CoinLedger + Hyperliquid: the smooth sync that hides a truncated record
updated november 2026 · ~5 min · basisproof research · not tax advice
Polish is the risk factor
CoinLedger's pitch is frictionless: connect, sync, file. That smoothness is genuinely useful — and it's also what makes a silent gap dangerous. A truncated import doesn't look broken; it looks finished. You get a green dashboard, a Form 8949, and no hint that the record underneath is a fraction of your year.
The caveat that keeps this honest: integrations get updated all the time, and CoinLedger may pull Hyperliquid data through multiple paths. Verify what YOUR sync produced — don't assume this page describes it.
Why a perps record truncates quietly
The venue's fill endpoint retains roughly the last 30 hours of fills — and still returns a successful response. An importer reads it, gets a complete-looking list, and shows no error. On an account with tens of thousands of lifetime fills, a sync can legitimately import 1–2% of the record and render it as 'done'.
Funding is a separate stream entirely. Perp funding settles hourly on its own endpoint — an importer that only reads fills reports a $0 funding line. On an account that held positions, zero funding is a dropped-data flag, not a fact.
And the record isn't just fills: builder-dex markets (xyz:, flx: and other HIP-3 venues) live outside the main perp book; sub-account and spot↔perp transfers can import as ordinary deposits and withdrawals, double-counting moves that never left your own wallets.
The three-minute verification
Count check. Does the imported transaction count look right against how much you actually traded? Hyperliquid scalpers run thousands of fills a month — a few hundred imported rows is a truncation tell.
Funding check. Any position held across funding boundaries produces funding events. A $0 funding line on a perps account means the stream was never read.
Symbol check. Search for builder-dex markets — xyz:, flx:, or whatever you actually traded. If none appear and you traded them, the importer enumerated only the main book.
Fix the data, then the filing
No sync settings fix a retention limit — a second run reads the same endpoint and gets the same 30 hours. The fix is upstream: fills from a chain indexer (full history), funding and ledger from the venue's own full-history endpoints, verified against an independent account-value identity.
Our free check recomputes your record fill by fill and prints the totals — run it against the same address and compare. Matching numbers are evidence your record is whole. A gap tells you exactly which stream was dropped, and the paid report produces the corrected CSVs and per-disposal basis records your preparer or a manual CoinLedger import can use.
Common questions
?Does CoinLedger support Hyperliquid?
CoinLedger supports a wide range of exchanges and wallets — but 'supported' describes the sync working, not the record being complete. The three checks above tell you which one you got.
?My import looks complete — is it?
That's exactly the failure mode: truncated imports render as finished imports. Compare the transaction count and funding line against your real activity before you trust the totals.
?Can I fix it without leaving CoinLedger?
CoinLedger accepts manual CSV imports — a complete recomputed export can replace the sync's partial record. The data has to come from a source that serves full history first.
Sources & method
See what the venue API isn't showing you — free
The fill endpoint forgets everything older than ~30 hours. The free gap check measures what fraction of your record it actually serves — seconds, not minutes.
Measure my gap