The one-sentence definition
CAGR is the constant annual growth rate that takes your starting balance to your ending balance over a given number of years, as if it grew by the same percentage every year.
How it's calculated
CAGR = (ending value / starting value) ^ (1 / years) - 1.
You divide the final balance by the initial balance, raise it to the power of one-over-the-number-of-years, and subtract one. The result is a yearly rate.
A worked example
Suppose $10,000 grows to $22,254 over four years - the kind of figure a multi-year backtest might report.
CAGR = (22,254 / 10,000) ^ (1 / 4) - 1 = (2.2254) ^ 0.25 - 1 ≈ 1.2218 - 1 = 0.2218, or about 22.2% per year.
So a total return of roughly +123% over four years is the same as compounding at about 22% a year. The total return sounds bigger; the CAGR is the honest yearly pace.
Why it matters for backtesting
Total return is misleading across different time spans: +50% in one year is far better than +50% over five years, but the headline looks identical. CAGR normalises for time, so you can line up a two-year backtest against a six-year one fairly.
It also keeps you honest about compounding. Big total returns over long periods often correspond to modest annual rates - and a modest, durable annual rate is usually what's actually repeatable.
Common mistakes
Where CAGR can mislead if you're not careful:
- Treating it as a smooth ride. CAGR is an average pace; the real path can swing wildly. Always read it alongside maximum drawdown.
- Using it over very short windows. A CAGR computed from a few months of data, then annualised, can produce absurd numbers.
- Ignoring the period chosen. A CAGR measured from a market bottom to a top flatters the strategy; the start and end dates matter enormously.