HONEST COMPARISON · FIGURES CHECKED 2026-08-29
ClearBasis vs Blockpit
EU and DACH filers who want a report written for their own tax office, priced in euros with a published ceiling. We are a different shape of product: narrower today, and built so that every number can be traced to the record it came from. Below is what they do better, where we differ, and a table where each figure is the one the vendor published, linked to the page it came from.
What Blockpit does better
Starting anywhere else would waste your time.
- Its report is written for a tax office rather than for a user: per-disposal cost basis, proceeds, holding period and gain with the statute cited inline (§23 EStG, §20 Abs. 2 EStG) — visible in their own published sample PDF.
- It is honest about depth versus breadth, naming the ten countries with real country-specific logic instead of hiding behind a '100+ countries' claim.
- Five flat euro prices with a published ceiling of 549€ and no open-ended top tier.
Where we are different
Every figure carries its evidence. Every figure links to the row it came from — source file, row number, tx hash, price source — and rulesets are versioned and dated (2026.08-r1) with the methodology memos (M-001…) published rather than internal. The country reports carry BOTH halves on the same line: the statute that decided the figure AND the source row that produced it, in the language of the office that reads it. Germany's report is in German, cites § 23 Abs. 1 Satz 1 Nr. 2 EStG per disposal with the holding period shown, and treats the Freigrenze as the THRESHOLD it is — reach it and the whole gain is taxable, not the excess, which is the error that is always in the taxpayer's favour and therefore never noticed until somebody with authority notices it. The UK is the one place that sentence does not describe a report, and we would rather say so on the same page than let two claims disagree: there is no UK report, deliberately — SA108 wants sterling at each transaction's own date and we will not invent it. What names which of s.104, s.105(1)(b) or s.106A(5) matched each disposal is the free public tool at /uk-share-pooling, against trades you type yourself. A citation tells a reviewer which rule was applied; it says nothing about where the number came from, and we print both.
The answer is free; the paperwork is paid. Everything computed and visible — true gain, the blank-basis overstatement, per-lot detail in preview, gaps and decisions. Payment unlocks the documents (Form 8949, workpapers, exports); nothing about the answer is hidden to sell the answer.
Your basis is portable by design. A documented read-only HTTP API over your own scans — transactions, disposals with FULL LOT LINEAGE, decisions, holdings and the completeness report, every figure carrying the source file, row, tx hash and price source behind it, and every response stamped with the dated ruleset that produced it. Read-only is a property of the design, not a scope string: the router registers GET routes only and the token table has no write concept to widen. Plus two free files from any saved scan even if you never pay — your normalised ledger as CSV in the universal template we publish, so it re-imports here and anywhere else implementing the format, and the Basis Passport, a signed evidence-linked basis history whose spec is public and which anyone, including these competitors, may implement.
Dead platforms are the specialty. Dead-platform rebuild is the specialty: FTX, Celsius, Cryptopia, Mt. Gox and others, reconstructed from what survives, with the gaps shown rather than filled in.
Side by side
| ClearBasis | Blockpit | |
|---|---|---|
| What you get without paying | Everything computed and visible — true gain, the blank-basis overstatement, per-lot detail in preview, gaps and decisions. Payment unlocks the documents (Form 8949, workpapers, exports); nothing about the answer is hidden to sell the answer. | Import unlimited sources and track the portfolio; the tax report requires a paid plan for that year, and 'Tax Figures' sit behind the separate Blockpit Plus subscription (47.90€/yr). |
| Price at ~1,000 transactions | $99 per tax-year bundle (≤1,000 tx) | $109 per tax year on their US page (Basic, ≤1,000 tx); the EU pages run a different ladder in euro |
| Above that | $249 (≤5,000 tx) · $299 (≤15,000) · $449 (≤50,000). Cut from $399 and $649 on 28 August 2026 after re-reading every vendor page in this table: at 20,000 transactions the market runs $259 at the cheapest against a median near $649, and our standing rule is to sit about half the comparable. We are not trying to be the cheapest row here — Blockpit is, at $269 for 25,000 — we are trying to be honestly priced against the middle. | US page: $269 (Pro, ≤25,000 tx) · $639 (Unlimited, ≤500,000). That is the best price-per-transaction at the top of this table by a wide margin. Their non-US pages use different tier names and currencies and were inconsistent between fetches, so treat euro figures as unverified |
| Four unfiled years — one payment or four? | Four payments, discounted. The first year is the band price and every additional year is half — so four years at the $99 band is $99 + three at $49.50, $247.50 in total. Said plainly, because it matters: for a straightforward low-volume catch-up a subscription competitor can be cheaper than us. What you get for the difference is one continuous ledger where the basis for a 2025 sale can be a 2021 purchase, and evidence attached to every figure. And the documents come out PER YEAR while the ledger stays whole, which is the part that is easy to get wrong in the other direction: ask for 2023 and you get 2023's disposals with the basis still carried from the 2021 purchase, because the engine replays the entire history and scopes only the output. A tool that instead filters the FILE to 2023 finds no acquisition for the coin sold that year, opens a zero-basis lot, and reports $25,000 of gain where there was $15,000. We also refuse to hand a multi-year ledger to a single document at all — a 2024 return must not carry 2022 disposals, and a TXF is a machine import that will not notice — so the year is asked for, and it is in the filename. Until 1 September 2026 we did neither: one 8949 came out with three years in it, named after the scan. If all you need is four separate reports, buy the cheaper thing. | Four payments. One-off licence per tax year, no subscription, any year back to 2011. Four years at the $109 US band is about $436. |
| Connections | 13 auto-detected CSV formats, including the 1099-DA itself and migration files from Koinly, CoinTracker, CoinLedger and CoinTracking — and, for everything else, COLUMN MAPPING: a file we do not recognise is not a dead end, it is one question. We show you your own first rows, say which column we think is which and why, and you correct us. This is the honest answer to a competitor advertising eight hundred exchanges: a list of eight hundred still has somebody standing just past the end of it, and a mapper has no end. Two things it deliberately will not do — it will not assign a numeric column by shape alone, because most exports have three and picking one would be a coin toss dressed as an inference; and it will never guess what a TRANSACTION TYPE means. Whether the word in your export is a sale or a transfer to your own wallet decides whether tax is owed and the row looks identical either way, so every distinct value is listed, we say what we think and how sure we are, and anything left unanswered is reported as skipped rather than assumed. Headers in a language we never listed still work: where the name means nothing to us the VALUES are read instead, and a column whose every cell parses as a date is offered as the date. Layouts we are taught are remembered so the next person with the same export gets a filled-in form — shown for confirmation every time, never applied behind anyone's back. DATES ARE THE OTHER HALF OF THAT, and we shipped it broken: a column of 12/31/2024 was offered with high confidence and then refused row by row, because the layer underneath read ISO and only ISO, so a European or US-slash export imported as NOTHING behind a form that looked right. It now normalises — and it will not do it row by row, because 03/04/2024 is the 4th of March in Boston and the 3rd of April in Riga, and choosing wrong moves disposals between tax years and decides whether a holding crossed the one-year line. The COLUMN decides where it can (one day past the twelfth settles the convention for the whole file) and where it cannot, we ask, in the same form, before anything is read. A column that contradicts itself is never normalised. A disposal written as a NEGATIVE quantity, which every accounting-shaped export does, used to be dropped as a non-positive amount — so a file imported with its purchases and without its sales, and the tax came out too LOW, which is the one direction our defaults are not allowed to fail in. And three of the nine actions the mapper offers — staking, interest, airdrop — were accepted and then silently discarded, which on a real file meant $0 of ordinary income against $1,620 and $2,040 of gain against $1,440: wrong in both directions at once. All three are fixed, all three have gates written against the MENU rather than the symptom, and we publish them because a mapper is only worth anything if the files it accepts actually arrive. Plus keyless wallet import on NINETEEN chains needing no API key at all — Bitcoin, Ethereum, Base, Optimism, Arbitrum One, Polygon, Gnosis, Scroll, zkSync Era, Mode, Unichain, Ink, Lisk, Soneium, Shape, World Chain, Celo, Immutable zkEVM and Rootstock. Every one of those was queried with a real address before it was listed, and the coin that pays for gas was corroborated against the explorer's own quoted price rather than taken from a documentation page: a chain whose gas coin is labelled wrong spends the wrong asset's lots on a fee and never says so. Bitcoin takes an xpub, ypub or zpub and derives the whole wallet, receive and change chains both, because a single Bitcoin address is one receipt out of hundreds; the key cannot sign anything and an xprv is refused on sight. Said plainly: nineteen chains, not 190, and no read-only exchange API connections. Solana is deliberately NOT shipped: we measured the keyless endpoints and the only one with usable history rate-limits transaction reads to about seven in ten, which makes a partial ledger the normal outcome — and a partial ledger that looks complete is the one result worse than no import. That is the remaining gap and it is a real one. | Their own pages state 700+ in one place and 160+ exchanges / 190+ blockchains / 2,500+ dApps in another. API and CSV supported. |
| Country depth | Thirteen cited country guides — Latvia, Estonia and Lithuania in five languages, plus Germany, Portugal, Spain, the UK, the Netherlands, Italy, Switzerland, the UAE, Singapore, Malta and South Korea in English — a global hub, and a fourteen-jurisdiction comparison with the exit taxes. NINE of those are now more than guides, and each is a genuinely different method rather than the American one reskinned: the UK pools under TCGA 1992 s.104 with s.105 same-day and s.106A(5) 30-day matching in front; Germany applies § 23 EStG, where more than a year is not taxable AT ALL rather than taxed at a lower rate, with the Freigrenze as the threshold it is; Canada averages every identical property into one adjusted cost base across ALL accounts under s.47(1) and denies superficial losses under s.54 — a rule with no US equivalent for crypto, so an engine ported north lets you claim a loss you are not entitled to. Australia discounts half a gain on a parcel held more than twelve months, but only AFTER capital losses are applied — the method statement in s 102-5(1) is a numbered list and doing step 4 before step 2 makes every dollar of loss do half its work, an error always in the taxpayer's favour and therefore never queried; its year also ends on 30 June, so a calendar-year split reports a different set of transactions rather than a slightly different answer. France is the one that breaks every engine built somewhere else, in two independent ways: article 150 VH bis computes the gain as a fraction of what the WHOLE portfolio cost, sized by its value at the moment of the disposal — the asset you sold appears nowhere in the formula, so a lot-based answer ported onto a French return is a different quantity wearing the same word — and crypto-to-crypto swaps without a soulte are not taxable at all, so a US-shaped engine invents a whole year of gains that do not exist in French law. A French net annual loss also dies on 31 December: every other jurisdiction here carries losses forward, and applying that habit to France makes somebody under-declare a later year. What France needs and no lot-based tool has is the portfolio's value at each disposal; it sits in the denominator, so we ask for it and refuse to estimate it, and a disposal without one is reported as needing it rather than quietly computed. Sweden looks like the simplest of them until you notice that a gain is taxed in full and a loss is deductible only to 70% — so a 10,000 kr gain against a 10,000 kr loss is not nil, it is 3,000 kr of capital income and 900 kr of tax on a year somebody made nothing; every other jurisdiction here nets krona for krona, and the cost basis is one average across all accounts with schablonmetoden expressly unavailable for crypto even though the general K4 guidance offers it. Finland is the only one where the usual bug runs the OTHER way: the hankintameno-olettama lets a disposal deduct 20% of the selling price instead of the real cost — 40% after ten years — and the filer takes whichever is larger, so an engine that only computes proceeds minus cost OVERSTATES Finnish tax, and an overstated bill is the one nobody queries because it looks like caution. Sweden and Finland are neighbours doing opposite things with the same shortcut, which is why a single European setting cannot be right for both. The Netherlands is the one that changes the question rather than the method: box 3 does not tax a capital gain at all, it taxes a DEEMED return on the value held at 00:00 on 1 January, so selling everything at a profit and losing everything in February produce the same tax — and the counter-evidence rule that lets you show a lower actual return is still not a capital gain, because it counts unrealised value change, allows no costs, cannot go below zero and is compared without the tax-free allowance. A tool that hands a Dutch filer a gain figure has answered a question the return does not ask. Denmark has the widest spread of any of them and is the one most likely to be reported wrong: a gain is anden personlig indkomst in box 20, taxed like salary at up to 53%, while a loss is a ligningsmaessigt fradrag in box 58 worth about 26% — and as a general rule they do not offset each other AT ALL, so 100,000 kr of gains against 100,000 kr of losses is roughly 27,000 kr of tax on a year somebody made nothing, where every netting engine reports zero. There is exactly one narrow exception — one purchase of a single asset, sold in stages in the same year with nothing bought in between — and we test its conditions rather than assume them. Free browser tools at /uk-share-pooling, /canada-superficial-loss, /australia-cgt-discount, /france-portfolio-formula, /sweden-k4, /finland-deemed-cost, /netherlands-box-3 and /denmark-box-20-58 show which rule matched what, and Germany, Canada, Australia, Sweden and Finland are downloadable reports in the language of the office that reads them. Said plainly: that is NINE jurisdictions with real statutory logic, and the UK, France and Netherlands reports are deliberately absent — SA108 wants sterling at each transaction's own date, France needs the portfolio's value at each disposal, and the Belastingdienst asks for the 1 January value «according to the rate on the exchange platform you used» — which is a number that has to come from the holder, not from our price source. We will not invent any of the three. Where Koinly is still ahead, and it is a real gap: it ships the authorities' own FILE FORMATS for around a dozen countries — Sweden's K4 as an SRU upload, a populated France 2086, Finland's Lomake 9, ATO reports — and we ship the statutory logic and a workpaper, not a file the tax office's system ingests. That is a deliberate line for the US (we will not produce a facsimile of an IRS form) and simply unfinished work everywhere else. Count them honestly: nine jurisdictions where we apply the actual statute and name what we refuse to decide, each with a free public tool anyone can check against their own trades, against about a dozen where they generate the filing artefact. The rest of the world is guides, and per-country work we have started rather than finished. | Ten countries with dedicated logic, named openly; a generic English report elsewhere. |
| Documents produced | A filing pack that carries every figure onto the line of the US return it belongs on: Schedule D lines 1b/2/3/8b/9/10 with the 7/15/16 totals, Schedule 1 line 8v for ordinary digital-asset income, Form 8949 itemised AND summarised per box, the Form 8453 transmittal facts, and the statement of transactions. It rules explicitly on which boxes still need detail to reach the IRS — including rows WE adjusted, because a conservative $0 basis is an adjustment and folding it into a summary total would be a compliance shortcut taken in your name; no competitor we checked makes that call in writing. Plus TurboTax CSV, the generic 8949 column set the online products import, and TXF v042 for the desktop ones. What we deliberately do NOT ship is a facsimile of an IRS form — the pack says on its first page that it is a workpaper and that the signature belongs on the real form. Opening with payments. | Country-specific PDF reports for the named jurisdictions, with statutory citations inline; PDF form-completion aids rather than electronic filing. |
| Does your import actually add up? | Every scan reports, free and before you pay, how many DOLLARS of the computed gain exist only because a cost record is missing — and names the asset, the venue and the date of the first sale that hit the gap, so you know which file to go and find. Coins held with no acquisition history are listed separately, because they cost nothing this year and everything in the year you sell. Nothing is ever filled in: a guessed basis would understate your tax and hide the gap that produced it. | Not published. |
| Evidence behind each figure | Every figure links to the row it came from — source file, row number, tx hash, price source — and rulesets are versioned and dated (2026.08-r1) with the methodology memos (M-001…) published rather than internal. The country reports carry BOTH halves on the same line: the statute that decided the figure AND the source row that produced it, in the language of the office that reads it. Germany's report is in German, cites § 23 Abs. 1 Satz 1 Nr. 2 EStG per disposal with the holding period shown, and treats the Freigrenze as the THRESHOLD it is — reach it and the whole gain is taxable, not the excess, which is the error that is always in the taxpayer's favour and therefore never noticed until somebody with authority notices it. The UK is the one place that sentence does not describe a report, and we would rather say so on the same page than let two claims disagree: there is no UK report, deliberately — SA108 wants sterling at each transaction's own date and we will not invent it. What names which of s.104, s.105(1)(b) or s.106A(5) matched each disposal is the free public tool at /uk-share-pooling, against trades you type yourself. A citation tells a reviewer which rule was applied; it says nothing about where the number came from, and we print both. | The closest of the group to an evidence product — statutory citation inline in the report — but still no versioned public ruleset, no change log and no per-figure source trail back to the imported row. |
| Getting your data out | A documented read-only HTTP API over your own scans — transactions, disposals with FULL LOT LINEAGE, decisions, holdings and the completeness report, every figure carrying the source file, row, tx hash and price source behind it, and every response stamped with the dated ruleset that produced it. Read-only is a property of the design, not a scope string: the router registers GET routes only and the token table has no write concept to widen. Plus two free files from any saved scan even if you never pay — your normalised ledger as CSV in the universal template we publish, so it re-imports here and anywhere else implementing the format, and the Basis Passport, a signed evidence-linked basis history whose spec is public and which anyone, including these competitors, may implement. | Report and transaction exports; no open published basis format. |
| Tax-loss harvesting | Free, and public before you have an account: /tax-loss-harvesting runs in the browser, and any saved scan gets the same figures computed off its own lots. Everyone in this table has a version of this — it is the easiest screen in crypto tax to build, because it is a subtraction. What none of them publishes is the thing that decides whether the subtraction means anything: WHICH OF YOUR POSITIONS CANNOT BE MEASURED AT ALL. When an acquisition is missing from your files every engine here, ours included, has to put something in the cost basis box, and the responsible value to put there is ZERO, because a zero basis overstates your gain rather than understating it — the safe direction to be wrong in for tax. It is the ruinous direction for this. Against a basis of zero a position CANNOT show a loss however far it has actually fallen, so a real loss in those holdings does not come out too small: it does not appear at all. That is not a subtlety we are asserting about competitors' products — it is arithmetic that applies to any engine that defaults a missing basis, and we found it by shipping the mistake ourselves. Our first version of this feature carried a card headed «loss resting on a basis nobody can support», which is structurally $0.00 forever and looks exactly like a check that ran and found nothing. It is now sized the only way that is ever true: by what those positions are WORTH TODAY, named asset by asset, so you know how much of your portfolio is unmeasurable and which purchase is worth going to find. We also refuse to value an asset we cannot document rather than dropping it silently — an unpriced position left out of a total is how somebody plans around holdings they have forgotten — and we never use today's unclosed candle as a price. Rates come from you; supply none and we make no tax claim at all. We also now run the cost-basis methods side by side, which is CoinTracking's idea and a good one — and we report the half it leaves out. A method that leaves you more to harvest is usually the method that realised more gain getting there, so ranking methods on harvestable loss alone points at the worse answer about half the time; and a method can look better purely by DEFERRING a gain into a later year, which a single figure shows as a win. On the worked example published at /tax-loss-harvesting, HIFO appears $20,000 better than FIFO and is not: add back the gain left standing and the two are identical to the cent. That equality is the honest headline — a cost-basis method moves money between tax years, it does not make any, and everything you are choosing between is timing. We print the deferred gain and the closing total beside the harvestable loss for exactly that reason, and we rank nothing. AND THE ANSWER IS NOT ONLY AMERICAN. Every harvesting screen in this table computes a United States answer and, at best, warns you that it is one. A warning is not an answer, and we were doing the same thing until we noticed we already held nine statutory engines that could give the real one. The report and the public page now put the SAME realised loss through all nine, and it lands in three honestly different places. It has a fixed value in Sweden and Denmark, where statute settles it with no personal rate involved at all: 70% deductible against income taxed at 30% makes a Swedish loss worth 21% of its face, while a Danish one is a box-58 deduction worth about 26% — and in Denmark it does not meet the box-20 gain you harvested it against, so harvesting there does not cancel anything. It is allowable but un-priceable in the US, UK, Canada and Australia, where the value turns on a marginal rate belonging to a whole return we cannot see — and each carries the rule that actually bites: the UK matches a repurchase within 30 days before the pool, Canada DENIES a superficial loss under s.54 and rolls it into the ACB of what you kept, and Australia applies the loss at full value BEFORE the 50% discount, so which gain you harvest against changes the answer. And it is REFUSED outright in three: the Netherlands taxes a deemed return on the 1 January value so there is nothing to harvest at all; France only has a loss when the whole portfolio is under water, and that loss dies on 31 December; Finland may not have the loss in the first place, because the deemed acquisition cost is a floor under the deduction and can exceed what you actually paid. Germany is the one that surprises people: held more than a year the GAIN would not have been taxable, so the LOSS is not deductible either and harvesting a long-held position achieves precisely nothing. Two disciplines make that table trustworthy rather than decorative. Every rate in it is imported from the jurisdiction module that owns it — a test fails if a statutory figure is re-typed in the harvesting code — and the public page is generated from those same modules at build time, so a table of tax rates cannot quietly go stale in HTML. It ranks nothing and tells nobody to move. The eight country tools each carry their own version of the warning — France, where a loss only exists if the whole portfolio is under water and dies on 31 December; Sweden, where the loss is deductible at 70% against a gain taxed in full; Finland, where the deemed-cost floor can delete the loss you were harvesting for; the Netherlands, where box 3 makes the exercise meaningless. Retrieved and written 31 August 2026. | «Tax Optimization» appears as a line item on Starter, Professional and Business — paid tiers, with an asterisk on the pricing page reading «Advanced features are exclusively available for the current tax year». What the feature actually shows is not described on the page we can cite, so we do not characterise it. Retrieved 31 August 2026. |
| Your ledger vs your Form 1099-DA | Upload your Form 1099-DA beside your exchange history and every form line is compared to your own ledger, free. Each line comes back with one of five verdicts and a different action behind it: it AGREES; the broker reported NO basis, so the form tells the IRS your whole proceeds were gain and you fix it on Form 8949; the basis DIFFERS and we show both numbers without claiming ours is right; the PROCEEDS differ, which is the one worth raising with the broker before you file because proceeds are what they actually paid you; or the sale is NOT IN YOUR HISTORY AT ALL — a sale the IRS has been told about and your return does not mention, which is the difference that gets a letter, and almost always a missing export rather than an error on the form. The three headline figures tie out arithmetically: what the form claims above what you owe equals the per-line differences, plus form lines we could not find, minus your own sales that are on no form. Those adjustments are printed rather than absorbed, because a summary figure that cannot be rebuilt from the rows under it is one nobody can check — and a test asserts the identity against the engine's own figure. With a real form in hand the headline stops being the blank-basis worst case and becomes a MEASUREMENT, labelled as such. What this will not do: decide that we are right, merge the broker's basis into your ledger, or pair a form line with a sale it is not confident about — an unmatched line is reported as unmatched, because pairing the wrong sale sends somebody looking for a problem that is not there. This also exists because building it exposed a live bug of ours: uploading the form beside real history used to put every one of those sales in the ledger TWICE. Measured on a one-buy one-sell ledger, $1,480 of gain became $2,980. Nothing failed — the scan completed, the documents generated, the audit was clean — it simply overstated the tax, which is the direction nobody queries. The form is now dropped from the ledger the moment any real history is present, the user is told in writing that it was, and its rows become this reconciliation instead. | Nothing published that we can cite about reconciling an uploaded Form 1099-DA against your own ledger. Blockpit's centre of gravity is European reporting, where the form does not apply, so treat this as out of scope for them rather than as a gap. Read 1 September 2026. |
| Dead platforms | Dead-platform rebuild is the specialty: FTX, Celsius, Cryptopia, Mt. Gox and others, reconstructed from what survives, with the gaps shown rather than filled in. | Not a stated specialty. |
| Support model | Self-serve; no sales calls. Practitioner tier (Workbench) is self-serve too. Volume is measured rather than claimed, and the measurement is one you can repeat: tools/make_scale_fixture.py --rows 120000 generates the same 134,648-transaction ledger every time from a fixed seed — three sources, with cross-file self-transfers, blank prices, crypto fees, staking, dust, scam airdrops with attacker-chosen symbols, and a deliberately missing acquisition — and tools/scale_run.py puts it through the real multi-source pipeline. It computes end to end in about six seconds on two 2.1 GHz cores (5.9–6.4s across runs; faster hardware, faster). The seconds depend on the machine and the count does not, which is why the count is the part stated exactly. What matters more than either is that the harness checks the ANSWER as well as the clock — every disposal fully allocated to lots, no negative open lot, gain equal to proceeds minus basis on all 23,194 8949 rows, the totals equal to the sum of the rows to the cent, and the planted missing history reported rather than absorbed. A run that is fast and wrong makes the timing number worthless. Our top band is 50,000. | Email/chat; Source-of-Funds reports sold per credit (19.99€). |
Blockpit figures from their own pricing page, checked 2026-08-29.
Which one should you actually use?
Choose Blockpit if you need dozens of exchanges connected by API today, or a statutory form for a country we have not built yet. We would rather tell you that than take your money for a worse fit.
Choose ClearBasis if your problem is missing cost basis — transferred-in coins, a dead exchange, a decade of history, or a 1099-DA that reads your whole sale as profit — and you need every figure to survive a question from a tax authority, with the evidence attached.
Use both if it helps. Their export imports here directly, and our Basis Passport is an open format anyone can read — including them. Switching costs are a business model, not a law of nature.
Other comparisons
ClearBasis vs Koinly · ClearBasis vs CoinTracker · ClearBasis vs CoinLedger · ClearBasis vs CoinTracking
compare-2026.1 · figures checked 2026-08-29 · Blockpit are trademarks of their respective owners; ClearBasis is not affiliated with, endorsed by or sponsored by any of them. Competitor figures are what each vendor published on its own site on the date shown and change without notice — follow the links and check before deciding. Where a vendor does not publish a figure in machine-readable form we say so rather than estimate it. ClearBasis is tax software, not a law firm, accounting firm or tax preparer.