Annualized Return & CAGR

Compound annual growth rate answers one question: what constant annual rate of return, compounded, would have turned the starting value into the ending value over the holding period? The formula is (Ending Value ÷ Beginning Value)^(1 ÷ Years) − 1. An investment that grows from $10,000 to $18,000 over five years has a CAGR of (1.8)^(0.2) − 1 = 12.5%. That does not mean it returned 12.5% in any particular year — it almost certainly did not. It means 12.5% compounded annually is the smooth-path equivalent of whatever actually happened.

The reason this matters is that the intuitive alternative — averaging the annual returns — is wrong, and wrong in a consistently optimistic direction. Consider an investment that gains 50% in year one and loses 50% in year two. The arithmetic average is 0%. The actual result is that $100 became $150, then $75: a 25% loss, or a CAGR of −13.4%. The arithmetic mean is not a small approximation error here. It describes an outcome that did not occur and could not occur, because returns compound multiplicatively rather than adding together.

The gap between the two has a name and an approximate size. Geometric return is roughly the arithmetic mean minus half the variance — CAGR ≈ arithmetic mean − σ²/2. This is volatility drag, and it grows with the square of volatility. Two portfolios with identical 10% average annual returns but volatilities of 12% and 30% deliver CAGRs of roughly 9.3% and 5.5% respectively. Over twenty years on $100,000, that is the difference between about $591,000 and about $292,000. The averages were identical; only the volatility differed. This is the single most important reason why reducing volatility is not merely a comfort measure — at equal average returns, the smoother path genuinely compounds to more money, and it is why managing risk is a return strategy rather than only a defensive one.

Annualizing converts a return measured over any period into its equivalent annual rate, which is what makes a six-month result comparable to a three-year one. The correct method is always geometric: raise one plus the total return to the power of the number of periods in a year, then subtract one. A 6% return over six months annualizes to (1.06)² − 1 = 12.36%, not 12%. The difference is the compounding of the second half-year on the first half-year's gains. Multiplying by two is a habit borrowed from simple-interest arithmetic and it understates every positive return.

Annualizing periods shorter than a year deserves particular caution, because it projects a short result across a full year as though the rate were durable. A fund that returns 8% in one quarter is often reported as '36.0% annualized.' That is arithmetically correct as a conversion and close to meaningless as a description, since it silently assumes three more quarters exactly like the first. This is why performance reporting standards discourage annualizing periods under twelve months, and why any annualized figure attached to a short track record should be read as a conversion rather than an expectation.

The other systematic error is endpoint sensitivity. Because CAGR uses only the first and last values, it is entirely determined by two dates and blind to everything in between. Shifting a measurement window by a few months across a market peak or trough can move a reported CAGR by several percentage points without anything about the investment changing. Any CAGR quoted from a period beginning at a market bottom is flattering by construction. The defense is to look at rolling returns — the distribution of CAGRs across many overlapping start dates — rather than a single window someone else selected.

CAGR is a smoothing device, and smoothing is the point: it strips out the path so that two investments can be compared on a single number. But the path is where investors actually live. Two funds can post an identical 10% CAGR over a decade while one drifted upward with a worst drawdown of 12% and the other doubled, collapsed 60%, and recovered. The number is the same; the experience is not, and the second fund's investors will mostly have sold near the bottom. This is why CAGR belongs next to maximum drawdown and volatility rather than on its own — it tells you where the investment ended, never what it cost to hold.

There is also a difference between the return an investment produced and the return an investor earned. Time-weighted return, which CAGR expresses, removes the effect of deposits and withdrawals and measures the strategy in isolation — appropriate for comparing funds, since a manager does not control when clients add money. Money-weighted return, or internal rate of return, accounts for the size and timing of cash flows and measures what actually happened to your capital. An investor who added heavily just before a drawdown can have a materially worse money-weighted return than the fund's published CAGR. Both figures are correct; they answer different questions, and the published one is rarely the one that describes your account.

Finally, CAGR carries no information about risk, leverage, or repeatability. A 40% CAGR earned across three years of a single sector's boom and a 12% CAGR earned across twenty years spanning several recessions are not comparable achievements, though the first number is larger. Sample length, regime coverage, and the volatility required to produce the result all sit outside the metric. Pair CAGR with the Sharpe ratio to see the return per unit of risk, and with maximum drawdown to see the worst case the strategy has actually delivered.