Tracking error

Also seen as: active risk, tracking risk

It measures a strategy against its parent Combined, not against a benchmark: the common assumption that tracking error is always measured versus a benchmark does not apply here.

How is tracking error calculated?

Tracking error is the standard deviation of the difference between the strategy's return and its parent Combined's return in each period. A strategy that moves almost in step with the whole produces small, steady differences; one that often diverges produces wide, variable differences. The daily return differences are annualised by multiplying by √252 (252 trading days a year), the same convention Fincanva uses for volatility and the Sharpe ratio.

$$\begin{aligned} \key{1}{\text{Tracking error}} &= \operatorname{stddev}(\key{2}{r_{\text{strategy}} - r_{\text{parent}}}) \\ &\quad \times \key{3}{\sqrt{252}} \end{aligned}$$

• the annualised volatility of the return gap
• the per-period gap between the strategy's return and the return of the Combined it sits inside
• the annualisation of the daily differences

How does Fincanva handle it?

• Reported on the Strategy analytics page only, in its Each sub-strategy, against the Combined card: one bar per strategy in the Tracking error column. A strategy analysed on its own still appears there, but it has no parent Combined to be measured against, so its tracking error carries no meaning — read it only for a strategy inside a Combined.
• Measured between a member strategy and its parent Combined — the strategy's return relative to the whole, not relative to a benchmark.
• Reported as the standard deviation of the daily return differences, annualised by ×√252 — the same convention as volatility and the Sharpe ratio.
• Pairs with the information ratio, which divides a strategy's excess return over the Combined by this tracking error.

What counts as a high tracking error?

A low tracking error means the strategy moves closely in line with its parent Combined; a high tracking error means it diverges from the whole. Neither is inherently good or bad — the value tells you how much a given strategy pulls the Combined away from its own average path.

What does it look like in practice?

A Combined holds three strategies. Two of them tend to move closely with the Combined as a whole; the third often zigs when the Combined zags. Day to day, the third strategy's return differs from the Combined's by a wide, shifting margin, while the first two barely differ at all. The standard deviation of those daily differences, annualised by multiplying by √252 — much larger for the third strategy — is its tracking error. A high tracking error flags the strategy that pulls the Combined around the most, independent of whether that strategy made or lost money.

A tracking error is not a limit on how far a strategy can diverge in future.