Tracking error and information ratio: deviation from the index

In depth7 min read · updated 17 September 2026 · 7 sources

The tracking error measures how much a fund's return fluctuates around that of its benchmark index. For an ETF it should be small; for an active fund it is the price of the chance to do better. Whether this deviation pays off is shown by the information ratio.

How to calculate tracking error

What matters is what is being measured: the tracking error measures the fluctuation of the difference, not the difference itself. A fund that lags the index by exactly 1% every year has a tracking error of zero and still a poor track record.

Ex post and ex ante

Measured from past returns, the tracking error is called ex post (realized). Asset managers, by contrast, steer the expected, ex ante tracking error: it follows from the deviations of the weights from the index and the covariance matrix of the investments.

This gives a practical rule of thumb: deviations between similar building blocks (two equity regions) create little tracking error, deviations between different investments (equities versus bonds) a lot.

Tracking error or tracking difference?

For ETFs, the tracking difference is often quoted: the average return difference from the index, usually slightly negative because of costs (see TER). The tracking error, by contrast, measures the fluctuation of this difference. An ETF can have a tracking difference of −0.2% and still a very small tracking error. It then lags by fairly exactly 0.2% every year.

The deviation cannot be avoided entirely. The index is a paper portfolio without costs, while the fund trades in real markets; Frino and Gallagher (2001) describe for S&P 500 index funds how these frictions make tracking error unavoidable. According to Elton et al. (2002), even the first US ETF lagged the index by its costs and because of how it handled dividends. Typical causes:

• Ongoing charges (TER): they are the largest and most consistent item.
• Withholding tax on dividends: indices are often calculated with one tax rate, while the fund may pay more or less depending on its domicile.
• Optimized sampling: if a fund does not buy all the securities, it deviates randomly.
• Cash holdings and trading times: inflows are invested with a delay, and index changes are not implemented exactly at the closing price.
• Securities lending and swaps: income from lending improves the tracking difference, swap costs worsen it.

Checking the tracking difference: what it costs in dollars

Small differences add up. If an index returns 7% a year and an ETF lags it by 0.2 percentage points every year, $10,000 grows to $19,306.90 in ten years instead of $19,671.51, i.e. $364.61 less. With a shortfall of 0.5 percentage points, it is already $900.14. An active fund with higher costs first has to make up this shortfall before it delivers any added value at all.

1. Choose the right index: the same one the fund tracks, in the same currency and with the same treatment of dividends (usually “net return”).
2. Compare whole calendar years and take the average of the annual differences. That is the tracking difference.
3. Check how much the difference fluctuates from year to year. That is the tracking error. Large outliers point to special effects such as taxes or index changes.
4. Compare with similar ETFs on the same index: the TER alone says little; the actual tracking difference is what counts.

Information ratio

It is the Sharpe ratio of the deviation: how much excess return per unit of active risk? Over many years, 0.5 is considered good and values above 1 exceptional. As with all metrics from short periods, there is a lot of chance involved.

Where a good information ratio can come from is described by the fundamental law of active management by Grinold (1989): it is roughly a manager's skill (the information coefficient, the correlation between forecast and outcome) times the square root of the number of independent decisions per year. Grinold and Kahn (2000) build an entire textbook on it.

What the research shows

If you minimize your tracking error against the index while seeking excess return, you do not automatically end up with a good portfolio. Roll (1992) showed that such portfolios have a beta above 1 and are not efficient in terms of return and total risk:

His remedy: limit the portfolio's beta. The second major finding concerns active funds as a whole. With a simple calculation, Sharpe (1991) shows that the actively managed dollars must, before costs, earn exactly the market return on average (together they hold the market, after all) and fall below it after costs. French (2008) estimates the cost of active investing in the US from 1980 to 2006 at 0.67% of market value a year. A positive information ratio after costs is therefore the exception, not the rule; more in the article active or passive.

For advanced readers: how long until there is proof?

The information ratio is closely linked to statistics: the t-statistic of the average active return is roughly the information ratio times the square root of the number of years. To show with 95% confidence that a manager with a true information ratio of 0.5 is better than zero, you therefore need about (1.96 / 0.5)² = 15.4 years of data, longer than most fund managers stay in charge of a fund.

Two further subtleties: with autocorrelated active returns (for example due to style tilts that persist for years), simple annualization underestimates the tracking error. And factor tilts create apparent alpha: a value fund measured against a market index has a high tracking error and, in good value years, a high information ratio, without the manager having achieved anything special. A factor regression separates the two.

In your portfolio

If you want to replicate a particular factor tilt with as little deviation as possible, you minimize the tracking error against a target portfolio. That is how the “Replicate factors” tool on the Factors page works. A tracking error is neither good nor bad here, but a deliberate decision for a different weighting. For private investors, the more practical question is usually the reverse: how far may my portfolio deviate from the global market before I regret the deviation? A tracking error of 5% against the global equity index means that in one year out of three your portfolio does more than five percentage points better or worse than the market. You have to be able to live with that.

How Portfolio Lab calculates it

In the Backtest, the active return is the difference between the period returns of the portfolio and the benchmark over all common periods. The tracking error is their standard deviation times √12 (or √252), and the information ratio is the mean active return times 12 (or 252) divided by the tracking error. The fund comparison on the Assets page shows beta, alpha, R², tracking error, information ratio and capture ratios against a comparison fund of your choice over the chosen time window. More in the metrics methodology.

This article explains a method; it is not investment or tax advice. Worked examples are simplified and rounded. All terms in the glossary.