Investing & Basis · September 30, 2026 · 13 min read · By CryptoTaxCalc Team

I Bought Bitcoin Every Week for Five Years. Tax Time Was a Quiet Nightmare.

Dollar-cost averaging is the most sensible way to buy crypto. It is also a sneaky way to accumulate 260 different cost bases. Here's what happened when I finally sold part of one weekly-DCA stack, and how the same sale gets taxed in the US, UK, Germany, Canada, Australia, Japan and India.

Image: Illustrative purposes. Rules cited from IRS Form 8949 and §1.1012 basis guidance, HMRC Cryptoassets Manual (Section 104 pool and 30-day rule), §23 EStG and the BMF letter of 19 August 2024, CRA adjusted-cost-base guidance, ATO CGT parcel rules, NTA cryptoasset reporting and India's VDA provisions.

January 2021, I set up the most boring financial decision of my life: $50 of bitcoin, every Monday, no exceptions. FOMO months, capitulation days, three exchanges, one hardware wallet, and a marriage later, I'd completed 260 weekly buys without really looking at the tax side. Buying isn't taxable, so why would I?

Then last autumn I sold a slice to cover a house deposit — about $13,000 worth — and sat down to answer what I thought was a simple question: what did those coins cost me? I opened the exchange export. Hundreds of rows stared back. One bitcoin at $32,000, another at $63,000, a sliver bought when BTC was $47,000, each with its own date and its own fee. I hadn't bought an asset. I'd bought 260 tiny assets, and a partial disposal could touch dozens of them.

Here's the thing nobody warns you about dollar-cost averaging and taxes: the strategy is simple. The basis accounting is not, and the answer changes completely depending on which country you file in. An American can legally choose which coins they sold. A Brit cannot. A German's answer depends on the exact calendar week. Let me walk you through what I learned, country by country, with the actual numbers from my sale.

Why DCA creates a bookkeeping problem, not an investing one

A lump-sum buyer has one acquisition date and one cost basis. A dollar-cost averager has a little ladder of them. That matters every time you sell less than the whole position, because the tax office needs to know which units left the building and what they cost.

Three things compound the problem. First, a partial sale often splits across short-term and long-term holdings — my $13,000 disposal reached coins both older and younger than a year, so the gain came in two flavors taxed at different rates. Second, my coins had lived on three platforms and a hardware wallet, and exchanges don't merge their books for you; the basis from venue A doesn't follow the coins to venue B (this is exactly how $0-basis 1099-DA forms are born). Third, life happened in between: a few staked coins, a bridge, some dust conversions — each event potentially resetting a clock or adding an income entry.

None of this is an argument against DCA. It's an argument for keeping the receipts. We'll get to the system I now use; first, the country mechanics, because they decide what you're even allowed to calculate.

United States: FIFO by default, HIFO if you did your homework

The IRS treats every purchase as a distinct lot. When you sell part of a stack, the default assumption is FIFO — first in, first out — meaning my sale reached January 2021 coins first. I'd bought those at around $32,000. Big gain, but long-term, so 0/15/20% rates.

Specific identification is the escape hatch. You're allowed to say "the coins I sold on October 14 were the ones I bought on these dates," and HIFO — highest in, first out — is just that rule pointed at your most expensive lots. Here's the catch people discover too late: the identification has to be real. You need the lot documented by settlement date, with date, time, amount, price and wallet, and the broker's records have to be able to back it. You can't sort a spreadsheet the night before April 15 and call it an election.

On my numbers, the two methods produced genuinely different answers (rounded):

Smaller gain, but taxed as ordinary income instead of long-term — so HIFO wasn't automatically the winner. I had to run both against my actual bracket, which is exactly the kind of check the free calculator is built for. The broader methods — FIFO, HIFO, specific ID — are broken out with a worked example in the cost basis methods guide. One more wrinkle for DCA-ers: under the per-wallet broker rules now arriving with 1099-DA, coins transferred into an exchange from my hardware wallet showed up with no basis until I attached my own transfer records.

United Kingdom: 260 buys collapse into one number

Britain is the country where my careful lot spreadsheet was simply useless. HMRC's Section 104 pool treats identical coins held in the same capacity as one pooled asset. Every Monday's purchase added quantity and allowable cost to the pool; the sale came out at the pool's average cost, full stop. I could no more choose my expensive 2025 coins than choose which litre of petrol I burned on the way home.

The pool actually makes the arithmetic easy once you accept it — my average acquisition cost was one number, and the gain was proceeds minus the proportionate pool cost. But there's no long-term rate as a consolation prize. A coin held since 2021 and one bought last week both face 18% in the basic band and 24% above it, with the £3,000 annual exempt amount softening the top. Two matching rules sit on top of the pool: the same-day rule matches a sale to coins bought the same day, and the 30-day bed-and-breakfast rule matches it to coins bought in the following month. A weekly buyer has to pay real attention here, because a buy within 30 days of a partial sale gets re-ordered ahead of the pool.

Canada: everyone gets one averaged price, forever

Canada's adjusted cost base works like a stricter British pool: every purchase, including fees, folds into a running average across the entire holding, and the average is recomputed each time you add units. There is no election, no specific ID, no HIFO. My $13,000 sale simply used the current ACB per bitcoin, and half of the resulting gain was included in income at my marginal rate.

Two DCA-specific traps. There is no long-term discount — averaging across five years earns you nothing rate-wise, so don't expect the American 0/15/20% schedule. And the superficial loss rule runs across the whole household: if I sold at a loss while my spouse bought the same coin inside a 61-day window (30 days either side of the sale), the loss was denied and rolled into the ACB. Weekly auto-buys are the classic way to trip this without noticing, so I pause the recurring purchase around any loss sale.

Germany: every Monday in 2021 has its own birthday

This is where my DCA calendar quietly became valuable. In Germany, each weekly purchase is a separate private-sale asset under §23 EStG, with its own 365-day holding clock. When I sold in autumn 2026, the FIFO presumption reached my earliest tranches first — and every single one of them was older than a year, so the entire gain was exempt. Not "taxed at a lower rate." Exempt, with no cap. A dollar-cost averager who sells after their youngest tranche has matured gets the cleanest result in Europe.

The fine print matters more than people think. Coins I staked or lent during the holding window can face a period of up to ten years under the August 2024 BMF letter, depending on the arrangement — plain self-custody staking is on better footing since the Cologne ruling, but exchange staking is the cautious case. And the €1,000 annual private-sale allowance is a cliff, not a deduction: if short-term gains total €1,001, the whole amount is taxable. My unstaked, long-held tranches sailed past both problems; a friend who sold fresh altcoins at €1,400 of gain paid tax on all €1,400. The Germany calculator models the allowance the way the tax office actually applies it.

Australia: parcels, and the 12-month split that rewards patience

The ATO treats each purchase as a separate CGT parcel, which is closer to the American model — but the prize for waiting is different. Parcels held longer than 12 months before the sale get the 50% CGT discount: only half the gain counts. On a partial sale of a DCA stack, you slice the disposal across parcels and apply the discount parcel by parcel. In my case, most of the sold units came from 2021 and 2024 tranches; the 2021 portion got the discount, the 2024 portion didn't, and the numbers were reported separately.

You can choose which parcels you dispose of, but only if your records genuinely identify them — dates, quantities, brokers. A weekly auto-buy with sloppy exports turns this advantage into guesswork. Fees fold into the cost base, and there is no de minimis: even small rebalancing trades are CGT events.

Japan and India: averaging at one end, flat tax at the other

Japan taxes crypto gains as miscellaneous income at progressive rates that run up to about 55% including inhabitant tax, and cost is calculated in yen using either moving-average or total-average across the year's buys — 260 weekly purchases, one averaged yen cost, applied consistently (switching methods needs permission). The ¥200,000 miscellaneous-income threshold is a filing floor, not a tax-free allowance, and resident tax can still apply below it. A reform scheduled for 2028 moves listed crypto to a flat around 20.315%, which will make the DCA math dramatically kinder; until then, a big DCA realization stacks on top of salary.

India is the bluntest system on earth for this question. Every gain is taxed at the flat 30% plus 4% cess — about 31.2% — no matter when you bought, no long-term reward, no rate bracket to optimize. Cost is acquisition cost, essentially nothing else deductible, and VDA losses cannot offset anything. DCA doesn't change the bill at all; what it does create is a 1% TDS record on every disposal, so the department already has a map of my sale before I file. The honest Indian version of DCA planning isn't lot-picking, it's knowing the 31.2% number before you press sell.

The system that made year five painless

Year one, I did nothing and paid for it with a miserable January weekend of reconstruction. After that I built the dullest system in crypto, and tax time now takes an evening:

• Quarterly CSV exports from every venue. All of them, even the exchange I barely use. Platforms prune or lock old records; the dead ones take your history with them. Each export goes into a year folder named the same way every time.
• One master buys sheet. Date and time in UTC, platform, units, price in fiat, fee. That's it. 260 rows, one per Monday. This is the raw material for FIFO, HIFO, the pool or ACB depending on where I file.
• A transfer log with transaction hashes. Every move between my own wallets and exchanges, dated, so a basis-less deposit can be tied to a specific old purchase. This single sheet is what keeps a $0-basis 1099-DA from landing in my life.
• Income events in a separate tab. Staking rewards, airdrops, card rebates paid in crypto — fair market value on arrival, because that number is both income and the new basis.
• Batch-checking locally. When it's time to test a sale, I drop the CSV into the batch import tab, which parses everything in the browser — nothing uploaded — and buckets the rows by holding period. I can compare FIFO against specific-ID scenarios before I sell, not after.

That last point is the actual lesson. The expensive DCA mistakes I see aren't caused by the strategy; they're caused by discovering the method question in April instead of October. Run the numbers while you still control the date of the sale.

The takeaway

DCA gets you better behavior. Good records get you the tax rate your behavior earned. Same weekly buys, same bitcoin, completely different bill depending on whether you're in Austin, Aberdeen or Adelaide — and depending on whether you can prove what every Monday cost.

Two things you can do tonight: export the last two years of CSVs from every exchange you've used, and put a repeating quarterly reminder on your calendar for the next one. Then, before your next partial sale, run the lot scenarios for your country in the calculator.

How do you handle the DCA stack — strict FIFO, specific ID where allowed, or just average everything like the pool countries do? Tell me in the comments which one matches your country; I'm collecting the weird edge cases.

General information, not tax advice. US 1099-DA basis reporting and the wash-sale proposals in H.R. 10357 are still phasing in, and Germany's treatment of staked tranches continues to evolve. For large disposals or amended returns, work with a tax professional in your filing country.

FAQ