# Performance Attribution Report: [Portfolio or composite name], [period]

> Periodic report that explains why a portfolio's return differed from
> its benchmark: returns and excess return, Brinson-Fachler and
> Brinson-Hood-Beebower equity attribution (allocation, selection,
> interaction), multi-period linking, currency and fixed income
> attribution, risk-adjusted measures (Sharpe ratio, information ratio,
> tracking error), contributors and detractors, factor exposure,
> commentary and methodology. Based on published academic methods
> (Brinson and Fachler 1985; Brinson, Hood and Beebower 1986; Carino
> linking), which are cited by name and expressed in original
> formulas. The worked example is a fictional strategy of the fictional
> adviser Kestrel Bay Capital. Replace every `[bracketed]` value and
> every row marked *Example*. This is a document format, not
> investment, legal or accounting advice.

<!--
Guidance: How to use this template
- Use it monthly or quarterly for an internal investment review, a
  client report or an investment committee pack. For a client report,
  keep sections 1, 2, 3, 6 and 9 and move the rest to an appendix.
- Fill section 10 (methodology) first: the model, segmentation,
  data sources and linking method decide every number in sections 3
  to 5. Change the model only at a period start, and say so.
- Attribution explains the past; it does not prove skill. Pair it with
  the risk-adjusted measures in section 6 and a long enough period.
- When the report is shown to clients or prospects, it is marketing
  material in most jurisdictions: route it through Marketing Review
  Checklist (`compliance/marketing-review-checklist`).
- Related documents: Performance Composite Report
  (`finance/performance-composite-report`), Fund Factsheet
  (`finance/fund-factsheet`), Client Portfolio Review
  (`finance/client-portfolio-review`), Investment Committee Memo
  (`finance/investment-committee-memo`), Investment Policy Statement,
  Institutional (`finance/investment-policy-statement-institutional`).
-->

## Document control

<!-- Guidance: Identify the portfolio, benchmark, period and attribution model. A rerun after a data correction is a new version with the reason recorded. -->

| Field | Value |
|---|---|
| Document ID | [ATR-KBC-GQE-2026Q3] |
| Version | [1.0] |
| Status | [Draft / Reviewed / Approved / Superseded] |
| Owner | [@performance-analyst] |
| Approvers/reviewers | [@head-of-performance], [@portfolio-manager], [@chief-investment-officer] |
| Last updated | YYYY-MM-DD |
| Classification | [Internal / Client confidential] |
| Portfolio or composite | [Name and code] |
| Benchmark | [Index name, variant, currency] |
| Period | [YYYY-MM-DD] to [YYYY-MM-DD] |
| Base currency | [USD / EUR / other] |
| Attribution model | [Brinson-Fachler, arithmetic, monthly, Carino-linked] |
| Data sources | [Holdings, prices, benchmark constituents, classifications, risk model] |

### Approvals

| Role | Name | Decision | Date |
|---|---|---|---|
| Performance analyst (preparer) | [@name] | [Submitted] | YYYY-MM-DD |
| Head of performance (reviewer) | [@name] | [Approve / Return] | YYYY-MM-DD |
| Portfolio manager (commentary) | [@name] | [Approve] | YYYY-MM-DD |
| Compliance (if client-facing) | [@name] | [Approve / Not required] | YYYY-MM-DD |

### Revision history

| Version | Date | Author | Change summary |
|---|---|---|---|
| 0.1 | YYYY-MM-DD | [@name] | First draft for [period] |
| 1.0 | YYYY-MM-DD | [@name] | *Example:* Approved after PM commentary and reconciliation |
| 1.1 | YYYY-MM-DD | [@name] | *Example:* Rerun after a corrected benchmark sector classification for one constituent |

## Contents

<!-- Guidance: Keep this list aligned with the headings. -->

1. [Summary](#1-summary)
2. [Returns versus benchmark](#2-returns-versus-benchmark)
3. [Equity attribution: allocation, selection and interaction](#3-equity-attribution-allocation-selection-and-interaction)
4. [Currency attribution](#4-currency-attribution)
5. [Fixed income attribution](#5-fixed-income-attribution)
6. [Risk-adjusted measures](#6-risk-adjusted-measures)
7. [Contributors and detractors](#7-contributors-and-detractors)
8. [Factor exposure](#8-factor-exposure)
9. [Commentary and actions](#9-commentary-and-actions)
10. [Methodology and data](#10-methodology-and-data)
11. [Glossary and references](#11-glossary-and-references)

## 1. Summary

<!-- Guidance: Give the reader the answer in five lines: return, benchmark, excess, the main sources of the excess and what the team will do. Questions: was the result driven by decisions the team intends to make (allocation, stock picks) or by unintended exposures? Pitfall: a summary that lists numbers without saying which decision produced them. -->

| Item | Value |
|---|---|
| Portfolio return (gross, base currency) | [x.xx%] |
| Benchmark return | [x.xx%] |
| Excess return (arithmetic) | [x.xx percentage points] |
| Excess return (geometric) | [x.xx%] |
| Largest positive effect | [Effect, segment, basis points] |
| Largest negative effect | [Effect, segment, basis points] |
| Key message | [One sentence] |

*Example:* in the third quarter of 2026 the Global Quality Equity
representative portfolio returned 3.26% gross in USD against 2.65% for
the benchmark, an arithmetic excess of 60 basis points (geometric
0.59%). Stock selection added 31.5 basis points, led by industrials
(22.0 basis points). Sector allocation added 15.6 basis points, mainly
from the overweight in information technology and the underweight in
consumer staples. Interaction added 13.3 basis points; the three
effects sum to the 60.4 basis point excess (section 3.4). The only
negative segment was financials (minus 6.8 basis points), where the portfolio's
banks lagged. The team keeps the sector positioning and has started a
review of the two weakest bank holdings (see section 9).

## 2. Returns versus benchmark

<!-- Guidance: Show the returns that the attribution explains, per sub-period and for the whole period. Questions: are the portfolio returns transaction-based or holdings-based? Do they reconcile to the official performance figures? Pitfall: attributing a holdings-based return that differs from the official return by more than the effects being explained. -->

### 2.1 Period returns

| Month | Portfolio (%) | Benchmark (%) | Excess, arithmetic (pp) |
|---|---|---|---|
| 2026-07 *(Example)* | 2.422 | 1.985 | 0.437 |
| 2026-08 *(Example)* | -0.712 | -0.774 | 0.062 |
| 2026-09 *(Example)* | 1.538 | 1.440 | 0.098 |
| **Q3 2026 linked** | **3.257** | **2.653** | **0.604** |

Quarterly returns link the monthly returns geometrically:
1.02422 x 0.99288 x 1.01538 = 1.03257 for the portfolio and
1.01985 x 0.99226 x 1.01440 = 1.02653 for the benchmark.

$$
R = \prod_{t=1}^{T} (1 + R_t) - 1, \qquad
\text{arithmetic excess} = R_P - R_B, \qquad
\text{geometric excess} = \frac{1 + R_P}{1 + R_B} - 1
$$

*Example:* geometric excess = 1.03257 / 1.02653 minus 1 = 0.588%.

### 2.2 Reconciliation to official performance

| Check | Attribution system | Official performance | Residual (bp) | Tolerance (bp) | Status |
|---|---|---|---|---|---|
| REC-001 *(Example)*: Q3 portfolio return | 3.257% | 3.262% | -0.5 | 5 | Pass |
| REC-002 *(Example)*: Q3 benchmark return | 2.653% | 2.653% | 0.0 | 1 | Pass |
| [REC-NNN] | [Value] | [Value] | [bp] | [bp] | [Status] |

A residual appears when a holdings-based calculation misses intraday
trades, fees or cash flows. Report it rather than spreading it across
the effects, and investigate when it exceeds the tolerance.

## 3. Equity attribution: allocation, selection and interaction

<!-- Guidance: Decompose the excess return into the decisions the team makes: how much to hold in each segment and what to hold within it. Questions: is the segmentation the one used to make decisions (sector, region, country)? Is interaction shown or folded into selection? Pitfall: using a segmentation the team does not manage to, which produces effects nobody can act on. -->

### 3.1 Model

For each segment $i$, with portfolio weight $w^P_i$, benchmark weight
$w^B_i$, portfolio segment return $r^P_i$, benchmark segment return
$r^B_i$ and total benchmark return $R_B$:

$$
A_i^{BF} = \left( w^P_i - w^B_i \right) \left( r^B_i - R_B \right), \qquad
A_i^{BHB} = \left( w^P_i - w^B_i \right) r^B_i
$$

$$
S_i = w^B_i \left( r^P_i - r^B_i \right), \qquad
I_i = \left( w^P_i - w^B_i \right) \left( r^P_i - r^B_i \right)
$$

$$
\sum_i \left( A_i + S_i + I_i \right) = R_P - R_B
$$

The Brinson-Fachler allocation effect ($A^{BF}$) rewards overweighting
segments that beat the whole benchmark, which matches how most
managers think. The Brinson-Hood-Beebower version ($A^{BHB}$) rewards
overweighting any segment with a positive return. Because active
weights sum to zero, both versions give the same total allocation
effect; they differ by segment. This report uses Brinson-Fachler.

### 3.2 Single-period detail, July 2026

Weights are the average of daily weights for the month. In this
*Example* they are held constant across the quarter to keep the
arithmetic visible. Effects are in basis points (bp).

| Sector | $w^P$ (%) | $w^B$ (%) | $r^P$ (%) | $r^B$ (%) | Allocation BF (bp) | Allocation BHB (bp) | Selection (bp) | Interaction (bp) |
|---|---|---|---|---|---|---|---|---|
| Information technology *(Example)* | 30 | 25 | 4.2 | 3.8 | 9.1 | 19.0 | 10.0 | 2.0 |
| Health care *(Example)* | 22 | 15 | 1.8 | 1.2 | -5.5 | 8.4 | 9.0 | 4.2 |
| Industrials *(Example)* | 20 | 18 | 2.5 | 2.1 | 0.2 | 4.2 | 7.2 | 0.8 |
| Financials *(Example)* | 18 | 27 | 1.2 | 1.6 | 3.5 | -14.4 | -10.8 | 3.6 |
| Consumer staples *(Example)* | 10 | 15 | 0.5 | 0.3 | 8.4 | -1.5 | 3.0 | -1.0 |
| **Total** | **100** | **100** | **2.422** | **1.985** | **15.7** | **15.7** | **18.4** | **9.6** |

Check: 15.7 + 18.4 + 9.6 = 43.7 bp, which equals the July excess of
0.437 percentage points. Worked line for health care:
allocation BF = (0.22 minus 0.15) x (1.2% minus 1.985%) = 0.07 x
(-0.785%) = -5.5 bp; the segment beat zero but lagged the whole
benchmark, so the overweight cost return under Brinson-Fachler while
BHB shows a gain of 8.4 bp.

### 3.3 Monthly effects

| Month | Allocation (bp) | Selection (bp) | Interaction (bp) | Total (bp) | Excess (bp) |
|---|---|---|---|---|---|
| 2026-07 *(Example)* | 15.7 | 18.4 | 9.6 | 43.7 | 43.7 |
| 2026-08 *(Example)* | -1.4 | 3.6 | 4.0 | 6.2 | 6.2 |
| 2026-09 *(Example)* | 1.2 | 9.1 | -0.5 | 9.8 | 9.8 |
| Simple sum | 15.5 | 31.1 | 13.1 | 59.7 | 59.7 |

The simple sum (59.7 bp) does not equal the quarterly arithmetic excess
(60.4 bp), because returns compound across months. Section 3.4 links
the effects so that they add up exactly.

### 3.4 Multi-period linking (Carino)

$$
k_t = \frac{\ln(1 + R_{P,t}) - \ln(1 + R_{B,t})}{R_{P,t} - R_{B,t}}, \qquad
k = \frac{\ln(1 + R_P) - \ln(1 + R_B)}{R_P - R_B}
$$

$$
E^{\text{linked}} = \sum_t \frac{k_t}{k} \, E_t
$$

*Example:* $k$ for the quarter = 0.97130; monthly $k_t$ = 0.97844
(July), 1.00749 (August), 0.98533 (September); scaling factors
$k_t / k$ = 1.00735, 1.03725, 1.01444.

| Sector | Allocation (bp) | Selection (bp) | Interaction (bp) | Total (bp) | Q3 $r^P$ (%) | Q3 $r^B$ (%) |
|---|---|---|---|---|---|---|
| Information technology *(Example)* | 10.2 | 12.4 | 2.5 | 25.1 | 5.17 | 4.68 |
| Health care *(Example)* | -6.5 | 9.2 | 4.3 | 7.0 | 2.31 | 1.71 |
| Industrials *(Example)* | -0.1 | 22.0 | 2.4 | 24.4 | 3.82 | 2.60 |
| Financials *(Example)* | 2.4 | -13.8 | 4.6 | -6.8 | 1.88 | 2.39 |
| Consumer staples *(Example)* | 9.7 | 1.6 | -0.5 | 10.7 | 0.79 | 0.70 |
| **Total Q3 2026** | **15.6** | **31.5** | **13.3** | **60.4** | **3.26** | **2.65** |

The linked total (60.4 bp) equals the arithmetic excess for the quarter
(3.257% minus 2.653%). Segment figures are rounded to 0.1 bp, so rows
and columns can differ from their totals by 0.1 bp (for example,
industrials shows 24.4 bp against a rounded row sum of 24.3 bp). The
August and September segment detail behind the linked figures is in
the appendix workbook; the linked totals reproduce from the monthly
totals in section 3.3 and the scaling factors above (for allocation,
15.7 x 1.00735 minus 1.4 x 1.03725 plus 1.2 x 1.01444 = 15.6 bp). The
Q3 segment returns are linked monthly returns; weighting them by the
constant weights does not reproduce the quarterly totals exactly
(3.24% and 2.64% against 3.257% and 2.653%), because each month
compounds on a different base. Other linking methods (Menchero, GRAP, or a
geometric attribution model that explains the geometric excess)
give slightly different segment numbers; name the method and keep it
stable.

## 4. Currency attribution

<!-- Guidance: Separate the return from currency decisions from the return earned in local markets when the portfolio takes active currency positions. Questions: does the manager decide currency exposure separately, or is currency a by-product of stock picks? Pitfall: showing a currency effect for a manager who does not manage currency, which blames or credits a decision nobody took. -->

Use this section when currency is managed separately (for example, by
an overlay). Otherwise state "Currency is not managed separately;
currency returns are included in the selection and allocation
effects", which is the case for the main example in section 3.

A simplified currency management effect compares portfolio and
benchmark currency exposures, where $e_c$ is the return of currency $c$
against the base currency and $E_B$ is the benchmark's total currency
return:

$$
C = \sum_c \left( w^P_c - w^B_c \right) \left( e_c - E_B \right), \qquad
E_B = \sum_c w^B_c \, e_c
$$

| Currency | Portfolio exposure (%) | Benchmark exposure (%) | Return vs USD (%) | Effect (bp) |
|---|---|---|---|---|
| USD *(Example)* | 60 | 65 | 0.0 | -0.3 |
| EUR *(Example)* | 25 | 20 | 1.2 | 6.3 |
| JPY *(Example)* | 15 | 15 | -2.0 | 0.0 |
| **Total** | **100** | **100** | **$E_B$ = -0.06** | **6.0** |

*Example:* $E_B$ = 0.65 x 0 + 0.20 x 1.2% + 0.15 x (-2.0%) = -0.06%;
EUR effect = 0.05 x (1.2% + 0.06%) = 6.3 bp. A full currency model
(for example Karnosky-Singer) also separates the forward premium,
the hedging cost and the local-market return; use it when forwards are
material.

## 5. Fixed income attribution

<!-- Guidance: For bond portfolios, explain excess return by the risks the team takes: income, yield curve, credit spread and selection. Questions: what duration and spread duration did the portfolio run versus the benchmark, and what did rates and spreads do? Pitfall: applying sector-weight Brinson attribution to bonds, which ignores duration, the main driver. -->

A duration-based decomposition for each portfolio and its benchmark:

$$
r \approx \underbrace{y \cdot \Delta t}_{\text{income}}
\; \underbrace{- D_{\text{mod}} \, \Delta y_{\text{gov}}}_{\text{curve}}
\; \underbrace{- D_{\text{spread}} \, \Delta s}_{\text{spread}}
\; + \; \varepsilon
$$

where $y$ is the yield, $\Delta t$ the period in years, $D_{\text{mod}}$
the modified duration, $\Delta y_{\text{gov}}$ the change in government
yields, $D_{\text{spread}}$ the spread duration, $\Delta s$ the change
in credit spread and $\varepsilon$ the residual (convexity, roll-down,
security selection and pricing).

*Example (Kestrel Bay core bond sleeve, one quarter, parallel fall in
government yields of 25 bp, spreads tighter by 10 bp):*

| Component | Portfolio | Benchmark | Difference (pp) |
|---|---|---|---|
| Yield (% p.a.) | 4.60 | 4.30 | |
| Modified duration (years) | 6.2 | 6.0 | |
| Spread duration (years) | 3.0 | 2.4 | |
| Income (yield x 0.25) (%) | 1.150 | 1.075 | 0.075 |
| Curve (-D x -0.25%) (%) | 1.550 | 1.500 | 0.050 |
| Spread (-D spread x -0.10%) (%) | 0.300 | 0.240 | 0.060 |
| Residual: selection and other (%) | 0.050 | -0.015 | 0.065 |
| **Total return (%)** | **3.050** | **2.800** | **0.250** |

The longer duration added 5 bp as yields fell and the larger credit
exposure added 6 bp as spreads tightened; both are deliberate risk
positions to be judged against the IPS risk budget, not as skill on
their own. Where key-rate durations are available, split the curve
effect into shift, twist and curvature.

## 6. Risk-adjusted measures

<!-- Guidance: Put the excess return in the context of the risk taken. Questions: over what window (36 months is common) and with which risk-free rate? Pitfall: annualising a short window, or comparing a Sharpe ratio built on gross returns with peers' net figures. -->

$$
\text{Sharpe} = \frac{R_P - R_f}{\sigma_P}, \qquad
TE = \sigma\left( R_{P,t} - R_{B,t} \right) \sqrt{12}, \qquad
IR = \frac{R_P - R_B}{TE}
$$

$$
\alpha_J = R_P - \left[ R_f + \beta_P \left( R_B - R_f \right) \right], \qquad
\text{Treynor} = \frac{R_P - R_f}{\beta_P}
$$

*Example:* trailing 36 months to 2026-09-30, annualised from monthly
data, gross of fees, USD. The risk-free rate is the average 3-month
government bill yield over the window.

| Measure | Portfolio | Benchmark | Calculation |
|---|---|---|---|
| Annualised return (%) | 9.80 | 8.60 | Input |
| Risk-free rate (%) | 3.90 | 3.90 | Input |
| Annualised volatility (%) | 14.20 | 14.90 | Input |
| Beta | 0.93 | 1.00 | Input |
| Tracking error (%) | 2.40 | | Input |
| Sharpe ratio | 0.42 | 0.32 | (9.80 - 3.90) / 14.20; (8.60 - 3.90) / 14.90 |
| Information ratio | 0.50 | | (9.80 - 8.60) / 2.40 |
| Jensen's alpha (%) | 1.53 | | 9.80 - [3.90 + 0.93 x (8.60 - 3.90)] |
| Treynor ratio (%) | 6.34 | 4.70 | 5.90 / 0.93; 4.70 / 1.00 |

Report the window, the frequency, the annualisation factor and whether
returns are gross or net. Ratios from fewer than 36 monthly
observations are unstable; label them.

## 7. Contributors and detractors

<!-- Guidance: Name the holdings that explain most of the result, both absolute and relative. Questions: which positions added or cost the most relative to the benchmark, including large benchmark names the portfolio did not own? Pitfall: listing only absolute contributors, which hides the cost of not owning a strong benchmark stock. -->

Absolute contribution is $w^P_j \, r_j$. Active contribution is
$\left( w^P_j - w^B_j \right) \left( r_j - R_B \right)$, using the
quarterly benchmark return of 2.65%. All company names are fictional.

| Holding | $w^P$ (%) | $w^B$ (%) | Q3 return (%) | Absolute contribution (pp) | Active contribution (pp) |
|---|---|---|---|---|---|
| Northgate Instruments *(Example)* | 3.2 | 0.8 | 14.5 | 0.46 | 0.28 |
| Larkspur Biologics *(Example)* | 2.6 | 0.4 | 9.8 | 0.25 | 0.16 |
| Tidewater Rail *(Example)* | 2.1 | 0.5 | 8.2 | 0.17 | 0.09 |
| Brightmoor Bank *(Example)* | 2.4 | 1.1 | -6.3 | -0.15 | -0.12 |
| Halden Foods *(Example)* | 1.9 | 0.6 | -4.1 | -0.08 | -0.09 |
| Orrin Semiconductors, not held *(Example)* | 0.0 | 1.5 | 18.0 | 0.00 | -0.23 |

*Example:* Northgate Instruments active contribution = (3.2% minus
0.8%) x (14.5% minus 2.65%) = 0.024 x 11.85% = 0.28 percentage points.
Not owning Orrin Semiconductors cost 0.015 x 15.35% = 0.23 percentage
points, the largest single detractor.

## 8. Factor exposure

<!-- Guidance: Show the style and macro exposures behind the result, from the firm's risk model. Questions: were the exposures intended? Did a factor explain most of the selection effect? Pitfall: calling a persistent factor tilt (for example quality or low volatility) stock-picking skill. -->

| Factor | Active exposure (standardised) | Factor return in period (%) | Contribution (bp) | Intended? |
|---|---|---|---|---|
| Quality / profitability *(Example)* | 0.45 | 0.40 | 18.0 | Yes, core to the strategy |
| Value *(Example)* | -0.20 | 0.30 | -6.0 | Accepted by-product |
| Momentum *(Example)* | 0.10 | 0.50 | 5.0 | No |
| Size *(Example)* | -0.15 | -0.20 | 3.0 | No |
| Beta / market sensitivity *(Example)* | -0.07 | 0.60 | -4.2 | Yes, lower-risk profile |
| Factor total *(Example)* | | | 15.8 | |
| Specific (stock-specific) *(Example)* | | | 44.6 | |
| **Excess explained** | | | **60.4** | |

*Example:* contribution = active exposure x factor return (for
quality, 0.45 x 0.40% = 18.0 bp). The factor total is 18.0 minus 6.0
plus 5.0 plus 3.0 minus 4.2 = 15.8 bp, and the specific part is 60.4
minus 15.8 = 44.6 bp. Name the risk model, its version and its
estimation universe in section 10; risk model factor definitions are
usually licensed and should be described, not copied.

## 9. Commentary and actions

<!-- Guidance: Turn the numbers into decisions. Questions: which effects reflect intended bets, which surprised the team, and what will change? Pitfall: commentary that restates the tables; explain cause and intent instead. -->

### 9.1 Portfolio manager commentary

[Three to six short paragraphs: market background in one paragraph,
then the main intended bets and how they worked, then the surprises,
then the outlook for positioning. Avoid forecasts presented as
promises.]

*Example:* Industrials selection was the largest positive effect,
driven by Northgate Instruments after it raised guidance on stronger
demand for test equipment. The underweight in financials helped
allocation, but the bank holdings we owned lagged their peers when
net interest margin expectations fell, so selection in the sector was
negative. Not owning the semiconductor name that led the benchmark was
the largest single cost; the stock does not meet our return-on-capital
threshold and we do not plan to buy it for benchmark reasons.

### 9.2 Actions and follow-up

| Action ID | Action | Owner | Due | Status |
|---|---|---|---|---|
| ATR-ACT-001 *(Example)* | Review the investment case for the two weakest bank holdings and report to the investment committee | [@portfolio-manager] | YYYY-MM-DD | Open |
| ATR-ACT-002 *(Example)* | Confirm that the momentum exposure is unintended and set a monitoring limit | [@risk-manager] | YYYY-MM-DD | Open |
| ATR-ACT-003 *(Example)* | Investigate the 0.5 bp reconciliation residual to the official return | [@performance-analyst] | YYYY-MM-DD | Closed |
| [ATR-ACT-NNN] | [Action] | [@owner] | YYYY-MM-DD | [Status] |

## 10. Methodology and data

<!-- Guidance: Document enough that the result can be reproduced and audited. Questions: which model, segmentation, frequency, linking, return basis and data sources? Pitfall: changing classification or linking method mid-year without disclosure, which breaks comparability. -->

| Item | Entry |
|---|---|
| Attribution model | *Example:* Brinson-Fachler, arithmetic, three effects shown (allocation, selection, interaction) |
| Segmentation | *Example:* Sector, from the benchmark provider's classification at period start |
| Frequency and linking | *Example:* Monthly, Carino-linked to quarter and year to date |
| Return basis | *Example:* Gross of management fees, after transaction costs, in USD |
| Holdings or transactions | *Example:* Daily holdings-based with transaction adjustment for intraday trades |
| Cash treatment | *Example:* Cash is its own segment against a zero benchmark weight |
| Currency | *Example:* Not separated; included in segment returns (section 4 shows the overlay sleeve only) |
| Fixed income model | *Example:* Duration and spread duration decomposition (section 5), residual reported |
| Risk measures | *Example:* 36 monthly observations, population standard deviation, annualised by the square root of 12 |
| Risk model | [Vendor or in-house model, version, estimation universe] |
| Data sources | [Holdings: accounting system; prices: pricing policy; benchmark: provider file dated YYYY-MM-DD] |
| Known limitations | [Residual from intraday trading; classification changes; corporate actions] |

```mermaid
flowchart LR
    H[Daily holdings and transactions] --> R[Segment returns]
    B[Benchmark constituents and weights] --> R
    R --> E[Monthly effects]
    E --> L[Carino linking]
    L --> C[Reconcile to official return]
    C --> Q{Residual within tolerance?}
    Q -- No --> X[Investigate data]
    X --> R
    Q -- Yes --> P[PM commentary and approval]
```

Present attribution numbers to clients in a way that is consistent
with the official performance figures, show the period and basis, and
do not imply that past attribution predicts future results. For a US
registered adviser, client-facing attribution of selected holdings may
be "extracted performance" under 17 CFR 275.206(4)-1; check the rule
and the SEC staff Marketing Compliance FAQ before publication.

## 11. Glossary and references

<!-- Guidance: Define the terms a non-specialist reader will meet and list the methods and rules used. -->

### 11.1 Glossary

| Term | Meaning in this document |
|---|---|
| Allocation effect | Return from holding a different weight in a segment than the benchmark |
| Selection effect | Return from holding different securities within a segment, measured at benchmark weight |
| Interaction effect | The joint effect of a different weight and different returns in the same segment |
| Brinson-Fachler (BF) | Allocation measured against the segment return minus the total benchmark return |
| Brinson-Hood-Beebower (BHB) | Allocation measured against the segment return itself |
| Linking | Method that makes single-period effects add up to a multi-period excess return |
| Tracking error | Annualised standard deviation of the difference between portfolio and benchmark returns |
| Information ratio | Excess return per unit of tracking error |
| Sharpe ratio | Return above the risk-free rate per unit of total volatility |
| Spread duration | Sensitivity of price to a change in credit spread |
| Basis point (bp) | One hundredth of a percentage point |

### 11.2 References

| Reference | Version or date | Use |
|---|---|---|
| Brinson, G. and Fachler, N., "Measuring Non-US Equity Portfolio Performance", Journal of Portfolio Management | 1985 | Allocation and selection model (cited by name) |
| Brinson, G., Hood, L. R. and Beebower, G., "Determinants of Portfolio Performance", Financial Analysts Journal | 1986 | Three-effect model (cited by name) |
| Carino, D., "Combining Attribution Effects Over Time", Journal of Performance Measurement | 1999 | Logarithmic linking |
| Karnosky, D. and Singer, B., "Global Asset Management and Performance Attribution" | 1994 | Currency attribution (cited by name) |
| Sharpe, W. F., "Mutual Fund Performance", Journal of Business | 1966 | Sharpe ratio |
| SEC Investment Adviser Marketing Rule, 17 CFR 275.206(4)-1 | As in force 2026-10 | Client-facing presentation (US) |
| SEC Division of Investment Management, Marketing Compliance FAQ | Updated 2025-03-19 and 2026-01-15 | Extracted performance and characteristics |

## Standard and licence note

Based on published academic attribution and risk measures (Brinson
and Fachler 1985; Brinson, Hood and Beebower 1986; Carino linking;
Sharpe ratio), which are methods and are expressed here in original
formulas, and on the SEC Marketing Rule (a US federal work) cited by
section. No textbook, curriculum or vendor material is reproduced. The
worked example (Kestrel Bay Capital) and all companies and figures are
fictional. The text is original. Check the current official versions
of the cited rules before use. This is a document format, not
investment, legal or accounting advice. Template licence: MIT.