How far from the benchmark are you actually straying?

You now have a fair benchmark and an active return. The next question is how much deviation produced that active return — because beating an index by 2 points while barely leaving it is a different act from beating it by 2 points while betting the portfolio on a handful of names.

Two measures answer it: one from returns, one from holdings.

Tracking error: deviation measured in returns

Build the series of active returns — portfolio minus benchmark, period by period. Tracking error is the standard deviation of that series, annualised.

From monthly data:

Tracking error (annual) = standard deviation of monthly active returns × √12

Worked example. Twelve monthly active returns with a standard deviation of 1.2%:

1.2% × √12 = 1.2% × 3.464 = 4.2% per year

Rough calibration, for orientation only: a well-run index tracker typically lands somewhere under 0.2%; a fund that calls itself active but hugs its index often shows 1–2%; a genuinely concentrated manager can run 6–10%. High and low are descriptions, not grades — an index fund's near-zero tracking error is exactly what it promised to deliver.

Active share: deviation measured in holdings

Active share = ½ × Σ | portfolio weight − benchmark weight |

The sum runs over every position in either portfolio; the half is there because every overweight is funded by an underweight, and you don't want to count the same deviation twice. The result reads as the percentage of the portfolio that differs from the index. A pure index fund scores near 0%. A portfolio sharing no holdings at all with its benchmark scores 100%.

Putting them together

Suppose a fund charges 0.90% a year, shows a tracking error of 1.1%, and an active share of 28%. The description that follows is purely arithmetic: roughly seven-tenths of the portfolio is the index, and the fee is being charged on all of it. Whether that is acceptable is a decision for the person paying it — the measurement only makes the situation visible.

Tracking error also sets up the next unit. It is the denominator of the information ratio, which asks the obvious follow-up: was the deviation worth the deviation risk?

Try it now

1. A fund and its benchmark are below. Switch both to Monthly and Measure the same three-year span on each; the difference is the active return over the window, in one number.

2. Tracking error is about the variation of that difference, not its size, so do it month by month: read the two monthly returns, subtract, and build the active series. Its standard deviation × √12 is the annualised tracking error.
3. Compare the number with the fund's stated objective. Write one neutral sentence: does the return series behave like a tracker, a mild deviator, or a genuine departure? A large active return with a small tracking error means something quite different from the reverse.