# Pnl Attribution

> Ingest market data, backtest a supplied multi-factor strategy with Qlib, and render the per-factor performance breakdown in `backtest_report` as a formatted attribution document. Use when a researcher wants an evidence-linked P&L attribution report; this workflow does not place trades.

- Skill: `knuckles-team/pnl-attribution` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add knuckles-team/pnl-attribution`
- Raw SKILL.md: https://api.skillmd.com/api/skills/knuckles-team/pnl-attribution/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Docs & Writing
- License: MIT
- Author: Knuckles-Team (https://skillmd.com/u/knuckles-team)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/knuckles-team/pnl-attribution

---


# PnL Attribution Workflow

Compose the named atomic skills without adding new attribution math here.

## Inputs

Provide the multi-factor strategy definition, asset universe, and date range.

## Steps

### Step 0: quant-data-ingest [skill: quant-data-ingest]

Invoke `$quant-data-ingest` with the workflow inputs to ingest market
data for the universe and date range.

Expected: `normalized_market_dataset`

### Step 1: qlib-backtester [skill: qlib-backtester] [depends_on: Step 0]

Invoke `$qlib-backtester` with `normalized_market_dataset` and the
multi-factor strategy definition to produce per-factor performance.

Expected: `backtest_report`

### Step 2: document-converter [skill: document-converter] [depends_on: Step 1]

Invoke `$document-converter` with `backtest_report` to render the
per-factor breakdown as a formatted attribution document.

Expected: `attribution_document`

## Output

Return `backtest_report` and `attribution_document`. Does not place trades.

## Execution

- **Run first:** Step 0 — `$quant-data-ingest`.
- **After level 0:** Step 1 — `$qlib-backtester`.
- **After level 1:** Step 2 — `$document-converter`.

**Execution:** If graph-os is reachable, offload the whole DAG via `graph_orchestrate action=execute_workflow` (or the `kg-delegate` skill) for true parallel/swarm execution. Otherwise execute the steps natively in dependency order: run steps with no unmet `depends_on` in parallel, then their dependents.