Portfolio a alokace
Portfolio Backtesting: How to Do It and Why Not to Trust It Blindly
Key takeaways
- A backtest is a useful tool for understanding portfolio characteristics, not a guarantee of future returns.
- Survivorship bias, data mining, and overfitting to history are the three main backtesting traps.
- The more parameters you optimise on historical data, the more you're testing the past, not the robustness of the strategy.
- A good backtest includes fees, taxes, and psychologically realistic rebalancing rules.
A backtest is a historical simulation of portfolio performance — it shows how a strategy would have performed in the past, given certain rules. It's a valuable but easily misused tool.
What a Backtest Is For
Used correctly, a backtest helps you:
- understand the strategy's volatility — how deep the drawdowns were;
- compare two allocations on a level playing field;
- verify that rebalancing rules worked historically as intended;
- estimate the maximum drawdown and recovery time.
What a backtest is not for: predicting future returns, or choosing a strategy simply because it had the best historical performance.
Three Big Traps
Survivorship bias — historical databases primarily contain companies and funds that survived. Failed businesses are missing, so average returns are inflated.
Data mining and overfitting — if you test 100 different strategies on the same historical data, statistically you'll find several that work — purely by chance. A strategy fitted to a specific history doesn't work in the future.
Look-ahead bias — in the backtest model you unintentionally use information that wasn't yet available at the time (e.g. today's dividend data applied retroactively).
How to Backtest Properly
Include all real costs — fees, spreads, dividend withholding taxes. Test over the longest possible history covering different market cycles (bull market, bear market, sideways market, inflationary and deflationary environments). Use out-of-sample validation — test the strategy on a different time period from the one on which you designed it.
Practical Use
For the average investor, backtests are freely available on tools like Portfolio Visualizer or Backtest.curvo.eu. They help verify the basic characteristics of a portfolio — but not to pick the "best" strategy. We discuss factors that backtests don't easily capture in the article on factor investing and on why not to forecast markets.
FAQ
What is a portfolio backtest?
A historical simulation showing how a portfolio would have performed in the past under given rules — allocation, rebalancing, contributions. It's useful for understanding the strategy's volatility and drawdowns, but does not guarantee future returns.
What is survivorship bias?
A distortion in historical data: companies and funds that went bankrupt or ceased to exist are missing from databases. The result is that historical average returns are inflated by the survivors — better than investors actually experienced at the time.
How do I run a reliable backtest?
Include real fees and taxes, test over the longest possible history covering different market phases, and avoid optimising too many parameters. If possible, validate the strategy on a different time period from the one on which you designed it.