Performance Attribution — The Detective Work of Portfolio Results

"If portfolio returns are a crime scene, performance attribution is the CSI team." — Your slightly dramatic investment TA

You're already fluent in translating policy into implementable portfolios — setting asset mixes, trading with an eye on implementation shortfall, and using derivatives overlays to shape exposures. Now comes the forensic part: performance attribution. It tells you why a portfolio beat or lagged its benchmark, who (or what) deserves credit or blame, and whether a manager’s process actually worked or was just lucky.

---

What Is performance attribution?

Performance attribution is the systematic decomposition of active return (portfolio return minus benchmark return) into components that explain the sources of that outperformance or underperformance. Think of it as converting a single suspicious number — say, +2.3% vs benchmark — into a courtroom-ready list of motives: allocation, selection, interaction, timing, trading costs, derivatives overlay effects, and cash/benchmark mismatch.

Why it matters now: You’ve already been selecting managers and monitoring implementation shortfall. Attribution completes the loop: did manager skill show up in stock picks (selection), or was it simply the result of favorable sector bets (allocation), or just a lucky macro wind?

---

How Does Attribution Work? (High-level mechanics)

There are two common approaches:

• Brinson-style (holdings-based) attribution — decomposes return into allocation, selection, and interaction effects by comparing portfolio and benchmark weights and returns at the security or sector level.
• Factor-based (returns-based) attribution — regresses portfolio returns on factor returns (e.g., market, size, value, momentum) to estimate exposures and returns attributable to systematic risk factors.

Both are useful. Holdings-based is granular (who did what), factor-based explains systematic sources of risk/return.

Core formulas (Brinson-Hood-Beebower flavor)

Allocation Effect_i = (w_p,i - w_b,i) * r_b,i
Selection Effect_i = w_b,i * (r_p,i - r_b,i)
Interaction Effect_i = (w_p,i - w_b,i) * (r_p,i - r_b,i)

Where w_p,i = portfolio weight in sector/security i, w_b,i = benchmark weight, r_p,i = return of portfolio holding i, r_b,i = benchmark return of i.

Sum across i to get total active return.

---

Examples of Attribution (short, memeworthy)

Imagine benchmark has 10% in Tech (return +8%) and 90% elsewhere (return +2%). Your portfolio put 20% in Tech and selected a few high-flyers inside Tech.

• Allocation effect: You overweighted Tech, which had high benchmark return — that gets you positive allocation credit.
• Selection effect: Within Tech, your picks outperformed benchmark tech names — that’s selection skill.
• Interaction: If your overweight also contained higher-than-benchmark returns, interaction amplifies results.

Table: simplified attribution summary

Effect	Contribution
Allocation	+0.80%
Selection	+0.60%
Interaction	+0.10%
Trading Costs / Shortfall	-0.25%
Derivatives Overlay	-0.05%
Total Active Return	+1.20%

---

Attribution for Derivatives & Implementation Shortfall

You already know derivatives overlays and implementation shortfall matter in portfolio implementation. Attribution must explicitly account for both:

• Derivatives: Convert derivative exposures to their equivalent underlying exposures (e.g., delta-equivalent equity exposure, duration equivalent for rates) or treat the derivative P&L as a separate line item. Otherwise, you mis-attribute returns to sectors that were actually driven by futures/forwards.

• Implementation shortfall / trading costs: These are real deductions from performance. Treat realized trading costs and realized slippage as negative contributions in attribution — otherwise you’ll overstate manager skill.

Pro tip: always reconcile executed trades and cash timing to holdings snapshots. Attribution is only as honest as your data.

---

Factor vs Holdings Attribution — Which to Use?

• Use holdings-based when you want granular, manager-level accountability (did the manager pick specific stocks?). Great for manager selection & monitoring.
• Use factor-based when the key question is exposure-driven: were returns due to macro or style tilts (value, momentum, carry)? Great when overlays or policy changes create factor exposures.

Often you do both: holdings-based to check the picks, factor-based to confirm structural exposures or to explain residuals.

---

Common Mistakes in Performance Attribution

1. Snapshot mismatch: using different times for portfolio and benchmark prices. Minute mismatch = messy attribution.
2. Ignoring cash and timing: large cash flows can distort attribution if not handled with multi-period linking.
3. Forgetting derivatives and overlays: attribute to the wrong bucket and blame the poor analyst.
4. Cherry-picking periods: short windows look like skill. Long-term consistency matters.
5. Data hygiene failures: wrong weights, stale prices, double-counting returns.

"If your attribution tells you the manager is a genius after you forgot to subtract commissions, the only genius here is the spreadsheet." — Notorious Data Lawyer

---

Ethics & Governance in Attribution

Attribution isn’t just math — it’s a responsibility.

• Transparency: Document methods (holdings-based vs factor-based), frequency, and assumptions. Clients must know how you measure.
• Avoid cherry-picking: Don’t report only the attribution from the best month. Show rolling results and explain volatility.
• Survivorship bias: Excluding closed managers or dead funds inflates apparent skill.
• Reconciliation: Reconcile attribution to custodial statements and confirm with independent verification.

Regulators and clients care about fair representation. Misstating attribution is not just sloppy — it’s ethically dangerous.

---

Closing — Key Takeaways (The Micro-Checklist)

• Performance attribution turns mysterious active returns into accountable causes: allocation, selection, interaction, factors, trading costs, derivatives.
• Use holdings-based for granular manager monitoring and factor-based for exposure analysis. Both are complementary.
• Always include derivatives, implementation shortfall, and cash timing in your attribution — otherwise your story is fiction.
• Keep methods transparent, data clean, and avoid ethical traps like cherry-picking or survivorship bias.

Final thought: attribution is the honest mirror for your investment process. If your manager’s CV brags about stock-picking skill, attribution is the blunt instrument that proves or disproves it. Do it well, and you turn performance noise into actionable insight.

---

Want a tiny homework nudge? Next time you review a manager, ask for a holdings-based attribution and a factor decomposition for the same period. If their story matches both, you may be on to something. If not, bring popcorn and questions.