Filing Strategy · September 30, 2026 · 13 min read · By CryptoTaxCalc Team

I Filed Crypto Taxes 3 Ways. Here's What DIY, Software and a CPA Actually Cost.

Year one I hand-typed a Form 8949 for free. Year four I paid software about $300. Year six I wrote a CPA a check for $2,200 and shook his hand. All three were the correct decision — for that particular year. Here's the honest price ladder and the five-question test I now use to pick.

Image: Illustrative purposes. Price ranges reflect commonly advertised 2026 US prep fees and software tiers, converted intuitively for UK, German, Canadian, Australian, Japanese and Indian equivalents.

Every November the same argument appears in my group chats: one guy insists crypto taxes are easy and free; another has just been quoted $4,000 by an accountant and is convinced he's being robbed; a third is using software but doesn't trust a single number on the report. All three are describing the same elephant from different sides of it. The method isn't good or bad — it fits or it doesn't.

I've now lived through all three routes, plus the hybrid. This is what each one cost, what it felt like, and the exact portfolio shape each one suits.

The price ladder in one table

These are typical 2026 ranges for a US individual investor; other countries are cheaper in absolute fees for simpler filings but follow the same shape — manual filing (HMRC self-assessment, ELSTER, MyTax, NETFILE, e-Tax, ITR) at zero, tracker subscriptions, then adviser fees. Your numbers will vary with transaction volume and city.

Route 1: pure DIY, the year it actually worked

My first crypto tax year contained eleven transactions, all on one US exchange: eight buys, two sells, one tiny staking payout shown clearly on the tax form the exchange sent. I downloaded the CSV, wrote the two sales onto Form 8949, and filed. Cost: nothing beyond my usual filing. Time: one Saturday afternoon, including reading the instructions twice. I still sanity-checked the gain with an online calculator, because typing a number you can't independently verify is how people transpose digits.

DIY is genuinely fine when all of these hold: one venue (or two with zero transfers between them); only fiat buys and fiat sells — no coin-to-coin trades, because those are taxable barters you'll have to price manually; no staking, mining, airdrops, NFTs, DeFi or wrapped tokens; a few hundred transactions maximum; and complete records. If that's you, paying for software is ego, and paying a CPA is burning money.

Where DIY quietly dies is the transfer. The moment coins move from Exchange A to a wallet and onward to Exchange B, the second venue's CSV shows a deposit with no basis — and a naive filing reads that deposit as income or a zero-basis asset. This is the single error behind the scary missing-basis 1099-DA notices that started arriving in 2026. If you've ever moved coins off-exchange, you have outgrown pure DIY even if your gain count looks small.

Route 2: software, the workhorse years

By year four I had three exchanges, a hardware wallet, some ETH swapped into half a dozen tokens, and weekly staking rewards. My spreadsheet was approaching nine hundred rows and I'd caught two of my own matching mistakes. I subscribed to a tracker, connected the APIs and imported the on-chain wallets. The paid tiers for reputable products run roughly $49 at the entry level to around $400–500 for thousands of transactions with full form export. I paid about $300 and would have paid more.

What software is brilliant at: ingesting dozens of venues, matching every lot under FIFO/HIFO/specific-ID or the UK §104 pool, pricing historical on-chain swaps, and emitting Form 8949 / international reports. The tool comparison page lays out how the main products differ, and the two I rotate between are in the cards below.

What it is not brilliant at, and where people get overconfident: garbage in is still garbage. My first import labeled three internal wallet transfers as sales, which would have invented $40,000 of phantom income; I caught it because I reconciled ending balances per coin against what the wallets actually held. Every expert I've spoken to does the same four-eye check: wallet by wallet, the software's ending balance must equal reality, or nothing downstream is trustworthy. I wrote my own full cleanup procedure into a weekly record-keeping system precisely so January's import takes an hour, not a weekend.

Software also can't make judgment calls. Whether you're an investor or in a trade or business, whether a particular chain interaction was a loan or a sale, whether a dead token qualifies for a worthlessness claim — the app will happily label whatever you click. You are signing the return, and the help desk is not your representative in an audit.

Route 3: the CPA, the year I stopped pretending

Year six: I'd used a couple of DeFi protocols across chains, received tokens from a bridge that required three pages of footnotes, moved between states, and wanted to claim a worthless-token position that had actually gone through a near-zero exit swap rather than a clean abandonment. The software produced numbers; what I needed was someone to argue with the classification. I found a CPA through a referral from another client in crypto, asked specifically how many digital-asset returns he'd signed that year, and paid $2,200 all-in for the federal return plus state. He caught two things worth roughly ten times his fee, including a cost-basis treatment I'd defaulted to wrong.

Crypto-knowledgeable CPAs and enrolled agents typically run $150–400+ per hour. A clean investor return built from software reports might be $500–1,500; DeFi, margin, NFTs and foreign-account schedules push it to $2,000–5,000. That sounds expensive until you see the next row of the table: people who skipped years and later need reconstruction — old wallet forensics, dead-venue claims, amended 1040s, FBARs, voluntary-disclosure-style remediation — routinely pay $5,000–15,000. Procrastination is the premium product in this market.

The honest caveat: hiring a CPA does not outsource your data problem. I still had to produce exports, label transfers and answer thirty questions. What I bought was judgment, signature responsibility, and a phone number that rings if a letter arrives. The audit playbook explains why that last item matters: representation is the thing software literally cannot sell you.

The hybrid most people actually need

Since then I've run the hybrid, and I think it's the unspoken best answer for the serious-but-not-professional investor. I buy the software, do the import myself in January, spend the cleanup hours, and then pay a professional for a review engagement — typically $500–1,500 on top of the subscription — to check methodology and sign. Total maybe $800 in a normal year. I get software's consistency and a human's veto over the two or three transactions where the tax code is genuinely gray. I book the review in February; by late March the good people are booked, and the people still free are sometimes available for a reason.

The five-question decision test

Run down this list before you spend a dollar:

1. Have coins ever moved between two venues, or lived in a self-custody wallet? If no and everything else is simple, DIY works. If yes, start from software, because basis doesn't travel with the coin.
2. Anything exotic — DeFi liquidity, bridging, NFTs, margin, mining as a gig, wrapped tokens? Software for the volume, professional review for the classification. I wouldn't file a pure DIY return with any of these.
3. Any unfiled years, an IRS/HMRC/ATO letter, or an amended return in mind? Professional, full stop. These have procedure rules (substantial-justification, discovery windows, voluntary disclosure) where a self-guided mistake compounds.
4. Did you change countries or states, or hold large foreign balances? Professional. Dual-residency years and foreign-account reporting are exactly the situation the cross-border guide warns about; the forms alone justify the fee.
5. What's at stake? If a single ambiguous position is worth, say, $50,000 of tax, a $2,000 review costs four percent of the exposure and is obvious insurance. If your total gain is $400, DIY with a calculator is obvious in the other direction.

If you hire: three screening questions

Anyone can put "crypto" on a website. Ask these before engaging: how many digital-asset returns have you signed in the last two filing seasons? (You want dozens, not two.) How do you handle a transferred-in asset showing zero basis on a 1099-DA? (They should immediately mention basis reconstruction and reconciliation — the procedure in our reconciliation guide.) And: do you represent clients in exams, and who signs the return? A vague answer to any of the three tells you everything. Also confirm the fee model up front — fixed per return with a stated scope beats open-ended hourly for a standard investor filing.

Buy the method that matches the year, not your ego

The expensive mistake isn't choosing any one of these three routes. It's using last year's method on this year's portfolio — DIYing a DeFi year out of habit, or paying a CPA for a year that contained four Coinbase sells out of anxiety. Complexity changed; the answer changes with it.

This week, pull up your actual transaction list and take the five-question test. If it's still simple, file free and feel smug. If it's outgrown you, set software up now while every API works and the calendar is empty — that's the difference between a $300 January and a $10,000 March.

This is general information, not personalized tax advice; the fee ranges are typical market figures, not quotes. Interview any professional and confirm their digital-asset experience in your own country before engaging.

FAQ