Part 7 - Checking Performance

Overview

In the previous post, the complete portfolio was detailed. It looks good on paper, but we need to investigate an important question: Does the portfolio actually do what it is supposed to do?

You might recall our primary objective from Part 1: achieving a 6% nominal return alongside a reasonable level of risk. In that same post, we estimated the expected nominal return for our asset allocation at 6.43%. So, will we actually get those returns?

One way to test this hypothesis is backtesting—a well-known method of evaluating an investment strategy using historical data.

I used two freely available tools for this: Curvo and a newcomer called EuroFolio. Backtesting has distinct limitations, but it provides valuable insights. I also entered the portfolio into justETF to examine structural parameters like replication types and fund sizes.

Let’s dive into the backtest results.

Backtesting Returns

Using EuroFolio and Curvo is straightforward after creating a free account. You enter the components of your portfolio with the target weights, and the tools run the historical analysis.

The table below summarizes key performance and risk metrics from both backtests:

The first thing to note is the testing timeline:

1. EuroFolio backtests from March 2024 onward because Xtrackers MSCI World ex USA UCITS ETF 1C was launched in early 2024
2. Curvo evaluated a broader 6.3-year historical dataset (Nov 2016 – Mar 2023) but did not offer more recent data.

The different annualised returns highlight the primary limitation of backtesting. EuroFolio measured an exceptional bull market (+17.10%), whereas Curvo tested a turbulent period containing the 2020 pandemic crash and 2022’s rate shock (+5.73%).

However, combining both perspectives confirms that our 6.0%–6.4% expected return baseline remains realistic.

Drawdown resilience is also notable. EuroFolio highlights a ~13.1% drop in mid-2025 during global tariff announcements, while Curvo captures the full 2020 pandemic drawdown (-17.90%).

What about Risk?

Let’s take a look at two often-used metrics.

Portfolio volatility across both tools rests between 9.55% and 10.74%. This is low compared to broad equity benchmarks like the MSCI World or S&P 500 (which historically range between 15% and 20% volatility) and emerging market equities (20% to 30%).

Our multi-asset structure successfully pulls overall volatility down to ~10%.

The Sharpe Ratio is a second metric we can look at. It measures risk-adjusted return (excess return earned per unit of volatility). The formula is:

• E[Rp - Rf]: Expected excess return over the risk-free rate.
• Rp: Total return of the portfolio.
• Rf: Risk-free rate (e.g., short-term government bonds or €STR).
• σp: Annualised standard deviation (volatility).

An example of excess return. If your portfolio gives you 8% as total return and short-term government bonds give you 2.5%, your excess return (the reward for taking risk) is 5.5%.

Why the discrepancy between EuroFolio (1.58) and Curvo (0.59)? Again, market regimes explain the gap. EuroFolio captured elevated market returns during a bull run, inflating the Sharpe ratio, whereas Curvo factored in negative-yielding risk-free rates and major market drawdowns.

The Sharpe ratio is probably around 1 for the complete period which is fine when compared to other retirement portfolios. The 1.58 is really a fluke. Fantastic but not sustainable.

Across a full market cycle, a portfolio Sharpe ratio between ~0.8 and 1.0 is a realistic expectation.

Limitations of Backtesting

When reviewing these figures, keep two core limitations in mind:

1. Past performance is not predictive

   Market conditions are never identical. A portfolio tailored to look optimal over historical data risks.

   Worse. I can design a portfolio specifically aimed at looking good when backtesting it. If I (July 2026) create a portfolio with 80% semiconductors and 20% gold it looks amazing when I backtest with 5 years of data. Does that mean it is a good portfolio? BTW, this is called “overfitting”.

2. Backtests ignore psychology

   Mathematical models do not account for investor behaviour during real drawdowns—such as panic selling, hesitation when re-balancing, or market timing temptation. All these things are not visible in the data.

justETF Structural Check

I also imported the complete portfolio into justETF for structural verification.

The diagnostic confirms:

• No synthetic fund replication across our main stock/bond core (100% physical holdings).
• All ETFs hold Assets Under Management (AUM) well over €500M.
• Currency exposure is well-diversified, keeping EUR exposure within a manageable ~30–40% range.

All charts are drill-down which is practical. One important thing is still missing and that is my regional allocation.

What is next?

In Part 8: Risk Management, we will dive into how I handle rebalancing, structure my cash buffer, and manage sequence-of-returns risk during withdrawals.