Sample UK Crypto Tax Reconstruction Workpaper

What came back

Six sources, 13,686 source rows normalised across three tax years (6 April 2023 to 4 April 2026), of which 2025/26 is the sample year. Pre-2025/26 history is loaded so the Section 104 pools carry into the year rather than starting empty. 180 Coinbase rows arrived twice and were de-duplicated on trade id. Two source rows were excluded at import, and they are named here rather than left silently absent.

All 1,079 disposals are included, reported as 2,996 disposal rows — a sale matched across several acquisitions is several rows, because each same-day, 30-day and Section 104 leg carries its own cost. The traced sale below is twelve. Acquisition cost and incidental costs of disposal are stored in separate columns and both are shown, so the reviewer decides how they are combined on the form. Net gain is positive, so the £0.00 carried forward is a computed result, not an unfilled line.

Miscellaneous income — staking and airdrops. 236 receipts, £5,407.52 in 2025/26. None unpriced, none valued at an approximate price. Reported separately from capital gains and not netted against them. One further airdrop, TXN-41629, was pooled as a capital acquisition rather than recognised as income, so it is not in this figure.

The file we were handed

Four wallets and two exchanges, one of which stops. The FTX export ends on 28 February 2024, which is the interesting part for a reconstruction: nothing it holds is disposed of inside 2025/26, but the cost it establishes flows into the Section 104 pools the 2025/26 disposals draw on. A reconstruction that started at 6 April 2025 would price those disposals against an empty pool.

Row counts are of normalised source transactions across all three tax years, not of disposals: 13,686 in total, 6 April 2023 to 4 April 2026. Being fully imported is a statement about coverage of the period, not about whether every row carries enough evidence to be taxed. Those are separate questions and the pack answers them separately.

What needed a decision

Every item below is something the pipeline could have quietly guessed at and did not. Four of the pack's six categories are shown here; the register in the workpaper carries all six, each with the evidence behind it, the handling we propose, and a decision column that is deliberately blank. A row that says matched means two records were tied to each other. It does not mean a treatment was approved.

One disposal, traced end to end

Every figure traces back to a specific source row. This is a single ARB sale that consumes all three UK matching rules in the statutory order — twelve legs for one sale, which is the shape that gets lost when a tool reports at the transaction level instead of the leg level.

These are the first three of twelve legs; the remaining nine are set out on page 6 of the workpaper. Two of the three show a loss because those acquisitions cost slightly more than the sale realised — each same-day and 30-day leg is costed at its own acquisition cost rather than at an average, while the two Section 104 legs share the pooled average of 1.178557 £/ARB. Proceeds − allowable cost − incidental costs = gain holds on every leg above, and on all 2,996 disposal rows in the year, with 0 rows failing and a maximum residual of 0.0000. That note is in the workpaper too — a reviewer should never have to discover a convention by tying out a total themselves.

If you want this on a real file

Two routes, and neither needs client data to leave your firm.