Arithmetic Averages of Realized Returns

What an Arithmetic Average Measures

When an investment has a different realized return in each period, one simple summary is the arithmetic average return. It gives the mean return across the observed periods, assigning the same weight to each period. This makes it useful for describing the typical period in a particular series of observations.

The arithmetic average answers a specific question: if the observed periodic returns are treated as a list of equally weighted values, what is their ordinary numerical average? It does not, by itself, tell you the rate at which an investment’s value compounded over the full span. Those are different interpretations, even though both involve returns.

A realized return is the return observed over a particular period, not a forecast or probability-weighted expectation. If period returns have already been calculated, the arithmetic-average calculation needs only those returns and the number of observations. For example, a return of 6% should be entered as either 6 percentage points when all returns are percentages or \(0.06\) when all returns are decimals. The two approaches produce the same average after expressing the result in the corresponding units.

For a period in which an investment has beginning value \(P_{t-1}\), ending value \(P_t\), and a cash distribution \(C_t\) received during the period, a simple holding-period return can be written as $$ R_t=\frac{P_t-P_{t-1}+C_t}{P_{t-1}}. $$ This expression treats the distribution as part of the period’s return and assumes it is measured relative to the beginning value; it does not model reinvestment of that distribution during the period. When the periodic returns are supplied directly, do not recalculate them unless the question asks you to.

Why the Formula Is an Equal-Period Mean

The arithmetic mean is the value that balances the signed deviations of the observed returns. To see why, suppose a candidate value \(m\) is used as the average. The deviations from that value are \(R_1-m,\ldots,R_n-m\). Their total is $$ \sum_{t=1}^{n}(R_t-m)=\sum_{t=1}^{n}R_t-nm. $$ The balancing value is the one that makes this total equal to zero. Solving gives \(nm=\sum_{t=1}^{n}R_t\), hence \(m=(1/n)\sum_{t=1}^{n}R_t\). This shows why the formula is the ordinary mean and why every period contributes one observation, regardless of the amount invested in that period.

This definition describes an average of the recorded periodic returns. If periods have unequal lengths, the arithmetic mean still averages the listed observations, but it does not automatically convert them to a common time unit. If one observation is a monthly return and another is an annual return, first establish what statistic the question intends rather than averaging incompatible periods without comment.

Using Period Returns That Include Cash Distributions

The average is calculated from periodic total returns, not from price changes alone when cash distributions are part of the return. A distribution can make the total return different from the percentage change in the investment’s price. The next example first identifies each period’s total return and then averages those returns. This keeps the return calculation separate from the averaging step.

Equal Weight per Period, Not per Dollar

In the arithmetic mean, each return observation gets the same weight. That remains true whether the investment’s dollar value grew or fell, and whether a larger amount happened to be invested in one period. This is a feature of the statistic, not a claim that the same dollar amount was invested in every period.

If the question asks for an arithmetic average, do not replace it with a dollar-weighted calculation. Weighting by the amount invested answers a different question. Similarly, do not silently remove a negative return, a zero return, or a period with a small gain: if it is one of the observed periods specified in the problem, it belongs in the sum and in the count.

What the Arithmetic Average Does Not Tell You

An arithmetic average is a concise description of the observed returns, but it does not make the path of returns irrelevant. Two sequences can have the same arithmetic average while having different sequences of gains and losses. The average also does not state what return will occur next, establish how likely a particular return is, or provide a compounded growth rate.

For example, the three annual returns in the cash-distribution example average to 3.3333% per year. That figure is the equal-weight mean of the three annual observations. The actual ending value must instead be traced through the sequence of prices and distributions, using the stated treatment of cash flows. The arithmetic mean should therefore be reported and interpreted as a mean observed periodic return, not as a substitute for the investment’s actual multi-period performance.

This distinction matters when comparing summaries. A mean calculated from monthly observations is a mean per month; a mean from annual observations is a mean per year. State the period and the averaging convention alongside the number. If returns are expressed as decimals, convert the final result to a percentage when that makes the interpretation clearer. For a final reported percentage, round only after summing and dividing; here, four decimal places are suitable when a repeating decimal appears.