Crypto Trading Bot Tax UK: Grid Bots, DCA Bots and HMRC Records

Grid bots on 3Commas, DCA bots on Pionex, signal-following bots on Cryptohopper — automation does not change the tax treatment, it changes the volume. A grid bot running on a £5,000 BTC/USDT pair can generate 200 disposals per month. Each one is a CGT event in the UK. Most bot users discover this at filing time. Here is how to stay ahead of it.

Every bot trade is a disposal

HMRC does not distinguish between a manual swap and a bot-executed swap. The same rules apply: token-to-token is a disposal, same-day matching, 30-day rule, Section 104 pool. The bot is just clicking the button for you.

This means a £5,000 BTC/USDT grid that flips 12 times in a day produces 12 disposals — not 1. Each must be matched, valued at the block-time GBP price, and entered into the pool calculation.

The trader-versus-investor question

Volume alone does not make you a "trader" for UK tax purposes — HMRC's badges-of-trade test asks about frequency, organisation, motive, finance, and short-term holding. But heavy bot use moves the needle. A retail investor with 50 manual disposals a year is almost certainly an investor (CGT). The same person running three grid bots that produce 5,000 disposals a year may be a trader (income tax + NI).

The practical threshold most UK accountants apply: if your bot activity looks like a part-time job — daily monitoring, capital recycled rapidly, profit-seeking as primary motive — HMRC will likely treat it as trading. If it is set-and-forget DCA, it is investing.

Trading status has both costs (income tax rates up to 45%, Class 4 NI) and benefits (losses offsettable against other income, business expenses allowable). It is rarely a clear win either way.

Records bots make hard

The single biggest practical problem with bot taxation is record fidelity. 3Commas exports an order-level CSV. Pionex provides trade history but truncates older data. KuCoin's API often returns inconsistent timestamps. By the time a UK filer realises they need every line, six months of data may already be gone.

The minimum you need per trade: timestamp, pair, side, base quantity, quote quantity, fee, and the GBP value of the base token at the timestamp. Without the GBP price feed at block time, your filing is approximate at best.

Funding and gas spend

Bot fees (3Commas subscription, Pionex spreads, exchange API fees) are not deductible against CGT — they would only be deductible if you are classified as a trader and the expenses are wholly and exclusively for the trade. For investors, fees are absorbed into the cost basis of the relevant disposal where they directly relate.

Stablecoin pair bots

A USDT/USDC grid bot still triggers disposals despite the negligible GBP movement. The Section 104 pool figures might net close to zero, but the disposals must be reported. CARF reporting from January 2026 means UK-regulated exchanges send the gross trade volume to HMRC — a return showing five fiat sales when the exchange reported £400,000 of token movement is the classic CARF mismatch.

Practical filing approach

Export everything monthly, not annually. Match the export to a tax engine that does block-time pricing rather than end-of-day pricing — bot trades happen at peaks and troughs and the difference can be material on tight margins. Decide trader-vs-investor before filing, not after; reclassifying later is expensive in time and accountant fees.

CryptoLens scans exchange API exports as ordinary disposal sets and applies block-time GBP pricing to each line, producing a single SA108 row per trade regardless of how many the bot generated.