Tracking Error and Active Share

Tracking error measures the volatility of active returns versus a benchmark, while active share measures the absolute difference in portfolio weights. They capture distinct dimensions of active management: factor risk versus holdings deviation.

Definitions

For portfolio return \(rp\) and benchmark return \(rb\), active return is \(ra=rp-rb\). Ex-ante tracking error follows from active weights \(a=wp-w_b\) and covariance \(\Sigma\):

TE_ex_ante = sqrt(a' Σ a)
TE_ex_post = stdev(r_p - r_b)
Active Share = 0.5 Σ_i |w_p,i - w_b,i|

import numpy as np

def active_metrics(portfolio, benchmark, covariance):
    active = portfolio - benchmark
    te = np.sqrt(active @ covariance @ active)
    active_share = 0.5 * np.abs(active).sum()
    return te, active_share

If the universe differs, map both portfolios to a common universe including zero weights. Fund cash, derivatives, short positions, and look-through holdings require explicit treatment.

Metric	High reading implies	Does not imply
tracking error	high active factor volatility	different securities
active share	different holdings/weights	high factor risk
information ratio	active return per TE	skill persistence

Why the metrics disagree

A sector-neutral manager can hold very different stocks from an index while maintaining low beta, sector, and style exposures: high active share, low TE. Conversely, a modest futures overlay can create high market beta and TE with almost unchanged cash holdings: low active share, high TE. Therefore a mandate that sets only one metric can be gamed or unintentionally restrict the strategy.

Risk budgeting

Decompose active variance by factor:

Σ = B F B' + D
active variance = a' B F B' a + a' D a

The first term is systematic active risk; the second is idiosyncratic risk. A manager claiming stock-selection alpha should explain why most active risk is not an unintended style, country, duration, or currency bet. The appropriate covariance model needs regularization; covariance shrinkage describes why sample covariance can create false precision.

Benchmark and implementation traps

Use the benchmark actually available at each date, including reconstitution timing. For international portfolios, align return timing and currency hedges. Ex-post TE is biased downward by stale prices and can miss jump risk. Ex-ante TE can be wrong in a regime shift. Report both, plus stress TE from historical shocks.

Annualize consistently: daily active-return standard deviation times sqrt(252) is a convention, not a guarantee of iid returns. Net returns should include trading costs required to maintain active weights.

Linking risk to conviction

Active share is descriptive, not a performance forecast. A portfolio can be highly active and wrong. Combine a risk budget with signal strength, breadth, capacity, and transaction cost estimates. In a Bayesian allocation framework such as Black-Litterman, views and their uncertainty determine how far active weights should move from equilibrium.

Key takeaways

• Tracking error measures active return volatility; active share measures holdings divergence.
• Use both metrics with factor-risk decomposition and a defined benchmark.
• Backtest with point-in-time benchmark data and realistic rebalancing costs.

Reporting template

Provide a holdings-based and return-based view on the same date. The holdings report should show active weights, factor contribution to tracking error, top active names, currency and derivative overlays, and benchmark mapping exceptions. The return report should split active performance into allocation, selection, currency, interaction, and trading cost where the attribution methodology supports it.

Set limits as ranges with escalation thresholds rather than treating an annualized TE target as a precise daily boundary. A sudden drop in realized TE can reflect stale prices or a market closure, while a spike may reflect a benchmark event rather than a manager decision. Review active share after corporate actions and index reconstitutions, which can change it without any discretionary trade.

For derivatives, calculate active exposure on a delta-adjusted or equivalent look-through basis and disclose the convention. A fund can appear index-like in cash weights while carrying substantial active beta through futures, options, or swaps. Without this mapping, active share becomes a cosmetic statistic rather than a useful governance metric.