← Glossary

Devancore Post-Trade Glossary

Performance Attribution Record

A performance attribution record is the controlled operating record that explains portfolio return using positions, weights, prices, cash flows, benchmarks, classifications, FX, corporate actions, and evidence.

Definition

A performance attribution record is the controlled operating record that explains portfolio return using source data that can be reviewed. The performance number alone is not enough. A firm needs to know which positions, weights, benchmark constituents, prices, FX rates, classifications, corporate actions, cash flows, model settings, and overrides created the explanation.

Performance attribution should be connected to the performance book of record, but it should not be treated only as analytics output. Institutional teams use attribution to explain investment decisions, client reporting, composite performance, risk review, mandate oversight, and recurring performance questions. Those outputs depend on operating data quality.

Attribution record inputs

Attribution record inputs

The record should explain both the result and the data used to produce it.

Input Examples Control question
Portfolio state Positions, weights, cash, unsettled trades, accrued income, and corporate actions Does the attribution use the same operating state as IBOR and PBOR?
Benchmark state Constituents, weights, returns, rebalances, classifications, and version Can the benchmark used for the calculation be reconstructed?
Classification Sector, industry, asset class, strategy, issuer, country, currency, and custom sleeves Are portfolio and benchmark labels aligned?
Market data Prices, FX rates, curves, yields, spreads, and valuation inputs Which source and timestamp created the return input?
Model output Allocation, selection, interaction, factor, income, currency, curve, spread, and residual effects Can each effect be traced to data and method?
Evidence Source files, versions, adjustments, approvals, comments, exceptions, and reports Can the report be supported without reconstructing the workflow manually?

The portfolio state is the foundation. Positions, weights, cash, unsettled activity, accrued income, fees, corporate actions, and account hierarchy determine what is being measured. If the PBOR does not align with IBOR and ABOR where expected, attribution may explain a data difference rather than an investment result.

The benchmark state is the comparison set. Constituents, weights, returns, rebalances, classifications, and versions should be preserved. A benchmark reconstitution, sector reclassification, or stale file can change allocation and selection effects without any portfolio decision changing.

Classifications are control data, not labels added for presentation. Sector, industry, issuer, country, currency, strategy, sleeve, asset class, and custom taxonomies determine how attribution buckets are calculated. If a security has one sector in the portfolio book and another in the benchmark file, attribution residuals may be operational noise.

Market data supplies the return inputs. Prices, FX rates, curves, yields, spreads, and valuation timestamps should be tied to source and version. Fixed income attribution is especially sensitive because return can be decomposed across rates, curve, credit spread, income, carry, roll-down, and currency effects.

Evidence completes the record. Source files, API events, versions, manual adjustments, approvals, exception comments, methodology versions, and report outputs should remain attached to the calculation. Without this chain, a team can show an attribution report but cannot defend how it was produced.

How it works

A performance attribution record works by turning portfolio and benchmark data into an explainable return record. The process starts before the model runs. It starts with controlled inputs.

Performance attribution control points

Performance attribution control points

The workflow should control the inputs before it explains the return.

Control point Record checked Failure mode
Position basis IBOR, ABOR, PBOR, settled state, unsettled trades, cash, accruals, and corporate actions Attribution uses a position set that does not match the operating books
Benchmark version Constituents, weights, returns, rebalance date, and provider file version Client report uses a benchmark state that cannot be reproduced
Classification alignment Sector, industry, asset class, issuer, country, strategy, sleeve, and custom taxonomy Selection or allocation effect is caused by mismatched labels
Price and FX source Price file, FX rate, valuation timestamp, curve input, and fallback rule Return differs because sources or timestamps differ
Override review Manual adjustment, reason, owner, approver, expiry, and affected accounts Performance result includes unsupported intervention
Reporting package Calculation version, methodology, inputs, exception state, and client-reporting output Attribution number cannot be tied to evidence

Position basis control checks whether the calculation uses the correct operating state. Some attribution workflows use beginning-of-period weights. Some use daily linked returns. Some need settled positions. Others need expected IBOR state. The record should show which basis was used.

Benchmark control preserves the comparison set. The benchmark file should carry constituent list, weights, returns, rebalance date, classification, and version. If a benchmark provider updates a file, the attribution record should show which version supported the report.

Classification control aligns portfolio and benchmark categories. Allocation and selection effects depend on buckets. Sector, industry, country, currency, asset class, strategy, issuer, and sleeve mappings should be governed because a label change can move basis points across attribution categories.

Price and FX control explains return input differences. A position return can differ because of price source, valuation timestamp, FX rate, holiday calendar, stale price, or fallback method. Those fields should be part of the record, not hidden in downstream analytics.

Override review is a controlled event. A price correction, benchmark adjustment, classification change, cash-flow correction, or model setting change may be appropriate. It still needs owner, reason, approver, effective date, affected accounts, and reporting impact.

Reporting package control ties the result to evidence. Client reporting, performance review, composite support, and supervision should point to a calculation version, input set, methodology, exception state, and approval trail.

Attribution data — record to evidence map

Devancore · message matrix

Rail Message Purpose Record
IBOR / PBOR portfolio state weight basis positions, cash, unsettled activity, accruals, corporate actions, account, and portfolio hierarchy
Benchmark index state relative comparison constituents, weights, returns, rebalance date, classification, and source version
Market data return inputs valuation prices, FX, curves, yields, spreads, timestamps, fallback rules, and source lineage
Model attribution effects return explain allocation, selection, interaction, factor, income, currency, curve, spread, and residual output
Control review package evidence exceptions, overrides, approvals, methodology version, reconciliation result, and reporting output

In Devancore™

Devancore supports performance attribution records by maintaining controlled operating data around positions, cash, accruals, benchmark state, prices, FX, classifications, corporate actions, model context, exceptions, approvals, and reporting outputs. The platform should be framed as a post-trade record and workflow layer, not as a performance consultant, investment adviser, index provider, pricing vendor, accountant, fund administrator, or reporting authority.

In a Devancore-style workflow, attribution starts from records that already carry lineage: IBOR positions, ABOR-confirmed state, PBOR performance state, cash flows, corporate action effects, benchmark files, classification versions, price sources, FX rates, and curve data. The attribution record can show which inputs fed the calculation and which output consumed the result.

This helps teams identify residuals caused by data issues, benchmark version mismatches, stale prices, incorrect classifications, missing corporate actions, FX timing, unsettled activity, or manual overrides. A performance analyst should not need to reconstruct those causes from separate systems after a client question arrives.

The same record can support performance review, client reporting preparation, mandate oversight, composite support, risk review, regulatory reporting support, and audit trail. The practical value is that the attribution result stays connected to the operating record rather than becoming a disconnected analytics artifact.

Conversational finance becomes useful when the attribution record is structured. A user may ask which securities drove selection effect, which benchmark version was used, which classification changes affected allocation, which fixed income positions moved because of credit spread, which FX rates were applied, or which reports include manual overrides. The answer should resolve to positions, benchmark records, source data, model context, owners, and evidence.