for ChatGPT, Claude, Perplexity…
CASE STUDY

A real Hyperliquid tax reconstruction, in numbers

· 7 min read · Not tax advice

Most "crypto tax" content is hypothetical. This one isn't: we pulled the public API history of a widely tracked Hyperliquid whale address — 0x5078c2fbea2b2ad61bc840bc023e35fce56bedb6 — on October 8, 2026, and reconstructed what its tax file actually has to contain. Every number below comes from Hyperliquid's own userFillsByTime, userFunding, and userNonFundingLedgerUpdates endpoints — anyone can reproduce them.

Key facts
  • One month, fully retrieved: 45,147 fills between March 13–31, 2025 — about 2,500 fills per day. 78% of them were a single market (kPEPE).
  • Exchange-reported realized PnL for that month: −$335,379 (a net loss). Trading fees: $104,168.
  • Funding is a separate firehose: $8,012,481 paid vs $251,756 received across 652 events (March 2025 → September 2026) — mostly carrying a large BTC long.
  • Ledger: 730 non-funding movements, including a single $946,927 USDC deposit.
  • Why it matters: a 10,000-transaction import cap silently truncates this wallet inside its first month — and the biggest line item (−$8M funding) never appears in trade fills at all.

March 2025, completely

Rather than trust a single API call (which caps at 2,000 fills), we paginated userFillsByTime forward until it returned empty — a bounded, complete month:

Metric (March 2025, UTC)Value
Fills45,147
Realized PnL (exchange-reported closedPnl)−$335,379
Fees paid$104,168
Top markets by fillskPEPE 35,079 · ETH 5,270 · BTC 2,706 · HYPE 1,197
Liquidation-flagged fills0

A fills-only reading says "lost $335k, paid $104k in fees." True — and wildly incomplete. The funding endpoint tells the real scale story:

Funding stream (retrievable window)Amount
Paid$8,012,481
Received$251,756
Net−$7,760,725
Largest by marketBTC −$5.82M · kPEPE −$1.33M · FARTCOIN −$421k · XRP −$170k · TRUMP +$110k
The failure mode in one sentence: an importer that only reads trade fills produces a file that is off by $8 million for this wallet — silently, with no error message.

Three things this case demonstrates

1. Funding is its own data stream — and it dwarfs everything else

Perp funding accrues hourly, out-of-band from fills, in userFunding's nested delta.usdc field. For this wallet it exceeds realized PnL by more than an order of magnitude. Whether funding paid is deductible and how is a classification decision (see funding tax treatment) — but a report missing −$8M of activity isn't a classification problem, it's missing data.

2. Limits hit high-volume wallets first, hardest, and silently

This wallet produced 45k fills in a single month — 4.5× the 10,000-transaction cap commonly cited for tax-tool Hyperliquid imports — and its history doesn't stop there: fills continue through September 2026. Nothing about an import says "truncated"; the totals just come out wrong. Even the API's own limits are easy to misread — a single call caps at 2,000 fills, and requests before the window return empty, not an error. (measured endpoint behavior · full export options.)

3. Exchange-reported fields are the ground truth

Every figure here is Hyperliquid's own output — closedPnl per closing fill, fee per fill, usdc per funding delta. Nothing was re-derived from prices. That's the property that lets a filer reconcile against the exchange's view, and the property importers lose when they re-derive (why re-derivation breaks).

Honesty notes

Run the same reconstruction on your address

The free scanner pulls fills, funding, and ledger movements, then totals them per tax year — read-only, no signup.

Scan my wallet

Common questions

How much trading history can one Hyperliquid wallet have?

We measured 45,147 fills for a single whale address (0x5078…edb6) in March 2025 alone — about 2,500 fills per day, 78% in one market (kPEPE). A 10,000-transaction import cap truncates this wallet inside its first month, silently.

What did the whale's actual numbers look like?

For March 2025: −$335,379 exchange-reported realized PnL and $104,168 in fees. But the funding stream dwarfs it — $8,012,481 paid vs $251,756 received across 652 events, mostly carrying a large BTC long. A fills-only import is off by $8 million for this wallet.

Can these numbers be verified independently?

Yes — every figure comes from Hyperliquid's public endpoints (userFillsByTime, userFunding, userNonFundingLedgerUpdates) for a publicly tracked address. The fills were paginated forward to exhaustion; funding figures cover all 652 events the endpoint exposed.

How this page was produced