Deutsche Fassung
Models & simulation

Backtesting and rebalancing

A backtest runs a strategy with current target weights backwards through price history. Rebalancing restores those target weights at regular dates. The result is a simulation, not real trading history, and it already knows the outcome.

How often do you rebalance?4.0times per year
Buy and hold
Without rebalancing, weights drift with the market. Winners keep growing, and over the years the portfolio becomes more concentrated and usually more aggressive than intended.
Annual to semi-annual
The usual rhythm for private portfolios. It keeps weights near target without generating meaningful trading costs.
Quarterly
Closer to target, at the price of more transactions. Whether that pays depends on your broker order fees and spreads.
Frequent
The additional benefit over quarterly has historically been small while costs rise linearly. On small portfolios, order fees swallow the effect quickly.

How it is calculated

At each date: holdings → target weights, free price path in between

A portfolio starts at 60 % equities and 40 % bonds. After a strong equity year the ratio has shifted. Without rebalancing, the portfolio carries more risk than planned.

Start
60 / 40
Equities +25 %
65 / 35
After 3 years
72 / 28
After rebalancing
60 / 40

What the number does not tell you

  • The backtest knows the outcome. Your current target weights contain everything you have learned since. Applying a strategy backwards to positions that turned out well is not a test but a confirmation.
  • Only survivors are in the dataset. Companies that went bankrupt and ETFs that closed no longer appear in the price history. Every backtest on securities that exist today therefore comes out systematically too kind.
  • Tax is usually missing. Every rebalance can trigger capital gains tax. In a tax-free model, frequent rebalancing looks considerably better than it does in a real account.
  • Rebalancing is not a source of return. It mainly keeps the risk profile stable. Whether it adds return depends on whether the asset classes converge again in between. Through long trending phases it costs return.

Related metrics

Time-weighted returnThe quality of the investment, independent of contribution timing.Monte Carlo simulationThousands of possible paths instead of a single number.Annualised return (CAGR)Total return converted into an average year.Sharpe ratioReturn per unit of volatility, after deducting the risk-free rate.

How Evergrova calculates it

The backtesting page compares several strategies over the same window, with end value, contributed capital, annualised return, volatility, Sharpe, maximum drawdown, Calmar and the number of rebalances. The benchmark is deliberately left unrebalanced.

Basis: simulated constant-mix strategy with current target weights on daily returns, not real trading history.

View the demo portfolio

Common questions

How often should you rebalance?

Historically the difference between annual and quarterly matters little, while costs rise with frequency. Many investors rebalance once a year, or only once a weight has drifted by more than five percentage points.

Why is the benchmark not rebalanced?

Because an index updates its composition by its own rules. Additionally resetting it to fixed weights would make it a different strategy rather than a yardstick.

What is constant mix?

A strategy that holds the chosen target weights constant through regular rebalancing. It automatically sells what has risen and buys what has fallen, without requiring any market view.

Last reviewed: 2026-08-08

This text is general information. It is neither investment advice nor a recommendation. Metrics describe past periods and allow no conclusion about future performance.