice, replacing roughly 32 of the 40 positions at each annual rebalancing.
- The ability to hold through drawdowns. This is the real challenge. In 2008, watching a portfolio fall 50% while the market falls "only" 43% requires solid conviction. The 2009 rebound rewards those who did not sell.
Conclusion
FCF Yield — in its FCF/EV version — delivers compelling results on European markets. Over 22 years, with realistic transaction costs, it outperforms the benchmark by +4.31 percentage points per year, with market exposure (beta 0.96) nearly identical to the benchmark itself. This is not leverage — it is signal.
The comparison with FCF/MktCap may be the strongest takeaway of this article: the simple variant shows a Sharpe ratio below the plain index (0.43 vs 0.49) and near-zero alpha. A single figure that summarises why capital structure cannot be ignored in a valuation ratio.
First episode, first positive verdict. The rest of the series will test other ratios under strictly identical conditions — enabling direct comparison between them.
📋 The 40 stocks as of 09/03/2026
The list of 40 stocks selected by the FCF/EV model is provided for informational purposes only and does not constitute investment advice. It is updated regularly.
Frequently asked questions
What is the difference between FCF Yield and P/E ratio?
The P/E ratio divides the share price by earnings per share. FCF Yield divides free cash flow by enterprise value (EV). FCF Yield is considered more reliable because free cash flow is harder to manipulate accountically than net earnings, and because EV accounts for the company's debt — which the P/E ratio ignores entirely.
Can this strategy be implemented with any European broker?
The strategy itself is broker-agnostic — you simply need access to European markets at reasonable cost. Brokers such as Interactive Brokers, Trade Republic or Degiro allow replication of a 40-stock European portfolio with commissions around 0.10-0.20% per trade, consistent with the backtest parameters.
What to do if a company's FCF is negative?
In this backtest, companies with negative or unavailable FCF are automatically ranked in the bottom decile and therefore never selected in the top 40. In practice, a one-off negative FCF is not necessarily disqualifying (intentional investment phase), but structurally negative FCF over multiple years is a serious warning signal.
Why is annual rebalancing better than monthly?
Without costs, the FCF Yield signal strengthens with rebalancing frequency. But once realistic costs are integrated (0.15% commissions + variable slippage), annual rebalancing wins. The model's annual turnover of ~80% means more frequent rebalancing generates costs that erode outperformance. The key takeaway for retail investors: fewer transactions, not more.
Does this backtest suffer from survivorship bias?
No. Portfolio123 uses a point-in-time database that includes companies that were delisted, went bankrupt or were acquired over the period. This is a necessary condition for avoiding survivorship bias — one of the most common flaws in amateur financial backtests.
En savoir plus sur dividendes
Subscribe to get the latest posts sent to your email.