What a portfolio backtest actually is
A backtest is a simulation that looks backwards. You take an allocation, say 60 per cent global equities and 40 per cent bonds, then apply it to real prices from the past. The result does not predict anything about tomorrow, it only describes in numbers how that combination would have moved.
What it is really for and where it falls short
It measures the risk and return profile over time, that is the main use. Capital growth, deepest losses, volatility, links between assets. It makes it possible to compare different allocations over the same time span. To see how the portfolio holds up during rough patches, 2000–2002, 2008–2009, 2020.
It will not tell you future returns. It will not pick the best instruments going forward. It will not promise that old numbers come back. Markets shift, correlations shift, macro conditions shift.
How BacktestFolio's engine works
BacktestFolio runs historical simulations on the portfolio you define, weights, symbols, base currency, period, optional recurring contributions DCA, rebalancing and costs TER. It uses price series from the integrated datasets.
Series are aligned to the first and last common business day across the included assets. One ETF with data from 2005 and another from 2010, the simulation starts in 2010. Consistency is guaranteed but the available window may shrink compared to the range you asked for.
How to read the main results
Shows the evolution over time of the simulated portfolio value, contributions, withdrawals, rebalancing included. Every metric derives from here.
Annualised geometric mean of the return over the entire window. It does not describe uniformity year by year, a portfolio with a 7 per cent CAGR may have done +30 some years and –25 others.
The largest percentage drop from the running peak of the curve, peak-to-trough. A drawdown of –35 per cent means that from the peak the value fell by a third before recovering. It indicates the worst loss you would have suffered entering at the worst moment.
Annualised standard deviation of returns. High volatility, large swings in both directions. It is not the same as permanent loss, it measures the instability of the path.
Limitations to keep in mind
Every backtest carries methodological biases.
- Survivorship bias: the data represents instruments that are still active. Closed ETFs or funds do not appear and this can overstate average category returns.
- Look-ahead bias: some series are reconstructed after the fact, backcalculated, embedding information that did not exist at the time.
- Period sensitivity: a backtest from 2010 to 2020, an exceptional decade for US equities, gives very different results from one running 2000 to 2010.
- Autocorrelation in rolling windows: metrics over overlapping windows, for instance 3-year rolling returns, share observations. The number of truly independent samples shrinks.
TER and modelled transaction costs are simplifications. They can underestimate real friction especially in illiquid markets or with high-turnover strategies.
In BacktestFolio
On the Backtest page you load your portfolio assets, assign weights, choose the period, set rebalancing and cost assumptions. The engine automatically calculates equity curve, drawdown, volatility, Sharpe, Sortino, VaR and CVaR, correlations, inflation impact on real value.
You can compare two allocations side by side. Download a PDF report with every metric and chart.
See also
- Glossary: CAGR, Drawdown, Volatility, TWRR, TER
The content on this page is for educational purposes only and does not constitute financial advice, investment recommendation or promise of return. See the Financial Disclaimer.