03 · Journal
Weekly P&L attribution
Split the week’s P&L into beta, funding, fees, and a residual, so a green number is not treated as skill.
- Level
- Advanced
- Cadence
- Once a week, on the same weekday, using the same benchmark
- Read
- 9 min
- Updated
Definition
Weekly P&L attribution is a rough cut of where the week’s money came from. Beta is a benchmark you chose, times a exposure you assume. Funding and fees come from exports. The residual is what is left. It is not a claim that you have an edge.
Who it is for
Traders who know the week’s P&L and still cannot say what produced it.
Job to be done
Attribute the week’s change in equity to a benchmark move, funding, fees, and a residual you do not call alpha.
Human gate. You accept the cut as a review document or you send it back for a missing export. You do not publish it as a track record.
Inputs and outputs
Inputs
- Start and end equity, or a daily equity export
- Funding and fee totals for the week
- A benchmark and the rough net exposure you are willing to assume
- Any transfers in or out, so they are not mistaken for P&L
Outputs
- Equity change excluding transfers
- A benchmark line labeled as an assumption
- Funding and fees as their own lines
- A residual, named residual, with the unexplained piece called out
Step-by-step workflow
- 01Freeze the windowSame start and end every week, in one timezone you write down. Partial weeks get labeled partial.
- 02Remove transfersDeposits and withdrawals come out of the equity change before you talk about trading.
- 03Book fees and fundingThese are the cleanest lines. If an export is missing, the attribution is incomplete, not zero.
- 04Apply a dumb betaBenchmark move times the net exposure you state. Call it an assumption. Do not fit a regression to flatter the week.
- 05Leave the residualEquity change minus transfers minus fees minus funding minus the beta line. Describe it. Do not praise it.
- 06Human gateWrite one sentence: what you will do differently, or that you will change nothing. The table does not decide.
Suggested tool types
These are categories of tools, not product picks. Crypto Agent Desk does not certify vendors and does not run an affiliate program.
| Type | Why it is here |
|---|---|
| Venue equity, fee, and funding exports | Attribution without exports is a mood. |
| Benchmark price series | The beta line needs the benchmark’s actual move. |
| Spreadsheet | The cut should be recomputable next month. |
| Chat model | Writes the narrative under the table. It should not rename the residual. |
Risks and failure modes
Calling the residual alpha
The residual includes sizing luck, marks, missing fees, and a bad beta guess. Name it residual.
Transfers as profit
Deposits during a good week are a classic self-deception. Split them out before any other line.
One benchmark for a mixed book
A book of alts attributed only to a major will look brilliant or terrible for no reason you can use. Say the assumption.
Operator notes
A green week that is mostly the benchmark is not a skill week. The table is there to make that sentence obvious.
Lines, in order
Transfers, fees, funding, benchmark assumption, residual. If a line is missing data, write “missing” and stop claiming the residual is meaningful.
Prompt you can paste
Attribute this week. Window: START to END, timezone TIMEZONE.
Equity start/end: EQUITY. Transfers: TRANSFERS.
Fees: FEES. Funding: FUNDING.
Benchmark: BENCHMARK moved MOVE. Assumed net exposure: EXPOSURE.
Subtract in that order. Name the last line residual, never alpha.
Say if any input is missing. Do not write a performance summary for other people.
End with a blank line labeled "my decision" for me to fill.
Done when
You can say how much of the week was the benchmark you chose, and you did not need a new adjective for the residual.
FAQ
How precise does the beta need to be?
Crude and stable beats a fitted number you will not remember. The point is to stop treating the whole equity change as discretionary.
What if marks are unreliable?
Say so. Unrealized P&L on a thin book is a different line from realized. Do not blend them without a label.
Can I show this to anyone as performance?
Not from this playbook. It is an internal cut with stated assumptions. A track record needs a standard you did not invent on Sunday.
Published . Updated . This is an operating playbook, not a recommendation to buy, sell, or hold any asset.