Hyperliquid Weekly Data Report — Week of September 28–October 4, 2026
Every metric below was fetched from a public API — the Hyperliquid official endpoint, DefiLlama, published venue tallies — on October 5, 2026, and the fetch is archived with a timestamp. This report prints only what a public source can verify: where a commonly cited metric has no public API, the table says so rather than estimating, and where a widely quoted figure uses a different construction, both are printed and named. The week in one line: HYPE absorbed the cycle's largest unlock (14.18M HYPE, ~$1.2B, Sep 29) and closed $90.57, −1.2%; platform open interest stood at $16.69B; and protocol fees for the full seven days totalled $17.92M, −22.5% against the prior week's provider-revised $23.13M.
The verified metrics
| # | Metric | Reading | Source | Note |
|---|---|---|---|---|
| 1 | HYPE price (Oct 4 close) | $90.57 | Hyperliquid candles | −1.2% WoW ($91.72 → $90.57) |
| 2 | Weekly high / low | $92.09 / $84.55 | Hyperliquid candles | High printed Mon Sep 28, low Wed Sep 30 |
| 3 | All-time high (intraday) | $98.00 | Hyperliquid candles | Unchanged (Sep 23); the close sits −7.6% below it |
| 4 | Market cap (at close) | $20.15B | Derived (222.446M × $90.57) | 222.446M circulating × Sunday close |
| 5 | Circulating supply | 222.4M | CoinGecko / venue tallies | Of 955.3M total; unchanged despite the Sep 29 unlock — see ledger |
| 6 | Token volume (24h) | $0.89B | CoinGecko | Oct 5 print; HYPE spot + perp token volume |
| 7 | Perp OI — platform total | $16.69B | Hyperliquid API | Core $12.80B + builder markets $3.90B |
| 8 | Perp OI — core order book | $12.80B | Hyperliquid API | 178 live markets of 234 listed; top three books sum $8.13B |
| 9 | Perp OI — HYPE book | $1.76B | Hyperliquid API | 19.84M HYPE × mark $88.53; −5.5% WoW |
| 10 | Perp OI — HIP-3 builders | $3.90B | Hyperliquid API | 23.4% of platform OI (was 23.1%) — first rise in three weeks |
| 11 | Live perp markets | 178 + 158 | Hyperliquid API | Core + builder markets with seeded books |
| 12 | Perp volume (24h) | $9.28B | Hyperliquid API | Core $6.83B + builders $2.45B |
| 13 | HYPE funding (8h) | +0.010% | Hyperliquid API | Rebuilt from ~0.000% — longs pay again, modestly |
| 14 | Protocol fees (7d) | $17.92M | DefiLlama | Sep 28–Oct 4, seven prints; −22.5% WoW vs revised $23.13M — see ledger |
| 15 | Protocol fees (30d) | $96.5M | DefiLlama | Seven-day run-rate annualizes to ~$0.93B |
| 16 | Implied buyback capacity (7d) | ~$17.7M | Derived | ~99% pass-through (third-party estimate) × fees; AQAv2 adds a measured $14.58M first payment (Oct 3) |
| 17 | TVL — Hyperliquid L1 | $6.85B | DefiLlama | −2.1% from $6.99B |
| 18 | TVL — Arbitrum bridge | $0.39B | DefiLlama | −30% from $0.56B — see ledger |
| 19 | Total TVL (protocol) | $7.24B | DefiLlama | L1 + bridge; −4.0% WoW |
| 20 | HYPE burned (cumulative) | ~48.7M+ | Press tally | ~4.9% of max supply; ~30.5K HYPE burned in the 24h around the unlock (HypedLaunches) |
| 21 | 2026 YTD perp fees | ~$493M | Hyperdash (press) | Jan 1–Sep 30; ~2.46 bps per dollar of ~$2T YTD perp volume |
| 22 | Next scheduled unlock | Oct 29 | Published schedules | ~2.6% of market cap per the published schedule — smaller than Sep 29's 2.7% |
Not verifiable via public API (printed as unavailable, not estimated): staked HYPE and staking APY, HLP vault TVL and return, Assistance Fund weekly USD purchases and cumulative ledger, weekly liquidations, 7-day perp volume, spot 7-day volume, active addresses, transaction count, HIP-3 weekly volume share as a single published series, and the exact token count of the Oct 29 unlock (the schedule publishes its market-cap share, not a headcount, until it lands). Rows 20 and 21 are third-party tallies, labeled as such: no public endpoint publishes the burn ledger, and Hyperdash's year-to-date fee series is a press figure, not interchangeable with row 15's API window.
The five-week chain
| Week | Close | High | Low | Fees (7d) | TVL L1 (close) |
|---|---|---|---|---|---|
| Aug 31–Sep 6 | $87.94 | $89.60 | $79.52 | $20.07M | $6.41B |
| Sep 7–13 | $77.26 | $88.31 | $76.59 | $18.99M | $6.33B |
| Sep 14–20 | $93.75 | $94.60 | $75.15 | $23.79M | $6.50B |
| Sep 21–27 | $91.72 | $98.00 | $89.77 | $23.13M | $6.99B |
| Sep 28–Oct 4 | $90.57 | $92.09 | $84.55 | $17.92M | $6.85B |
Fee cells print the provider's latest revised series on every week; the as-published value of each week remains visible in that week's own report (Sep 21–27: $23.13M revised from $19.46M; Sep 14–20: $23.79M from $19.34M; Sep 7–13: $18.99M from $12.9M; Aug 31–Sep 6: $20.07M from $16.5M) — see ledger.
The shape: a two-week round trip through the $75–76 lows, a +21.3% reclaim into a record $98.00, a place-holding week, and now an absorption week — a lower high ($92.09 vs $98.00), a higher low than any week since the rally began ($84.55), and a close inside the prior week's range. The five-week fee total is ~$103.9M across the window on the revised series, and the five-week price range is $75.15–$98.00. Note what did not happen: the $89.77 low of the record week was never respected as support — the market traded through it to $84.55 — and yet the week closed above $90. Both the bears' level and the bulls' level failed; the range held.
Inside the week: the daily path
| Date | Close | Δ | Protocol fees (day) |
|---|---|---|---|
| Mon Sep 28 | $87.47 | −4.6% | $3.09M |
| Tue Sep 29 | $86.03 | −1.6% | $2.98M |
| Wed Sep 30 | $90.94 | +5.7% | $2.19M |
| Thu Oct 1 | $87.63 | −3.6% | $2.43M |
| Fri Oct 2 | $89.06 | +1.6% | $4.06M |
| Sat Oct 3 | $89.28 | +0.2% | $1.40M |
| Sun Oct 4 | $90.57 | +1.4% | $1.76M |
Three structural notes on the daily pattern. First, the week's selling was front-loaded: both down legs before Wednesday came before or during the unlock, and every session after the $84.55 probe closed higher than the last except Thursday's give-back. Second, the largest fee print ($4.06M, Friday) landed on a +1.6% day — the fee line again tracked repositioning rather than price size, consistent with the multichain deployment landing the same day. Third, the week's two gap-like moves (Wednesday +5.7%, Thursday −3.6%) bracket the unlock resolution: the market rejected $84–85, probed whether the rejection was real, and spent the weekend firming the answer.
The same seven days for BTC and ETH
| Date | HYPE Δ | BTC close | BTC Δ | ETH close | ETH Δ |
|---|---|---|---|---|---|
| Sep 28 | −4.6% | $83,457 | −1.2% | $2,689 | +0.0% |
| Sep 29 | −1.6% | $83,620 | +0.2% | $2,678 | −0.4% |
| Sep 30 | +5.7% | $83,607 | −0.0% | $2,687 | +0.3% |
| Oct 1 | −3.6% | $84,837 | +1.5% | $2,707 | +0.7% |
| Oct 2 | +1.6% | $84,482 | −0.4% | $2,669 | −1.4% |
| Oct 3 | +0.2% | $84,716 | +0.3% | $2,687 | +0.7% |
| Oct 4 | +1.4% | $86,504 | +2.1% | $2,727 | +1.5% |
Weekly totals on the Sunday-close clock: HYPE −1.2%, BTC +2.4% ($84,437 → $86,504), ETH +1.4% ($2,688 → $2,727). The cross reads cleaner than last week's: HYPE's two worst sessions (Sep 28, Oct 1) were idiosyncratic — BTC fell 1.2% on the first and rose 1.5% on the second — and the token's recovery (Sep 30, +5.7%) came on a flat majors tape. This was HYPE trading its own event, not the market's. The unresolved part is the close: the week ended with BTC at its best (+2.1% Sunday) and HYPE merely following, which keeps last week's underperformance question open for another week.
The HIP-3 builder census, measured
Builder-deployed markets are measured directly from the API's perpDexs and metaAndAssetCtxs endpoints, summed across every registered dex. This week's census, run Oct 5:
| Builder dex | Live markets | Open interest | 24h volume |
|---|---|---|---|
| xyz | 114 | $3.81B | $2.42B |
| io | 9 | $0.06B | $0.02B |
| para | 30 | $0.02B | $0.00B |
| mkts | 5 | $0.01B | $0.00B |
| Total (4 of 10 registered) | 158 | $3.90B | $2.45B |
The largest builder books this week: SP500 ($373.4M), GOLD ($312.6M), SKHX ($305.5M), XYZ100 ($241.0M), BRENTOIL ($192.3M), SILVER ($159.4M), SKHY ($158.6M), SNDK ($155.3M), NVDA ($153.7M) and CL ($136.0M) — the composition continues to rotate within the non-crypto stack, with GOLD overtaking SKHX for second place and new entrants (SNDK, NVDA, CL) displacing names from last week's list. Builder markets held 23.4% of platform open interest ($3.90B of $16.69B), up from 23.1% — the first rise in that share after two consecutive declines. Four of the ten registered builders have live markets; the other six show zero open interest and zero volume. On the total-OI question, see the ledger below — the platform figure here sums every live market, and widely published figures measure a different, narrower set.
Valuation arithmetic on verified inputs
From verified inputs only. Market cap at the weekly close ($90.57 × 222.446M circulating) is $20.15B. The seven-day fee total is $17.92M; annualised at 52 weeks that is a $0.93B run-rate, and the 30-day base ($96.5M × 365 ÷ 30) annualises to $1.17B. On fully diluted supply of 955.3M the same market value is $86.5B.
| Inputs | Value | Multiple on circulating | Multiple on FDV |
|---|---|---|---|
| Market cap / 7-day fee run-rate ($0.93B) | $20.15B | ~21.6x | ~93x |
| Market cap / 30d fee base ($1.17B) | $20.15B | ~17.2x | ~74x |
| Market cap / W6 revised run-rate ($1.20B) | $20.15B | ~16.7x | ~72x |
| Fully diluted value / 7-day fee run-rate ($0.93B) | $86.5B | ~93x | n/a |
Read the first and third rows together, because they bracket the honest answer better than any single number. On this week's decayed fee pace the multiple is ~21.6x; on the prior week's provider-revised pace it is ~16.7x. The 5-point spread across adjacent weeks is not the market de-rating and re-rating — it is a fee series with 22.5% weekly variance passing through a near-total pass-through token model. Any single multiple quoted without its fee window is unfalsifiable; this table exists to make that point rather than to make a forecast. The FDV rows assume all 955.3M outstanding tokens trade at the same price, which is not how unlocks work.
Data quality ledger
Every open item in this report, stated plainly rather than smoothed over:
- Prior-week fees — revised upward, again. DefiLlama now prints Sep 21–27 at $23.13M, against the $19.46M published when our W6 report went out — an 18.9% upward revision, distributed across every daily print (e.g. Sep 21: $4.65M → $5.14M; Sep 27: $1.75M → $2.29M). This is the second consecutive week the provider has revised a week upward after publication (W5: $19.25M → $19.34M). All week-over-week comparisons in this report use the revised $23.13M. The pattern matters for reading fee trends in real time: recent weeks' prints should be treated as provisional, and this site will keep publishing the as-fetched figure with the revision disclosed rather than silently updating. The same re-read shows the revisions reach further back than this report's first ledger entry acknowledged: Aug 31–Sep 6 now prints $20.07M (published $16.5M), Sep 7–13 $18.99M ($12.9M), Sep 14–20 $23.79M ($19.34M), and August as a whole $77.8M ($67.3M as published in the August report). The five-week chain above uses the revised series; the monthly report's week-over-week comparison does the same.
- Open interest — two constructions, both printed. This report's $16.69B sums every live market on the platform ($12.80B core + $3.90B builder) from the API. Figures quoted elsewhere that track only the largest books run several billion lower; the core book's top three (BTC $3.30B, ETH $3.07B, HYPE $1.76B) already sum to $8.13B. Neither is an error; they are different measures.
- Circulating supply — unchanged through a 14.18M unlock. CoinGecko still prints 222.4M circulating, the same figure as before Sep 29, while the vesting schedule released 14.18M tokens on that date. The provider's circulating series has not absorbed the unlock; unlocked tokens are not automatically counted as circulating. Some exchange tickers list 251.2M and quote correspondingly higher market caps. Row 4's $20.15B uses the 222.4M basis; read any market-cap rank against its supply definition.
- All-time high — unchanged. The record remains $98.00 (Sep 23) on the Hyperliquid candle series; no venue printed higher this week, and the week's high of $92.09 stayed 6.0% below it.
- Fetch timing. Price rows are in-window daily closes (Sep 28–Oct 4). Open interest, funding, TVL, supply and the 24h volume rows are fetch-time snapshots taken Oct 5, 2026, one day after the Sunday close — the OI and funding series are not stored historically by a public endpoint, so each week's row reflects the measurement date printed beside it.
- Bridge TVL — down 30%. The Arbitrum bridge balance fell from $0.56B to $0.39B, the largest weekly move in that line this site has recorded. The API prints the levels without a cause; we print the same. It is flagged here rather than buried in row 18.
Frequently asked questions
Why do the fees keep getting revised after publication?
The fee series on DefiLlama is assembled from on-chain accounting that continues to settle after a week closes, and the provider revises its chart retroactively — twice in a row now, always upward (W5 +0.5%, W6 +18.9%). Our policy is to publish what the API says at fetch time, label it, and disclose the revision when the next week's report re-reads the window. The alternative — waiting for the series to stabilise — would delay every report by weeks for a data point that is directionally stable even when its level is not.
Is a −22.5% fee drop a warning?
On its own, no — it is the arithmetic of comparing against a revised record week. The 30-day total ($96.5M) is the second-highest window the platform has printed, and the daily prints inside this week ($1.40M–$4.06M) sit in the normal post-spike band. What would upgrade the concern: a further decay week stacking on this one, taking the 7-day run-rate materially below the ~$0.93B this report annualises.
Why did the circulating supply not move when 14.18M tokens unlocked?
Because "unlocked" and "circulating" are different ledgers. The vesting schedule released eligibility on Sep 29; the supply aggregators that publish 222.4M have not reclassified those tokens into their circulating series (and may apply their own criteria for when unlocked tokens count). The discrepancy is visible in this report: row 5 is unchanged while row 22's predecessor event completed. We print both facts rather than reconciling them by assumption.
Why does this site never print a single "Hyperliquid open interest" without qualification?
Because the same three words describe different sums that differ by billions, and the difference is entirely a definition. We print our construction (every live market, summed), name the top books so readers can sanity-check, and let the narrow constructions be compared on their own terms. A single unqualified number is the most common way to be both accurate and misleading.
Is this financial advice?
No. This is data analysis for informational purposes. Nothing here is a recommendation to buy, sell or hold HYPE or any other asset.