Skip to content

CFA Level II Exam · Exchange-Traded Funds: Mechanics and Applications

ETF Trading Costs, Tracking Difference and Tracking Error

Updated 7 October 2026 · Fact-checked

ETF cost has two parts: costs of trading (bid-ask spread, commissions, market impact) and costs of holding (expense ratio, tracking difference). Tracking difference is the ETF return minus the index return over a period. Tracking error is the standard deviation of those differences. Add up the costs over your holding period.

Understand ETF Trading Costs and Tracking Error

An ETF tries to deliver the return of an index. It never does so perfectly. You face two kinds of cost: what you pay to get in and out, and what you lose while you hold.

Trading costs come from the bid-ask spread, brokerage commissions and market impact on large orders. You buy at the ask and sell at the bid, so a round trip costs about one full spread. The spread is wider when the underlying securities are illiquid, when markets are stressed, or when the ETF has few market makers. Spreads are paid every time you trade, so they matter most for short holding periods.

Holding costs start with the expense ratio, the annual management fee charged inside the fund. It is not billed to you; it reduces the fund's NAV return. Other drags include trading costs inside the fund when it rebalances, and withholding taxes on dividends. Offsets include securities lending income and sampling choices that cut costs. Cash held for dividends or redemptions can also drag in rising markets.

Tracking difference is the ETF return minus the benchmark return over a stated period. It is a single signed number, and it is often negative, roughly equal to the expense ratio minus offsets. Tracking error measures the variability of the period-by-period differences. It is their standard deviation, so it shows how consistent the tracking is, not how large the average gap is.

A fund can have a large tracking difference and a small tracking error if it lags the index by the same amount each period. A fund can also have a near-zero average difference but a high tracking error if the gaps swing both ways. Sources of tracking error include fees, sampling or optimised replication, imperfect rebalancing, index changes, cash drag, securities lending, withholding taxes, and differences in valuation timing or currency. Also note that a market price can differ from NAV, which adds premium or discount risk for the investor.

Key formulas to remember

Tracking difference
Tracking difference = ETF return − Benchmark return
Measured over a stated period, for example a year. Signed value. Usually negative for a fund after fees.
Tracking error
Tracking error = standard deviation of (ETF return − Benchmark return) over the periods
Use the sample standard deviation, dividing by n − 1, of the periodic differences. Annualise by multiplying by √(periods per year) if periodic data is used.
Round-trip spread cost
Round-trip cost ≈ Ask − Bid, or as a % = (Ask − Bid) ÷ Midpoint
Buying at the ask and selling at the bid costs one full spread. Half-spread is the one-way cost against the midpoint.
Total cost of ownership (approx.)
Total cost ≈ Spread cost (annualised over holding period) + Expense ratio + Other drag
Divide the round-trip cost by the number of years held to compare with the annual fee.
Expected tracking difference
Tracking difference ≈ − Expense ratio + Securities lending income + Other offsets − Other drags
A rough guide, not a rule. Check each item's sign.

How to solve ETF Trading Costs and Tracking Error questions

Use this method for any item set question on ETF costs or tracking.

  1. 1Read the question and decide whether it asks about trading cost, holding cost, tracking difference or tracking error.
  2. 2Find the numbers in the vignette or exhibit: bid, ask, midpoint, expense ratio, holding period, and the ETF and index returns.
  3. 3For tracking difference, subtract the benchmark return from the ETF return period by period and keep the sign.
  4. 4For tracking error, compute the differences first, then their sample standard deviation. Do not use the returns themselves.
  5. 5For trading cost, convert the spread into a percentage of the midpoint or price, and spread it over the holding period if you compare with an annual fee.
  6. 6Add the cost components that apply to the investor, not to the fund, and check that all are in the same time unit.
  7. 7Match the result to the answer options and check the sign and the direction of the sources, such as lending income reducing the gap.

Quickest way: Sign and size check for ETF cost questions

When to use it: Use when the options differ in sign, scale or which cost is included.

  1. Decide whether the answer is a difference (signed), a standard deviation (never negative) or a cost (positive).
  2. Estimate the spread as a percentage: (ask − bid) ÷ midpoint.
  3. For total cost over several years, divide the round-trip spread by the years held and add the expense ratio.
  4. Eliminate options that use the wrong holding period or confuse tracking error with tracking difference.
  5. Only do the full standard deviation if the options are close.

Common mistakes in ETF Trading Costs and Tracking Error

  • Treating tracking error and tracking difference as the same thing.

    Both measure how far an ETF is from its index, and the names sound alike.

    Fix: Tracking difference is a signed return gap over a period. Tracking error is the standard deviation of the periodic gaps.

  • Computing tracking error as the standard deviation of ETF returns.

    Students apply the usual risk formula to the wrong series.

    Fix: Subtract the benchmark return from the ETF return first. Then take the standard deviation of that difference series.

  • Counting the spread once for a round trip when it is quoted against the midpoint.

    The half-spread and full spread get confused.

    Fix: A buy at the ask and sale at the bid costs the full spread. Against the midpoint, each leg costs half.

  • Comparing the one-off spread directly with the annual expense ratio.

    The units differ but the numbers look comparable.

    Fix: Divide the round-trip spread by the years held, then add the annual expense ratio.

  • Assuming tracking difference is always equal to the expense ratio.

    It is a handy rule of thumb.

    Fix: It is only approximate. Securities lending income, sampling, cash drag and withholding taxes also change it, so the gap can be smaller or larger than the fee.

  • Thinking a low expense ratio means low tracking error.

    Fees are the most visible cost.

    Fix: Fees mostly create a steady negative difference. Tracking error rises with sampling, cash drag and other uneven drags.

Worked examples

Example 1

Vignette: An analyst considers an equity ETF for a 4-year hold. The ETF quotes a bid of 49.90 and an ask of 50.10. Its expense ratio is 0.20% a year. Q1: What is the round-trip spread cost as a percentage of the midpoint? Q2: What is the approximate annual total cost, counting the spread and expense ratio?

Show the solution
  1. Midpoint = (49.90 + 50.10) ÷ 2 = 50.00.
  2. Spread = 50.10 − 49.90 = 0.20.
  3. Round-trip cost = 0.20 ÷ 50.00 = 0.40%.
  4. Spread spread over 4 years = 0.40% ÷ 4 = 0.10% a year.
  5. Annual total cost ≈ 0.10% + 0.20% = 0.30%.

Answer: Q1: 0.40% round-trip. Q2: about 0.30% a year.

Example 2

Vignette: Over four years an ETF returned 8%, 12%, −3% and 7%. Its benchmark returned 8.5%, 12.3%, −2.6% and 7.4%. Q1: What is the tracking difference in each year? Q2: What is the tracking error as a sample standard deviation? Q3: What does the pattern suggest?

Show the solution
  1. Differences: 8 − 8.5 = −0.5; 12 − 12.3 = −0.3; −3 − (−2.6) = −0.4; 7 − 7.4 = −0.4.
  2. Mean difference = (−0.5 − 0.3 − 0.4 − 0.4) ÷ 4 = −0.40.
  3. Deviations from the mean: −0.10, +0.10, 0.00, 0.00.
  4. Squared deviations: 0.01 + 0.01 + 0 + 0 = 0.02.
  5. Sample variance = 0.02 ÷ 3 = 0.006667.
  6. Tracking error = √0.006667 = 0.0816%, about 0.08%.
  7. The gap is steady near −0.4% a year, consistent with a fee-like drag and low variability.

Answer: Q1: −0.5%, −0.3%, −0.4%, −0.4%. Q2: about 0.08% (tracking difference averages −0.40%). Q3: a persistent small negative difference with low tracking error, typical of a fee drag.

Exam tips

  • Always identify which measure the question asks for: a signed difference, a standard deviation, or a cost.
  • Take the bid and ask from the exhibit and use the midpoint as the base unless told otherwise.
  • Check the holding period before comparing spread costs with the expense ratio.
  • When a question asks for a source of tracking error, link it to the fund: sampling, cash drag, fees, lending, taxes or rebalancing.
  • There is no penalty for wrong answers, so always pick an option, but eliminate by sign first.

ETF Trading Costs and Tracking Error in other exams

The same ground in other exams, if you are preparing for more than one or want another angle on it.

ETF Trading Costs and Tracking Error: frequently asked questions

What is the difference between tracking error and tracking difference?

Tracking difference is the ETF return minus the index return over a period, and it carries a sign. Tracking error is the standard deviation of the periodic differences. One shows the size of the gap, the other its variability.

How do I calculate ETF tracking error?

Compute the ETF return minus the benchmark return for each period. Then take the standard deviation of those differences. Annualise by multiplying by the square root of the number of periods per year if needed.

Does the bid-ask spread count as part of the expense ratio?

No. The expense ratio is an ongoing fund fee that reduces NAV. The spread is a transaction cost you pay when you trade the ETF on the market. Both belong in total cost of ownership.

What causes ETF tracking error?

Common sources are fees, sampling or optimised replication, cash drag, rebalancing and index changes, withholding taxes, and differences in timing or currency. Securities lending income can reduce the drag.