Weighted Averages Explained: Weights, Moving Averages, and Grades

An ordinary average treats every value equally. A weighted average does not: it gives each value a weight, a measure of how much it should count, and divides by the total weight. Grades are the familiar case — a final exam counts more than a quiz — but the same idea prices a portfolio, averages a cost over purchases of different sizes, and smooths a time series. The formula is short, the mistakes are predictable, and the difference between a weighted and an unweighted average is often the difference between the right decision and the wrong one. Grade-point calculations for a specific institution’s scale are a separate subject, and this guide does not replace a university’s own GPA rules.

The formula, and the one division that matters

The weighted mean multiplies each value by its weight, adds the products, and divides by the sum of the weights. If the weights are percentages that sum to 100, the division is by 100. If they are credit hours, the division is by the total credit hours. If they are arbitrary numbers — 3, 2, 1 — the division is by 6. The division is what makes the result an average rather than a total, and it is the step most often skipped.

Worked example: three grades, weighted by credit hours. A 90 in a 3-credit course, an 80 in a 3-credit course, and a 70 in a 4-credit course. The products are 270, 240, and 280, totalling 790. The total weight is 10 credits. The weighted mean is 79.0. The ordinary mean of the three grades is 80.0, which overstates the result because the lowest grade carries the most weight. That one-point difference is the whole point of weighting: the 4-credit course counts more, so it pulls the average further.

Weights can also be entered as percentages, and the same example shows the equivalence. The weights are 30%, 30%, and 40%. The products are 27, 24, and 28, totalling 79, and dividing by 100 gives the same 79.0. The two forms are the same calculation with the weights scaled, and the calculator accepts either form. What it cannot accept is weights of zero or negative weights, which have no meaning in an average — a value that should not count should be left out, not weighted at zero in a way that hides the omission.

Try the method on your own figures with the Weighted Average Calculator.

Grades, costs, and returns

Grades are the case most people meet first. A course grade is a weighted average of its components — homework, quizzes, midterm, final — with weights set by the syllabus. A semester average is a weighted average of course grades by credit hours. A degree classification is a weighted average of years, often with later years weighted more heavily. In each case, the weight is the syllabus’s or the institution’s statement of importance, and using any other weight — equal weights, or weights from memory — gives a different answer from the official one.

Average cost is the same idea in money. Buying 100 shares at $20 and 300 shares at $30 gives an average cost of (100 × 20 + 300 × 30) ÷ 400, which is $27.50 — not the $25 that the ordinary mean of the two prices suggests, because three times as many shares were bought at the higher price. The same calculation averages the cost of inventory bought at different prices, the cost of a loan drawn in tranches, and any other quantity where the price depends on how much was bought at it.

Investment returns weight by money, not by time. A portfolio that is 60% in an asset returning 10% and 40% in an asset returning 4% returns 0.6 × 10 + 0.4 × 4, which is 7.6% — if the weights are the portfolio shares at the start of the period and no money moved during it. When money moves in and out, the calculation becomes a money-weighted return, and the timing of the flows matters as much as their size. The simple weighted mean is the starting point, not the whole of performance measurement.

Moving averages: weighting over time

A moving average is a weighted average of the most recent values in a series, recomputed as each new value arrives. A simple moving average weights the values in its window equally: the 5-day average is the mean of the last five days. An exponential moving average weights recent values more heavily, with the weight falling exponentially as values recede, so it reacts faster to change while still smoothing noise.

The trade-off is lag against smoothness. A longer window smooths more and lags more: a 50-day average of a rising price sits below the current price for the whole rise, and only catches up when the rise ends. A shorter window follows the price closely but jumps with every noisy day. There is no window that both smooths and follows perfectly, and the choice depends on the purpose: trend identification favours longer windows, timing favours shorter ones, and neither predicts the next value.

The same idea appears outside finance. A 7-day average of daily cases smooths the weekend reporting pattern. A rolling average of temperatures smooths daily weather into climate signal. In every case, the average is a description of the recent past, weighted toward the present, and it should not be read as a forecast. The most recent value is the forecast’s starting point; the average is its context.

Mistakes: weights, averages of averages, and Simpson’s paradox

The first mistake is averaging averages. The mean of two group means is the overall mean only if the groups are the same size. Two classes averaging 80 and 70, with 10 and 30 students, average 72.5 overall, not 75 — because the larger class counts more. Whenever the groups differ in size, weight by size, or go back to the original values.

The second mistake is using weights that do not mean what they claim. Weighting grades by the number of assignments rather than by the syllabus weights, weighting returns by the number of holdings rather than by money, or weighting survey responses equally when some groups were oversampled — each produces an average that answers a different question from the one asked. The weight must be the quantity that makes a value count more, and if it is not, the average is wrong however carefully it is computed.

The third is Simpson’s paradox, where the overall average reverses the pattern in every subgroup. A treatment can look better overall and worse in every subgroup, if the subgroups are weighted differently between treatments. The paradox is a warning, not a curiosity: an average without its weights, and without the subgroup breakdown, can mislead even when the arithmetic is correct. Always ask what the weights are, and whether the subgroups tell a different story from the total.

Frequently asked questions