Tracking error is the volatility of the difference between your portfolio’s returns and the benchmark’s. A portfolio that mirrors SPY has tracking error near zero; one full of concentrated bets marches to its own drum and shows a high one.
It measures activeness, not quality — there is no “good” or “bad” tracking error in isolation.
Annualized from monthly returns against SPY, in percent per year.
A broad 40-stock portfolio with mild sector tilts might run a tracking error of 4% a year. A 8-position conviction portfolio can easily run 12% — in any given year it may beat SPY by 15 points or trail by as much, purely from the width of its bets.
Reference bands (percent per year, vs SPY):
| Tracking error | Reading |
|---|---|
| < 2% | effectively an index fund |
| 2–6% | moderately active — diversified with tilts |
| 6–10% | active stock picking |
| 10–15% | concentrated portfolio — normal for 5–15 positions |
| > 15% | lives a life of its own (heavy concentration, sector bets, crypto-like assets) |
Always pair it with alpha: tracking error is the risk budget of active investing, alpha is what that budget bought.
Most telling when your returns diverge from the index and you want to know whether that divergence is structural (high TE by design) or accidental. Also the honesty check for “closet indexing” — active fees on near-zero TE.
Needs at least 3 overlapping months with the benchmark. Monthly data understates it relative to daily. It is symmetric: beating the index by 10 points raises TE exactly as much as trailing by 10 — which is why it must never be read as a quality score.