What Is Tracking Error? How Far a Portfolio Drifts From Its Benchmark

Tracking error measures how much a portfolio’s returns vary from its benchmark’s returns, period by period. It is not a measure of underperformance and it is not a measure of error in the everyday sense. It is a measure of difference, and difference in either direction counts the same.

The practical reading is simple. A low tracking error means the portfolio behaves almost exactly like its benchmark. A high tracking error means it does its own thing, whether that turns out well or badly.

What it measures

The definition in words. For each period, day, week or month, take the portfolio’s return and subtract the benchmark’s return over the same period. That difference is the active return for that period. Do this for every period in the history. You now have a series of differences, some positive, some negative.

Tracking error is the standard deviation of that series. In plain terms it is a measure of how spread out those differences were around their own average. It is then usually annualised so it can be quoted as a yearly figure and compared across portfolios.

Three features of that construction matter.

It is a dispersion measure, not a level measure. A portfolio that beat its benchmark by exactly the same amount every single month would have an average active return that is positive and a tracking error close to zero, because the differences barely varied. Consistency of difference produces low tracking error even when the difference itself is large.

It treats upside and downside deviation identically. A month where the portfolio outpaced the benchmark contributes to tracking error just as much as a month where it lagged by the same amount. This is the same symmetry that standard deviation carries generally, and it is the reason some analysts prefer downside-only measures for certain questions.

It depends on the measurement frequency. Daily and monthly data on the same portfolio will not produce identical annualised figures, because the annualisation assumes the periods are independent and they usually are not quite. Two tracking error numbers are only comparable if they were built the same way, over the same window, against the same benchmark.

How to read it

Start with the mandate. Tracking error is meaningless without knowing what the portfolio was supposed to do. For a fund whose stated job is to replicate an index, low tracking error is the job, and any meaningful deviation reflects cash holdings, replication method, trading costs, or the timing of index changes. For a fund whose stated job is to build a differentiated portfolio, low tracking error is arguably the problem, because it suggests the investor is paying active fees for something close to the index. The same number is good news in one case and bad news in the other.

Do not read it as a quality signal. Tracking error tells you how far a portfolio moved away from its benchmark. It says nothing about which direction. High tracking error is neither skill nor failure. It is simply the amount of active risk being taken.

Pair it with active return. The natural companion is the information ratio, which divides active return by tracking error. That ratio answers the question tracking error alone cannot: was the deviation rewarded? A portfolio with a large deviation and little to show for it is taking active risk without compensation. A portfolio with modest deviation and steady active return may be doing more with less.

Read it alongside active share. Tracking error is computed from returns. Active share is computed from holdings, measuring how different the actual positions are from the index constituents. The two can disagree. A portfolio can hold very different names that happen to move similarly, producing high active share and low tracking error. Looking at both gives a fuller picture of how differentiated a portfolio really is.

Check the benchmark is the right one. If the benchmark does not match the portfolio’s universe, the tracking error is mostly measuring the mismatch rather than the manager’s decisions. A small-company portfolio measured against a large-company index will show a large figure that reveals nothing about the portfolio itself. This is the same problem that distorts alpha, and the fix is the same: choose the benchmark deliberately.

Tracking error is the size of the bet against the benchmark. It never tells you whether the bet paid.

What it does not tell you

It does not tell you direction. This is the single most common misreading. Tracking error is symmetric. A portfolio that consistently outperformed and a portfolio that consistently underperformed, by the same varying amounts, produce the same figure. Anyone who reads a high tracking error as bad news has read a dispersion number as a performance number.

It does not tell you whether the risk was rewarded. That requires pairing it with active return, which is exactly why the information ratio exists as a separate metric.

It does not tell you the source of the deviation. A portfolio can deviate from its benchmark because of stock selection, sector weights, a persistent tilt toward smaller companies or cheaper valuations, cash holdings, or currency. Tracking error aggregates all of it into one number and cannot decompose it. Understanding the drivers requires factor exposure analysis and position-level attribution.

It assumes the past pattern of deviation continues. Tracking error is estimated from history. A portfolio that has just changed its process, its universe, or its concentration will have a historical figure that no longer describes it. The number is backward looking and adapts slowly.

It is affected by how returns are distributed. Standard deviation is a well-behaved summary when differences cluster around an average in a roughly symmetric way. When active returns are dominated by a small number of very large deviations, a single dispersion figure understates how lumpy the experience actually was.

It says nothing about drawdown. A portfolio can post a modest tracking error while still suffering a deep absolute fall alongside its benchmark, because tracking error only measures the gap between the two, not the level of either. Absolute risk needs an absolute measure such as maximum drawdown.

It is sensitive to window and frequency. Change the period, change the data frequency, or change the annualisation convention, and the number moves. Two figures from two sources are not comparable unless both were computed the same way.

Kept in its lane, tracking error is one of the more honest metrics in the toolkit. It states plainly how much a portfolio is willing to look different from its benchmark, which is the first thing anyone evaluating an active mandate should want to know, and the last thing that should be mistaken for a verdict.

Related reading

• Portfolio metrics explained: the hub that maps how return, risk and trade statistics fit together.
• What is alpha in investing: the benchmark-relative return that tracking error puts in context.
• What is the information ratio: active return per unit of tracking error, the pairing that gives the number meaning.
• Benchmark selection for portfolios: why the reference index decides what the figure means.
• Active share explained: the holdings-based companion to a returns-based deviation measure.

This article is educational. Altys Labs is not a registered research analyst or investment adviser, and nothing here is investment advice or a recommendation to buy, sell, or hold any security.