Hyperliquid Weekly News Review — Week of September 7–13, 2026

This review covers the Hyperliquid ecosystem and the HYPE token for the week of September 7–13, 2026, in sixteen stories. Every number below is pulled from a named primary source — the Hyperliquid API, DefiLlama, CoinPaprika — and each entry names where it can be re-verified. The week in one line: HYPE gave back 12.1% to close at $77.26 after Sunday's $89.60 intraday all-time high, while the protocol's fee run-rate slowed for a fourth straight week and the derivatives book barely moved.

1. The week in sixteen stories

Sixteen items, each with the verified number first and an editorial take second. The takes are interpretation only — no take introduces a new figure; every fact above it is re-verifiable at the named source.

1. HYPE closes the week at $77.26, −12.1% Price

HYPE opened the week at $87.96 and closed Sunday Sep 13 at $77.26 — a −12.1% week on the Hyperliquid daily candles, measured close-over-close like every weekly number on this site. The four-week closing series frames it: $82.37, $79.99, $87.94, $77.26. The path was concentrated: three quiet sessions ($85.18, $85.12, $83.52), one −6.0% break on Thursday, and a weekend that drifted to the $76.59 weekly low.

Our take: a −12% week one week after a +9.9% record week is a round trip, but the closing level still sits roughly 35% above the $57.10 close four weeks ago. Weekly closes are the cleanest signal on a noisy tape, and this one says the market paid up for the record and then partially unwound. Whether that unwind touched anything structural is what the next twelve entries answer — and on the evidence, stories 5 through 13 say it did not.

2. Withdrawal from the record Price

From the $89.60 intraday all-time high of Sep 6 — a print the prior week's review covers in full — the weekly close sits −13.8% lower. Measured from the $87.94 close of that same Sunday instead, the give-back is −12.1%. Both clocks are printed here because both circulate: the intraday anchor produces the bigger number, and the close-over-close anchor is the convention this site uses everywhere else.

Our take: intraday highs are the most flattering anchor in an uptrend and the harshest one in a drawdown, which is exactly why headlines like them. Measured from the last close instead — the convention we use everywhere on this site — the give-back is −12.1%. Both numbers are true. The bigger one will circulate more widely, and knowing which clock produced it is most of the work of reading crypto price coverage.

3. Thursday was the damage day Price

Sep 10 closed $78.52, −6.0% in a single session from $83.52, with a $78.17 intraday low — the week's only large daily move. The week's low of $76.59 printed three days later, on Sunday Sep 13, on a thin weekend book. The sequencing matters: the damage came first, the low came after, and the week's highest fee print ($2.81M) landed on the damage day itself.

Our take: one session carrying most of the week's damage, with the low arriving on a weekend book, is the signature of a positioning unwind rather than a shock event — no headline broke into the tape on Sep 10. And a Saturday low deserves less weight than a weekday print: weekend order books are thin, and lows set there tend to overstate where real clearing happened. The daily decomposition at the bottom of this page makes the same point hour by hour.

4. Local excess, not market beta Cross-asset

The same week, on the same daily candles: BTC fell −4.4% (to $76,810) and ETH fell −1.6% (to $2,476). The daily detail sharpens the contrast: BTC's worst session was Sep 10 at −2.2%, ETH's was the Sunday close at −1.9% — HYPE's −6.0% on Sep 10 was nearly triple BTC's worst day. Over the full week the gap is 2.7x BTC's decline and 7.6x ETH's.

Our take: this is the most important story on the list. A triple-beta week, once, is noise. But combined with what did not fall — open interest (story 8), funding (story 10), TVL (story 11) — the reading is a de-risking of a crowded winner, not a crypto crash. The corollary deserves equal attention: the +44.3% launch week three weeks ago built exactly the crowding that unwinds like this. Outperformance and its unwind are one mechanism, not two.

5. Protocol fees fell a fourth straight week Fees

Trailing-7-day protocol fees printed $12.9M — the fourth consecutive decline, from $24.6M to $21.2M to $16.5M to $12.9M (DefiLlama). The daily series inside the week: $1.83M, $2.35M, $2.32M, $2.81M, $1.44M, $0.73M, $1.36M — summing to $12.8M, consistent with the weekly print at rounding. No single day explains the decline; the baseline itself moved down.

Our take: the fee series, not the price series, is what repriced this week. A 48% drawdown in the gross revenue run-rate inside one month is a fundamental event, and price simply caught up to it. Anyone holding HYPE holds a claim on this series — directly, through the buyback mechanics in stories 12 and 13 — which is why we print the fee ledger in every review even when nothing else moves.

6. Saturday was the fee trough Fees

Sep 12 printed $0.73M in daily protocol fees — the weakest single day of the four-week window, against a weekday baseline near $2M (DefiLlama). It was a Saturday session on a weekend book, and the week's Monday had printed $1.83M — weak, but two and a half times the trough. The weekly aggregation (story 5) is the honest read; single prints at the extremes are calendar effects as often as signal.

Our take: single-day fee prints overstate both booms and busts. $0.73M on a Saturday against a weekday baseline near $2M is mostly calendar — weekend books are thinner everywhere — not evidence of collapse. The weekly series in story 5 is the honest aggregation; this story is a caution against extrapolating any one print in either direction, including the bullish ones.

7. 30-day fee base: $76.1M Fees

The trailing 30-day fee total stands at $76.1M, which puts the annualized run-rate on the most recent seven days at roughly $0.67B/yr (DefiLlama). The spread across the window is the context: $6.20M was the best single day of the month (Aug 21) and $0.73M the worst (Sep 12) — an 8.5x range inside 22 days. The $76.1M base, not either extreme, is the number valuation arithmetic sits on.

Our take: this is the number that every valuation argument sits on, whether or not the argument states it. A month ago the same construction annualized closer to $1.28B; the base moved, so every multiple built on top of it moved with it. Any HYPE valuation claim that does not disclose its fee base is not an argument — it is an aesthetic preference. We keep this story in every review for exactly that reason.

8. Open interest barely moved Derivatives

Total perp open interest measured $10.32B on the core DEX as of Sep 14, with Hyperliquid remaining roughly the third-largest perp venue (Galaxy Research). The full platform stock — including the $4.12B of HIP-3 builder markets (story 14) — measured $14.44B in the same census, so the week's −12% price move ran against a derivatives book that measured essentially intact.

Our take: a −12% price week against a flat derivatives book is the strongest single argument against reading the week as a confidence event. Liquidation cascades shrink the stock of positions; closing flows merely slow the flow through it. This story, plus story 10, is why we keep calling the drawdown local de-risking — the committed capital did not flinch.

9. HYPE-perp OI $1.67B Derivatives

Open interest in the HYPE book itself stood at $1.67B across 178 listed perp markets; platform-wide 24-hour perp volume printed $4.39B (Hyperliquid API). Both are prints from the same census: the flow fell with the week's activity while the stock held — the combination stories 8 and 10 read together, and the reason this review keeps them adjacent.

Our take: volume fell harder than open interest, and that ratio — flow over stock — is the activity gauge worth watching. When volume compresses faster than OI, positions are being held rather than traded; they resolve either in a trend or in chop. Near-zero funding (story 10) says the market is not paying much to force a resolution either way. Stalemates like this one usually end in volatility; the direction is the unknown.

10. Funding stayed benign Derivatives

HYPE perp funding printed 0.0000125%/hour — roughly 0.010% per 8-hour interval, effectively neutral (Hyperliquid API). At that rate, a leveraged long pays on the order of 0.03% per day to hold — noise against a week in which the underlying moved 12%. The print is the absence of leverage stress: crowded books show up in funding before they show up in price, and this one never did.

Our take: near-zero funding during a −12% week is the absence of a forced unwind. Crowded longs pay funding; nobody was paying. Read together, stories 8, 9 and 10 form the cleanest evidence in this review: the price fell, the leverage did not get punished, and the positions did not get liquidated — traders chose to leave, they were not expelled.

11. TVL range-bound, not exiting Deposits

Hyperliquid L1 TVL held a $6.28–6.44B band all week and closed Sunday at $6.33B (DefiLlama). The band brackets the closes of the prior three weeks — $6.16B, $6.43B, $6.41B — so a −12% token week left deposits inside the range they had already established. Price left the range; deposits did not.

Our take: deposits are the least sentimental capital on the platform, and they stayed put through a −12% token week. Keeping price and platform separate is the discipline this story enforces: one repriced, the other did not move. If TVL ever confirms a fee decline — both falling together — that is the sequence that would upgrade our language from "local de-risking" to something more structural.

12. The buyback engine's new fuel Tokenomics

Under the USDC aligned-quote-asset arrangement, Circle and Coinbase share roughly 90% of the reserve yield on USDC collateral back to the protocol — an estimated ~$160M/yr of additional HYPE buyback-burn capacity (Galaxy Research; estimate, not a published ledger). The underlying base was sized in the prior week's review at ~$6.21B of USDC in the AQA v2 wallet, with the first payment scheduled for Oct 3 — the date that turns the estimate into a measured print.

Our take: this line matters more in a falling-fee week than it did in a record one. Trading fees are pro-cyclical — stories 5 through 7 just demonstrated it — while a reserve-yield stream is comparatively steady. A revenue mix with a counter-cyclical component is structurally better than its worst week. The caveat stands: this is an estimate until the first payment lands, and it is labeled as one.

13. AF mechanism: ~99% of revenue Tokenomics

Virtually all protocol revenue routes to the Assistance Fund for HYPE buyback-burn; HLP now receives only ~1% of fees (third-party on-chain estimates). The cumulative ledger, as printed in this site's earlier reviews: ~$1.1B spent, 45M+ HYPE repurchased, 48.45M burned (4.84% of the max supply), $379M repurchased in 2026 year-to-date. The pass-through is why this week's fee decline (story 5) transmits to the buyback ledger at close to one-for-one.

Our take: near-total pass-through is unusual in this industry, where treasuries and middlemen routinely sit between revenue and holders. It aligns the token with gross revenue almost one-for-one — which is precisely why the fee entries dominate this week's story. It also concentrates the risk: there is no buffer between a weak fee print and the buyback ledger. The alignment cuts both ways, and this week showed the downward edge.

14. HIP-3 keeps scaling non-crypto markets Structure

$4.12B of the $14.44B platform open interest — 28.5% — sits in builder-deployed markets covering the S&P 500, single-name equities, commodities and indices, measured directly from the API (full census in our HIP-3 deep dive). The census detail behind the aggregate: 10 active builder DEXs, with xyz running 104 of the markets and ~98% of builder OI — top books include SP500 at $421.8M, the SK Hynix single-name at $380.6M and GOLD at $337.4M — and HyENA at zero active markets, the first full deployer exit.

Our take: 28.5% is not a crypto metric; it is a market-structure metric. A quarter-plus of the book's open interest now depends on non-crypto underlying, which means the venue's growth story is partially independent of the token's price story — and this week proved it: the diversification argument survived a −12% HYPE week untouched. The next structural question is concentration, which the deep dive measures at roughly 98% of builder OI in one deployer.

15. Supply picture: $17.7B market cap Supply

222.4M HYPE circulates of 955.3M outstanding against the 1B maximum; rank #11; 24-hour token volume $208M, +32.6% (CoinPaprika). The float is unchanged from the prior weeks' reviews, and this week carried no unlock event — the last two, Aug 29 (14.18M) and Sep 6 (9.92M), are the recent supply tests. With the float flat, the week's move to a $17.7B market cap reduces to price times an unchanged base.

Our take: with circulating supply effectively unchanged and no unlock event in this week's story, market-cap movement reduces to price times an unchanged float — which makes this story a restatement of story 1, and is printed as a baseline rather than a development. The supply entries that matter arrive on the calendar, not in the tape; August's two unlock tests (see the Aug 24–30 and Aug 31–Sep 6 reviews) are the recent precedent for how this market digests them.

16. Monday open: stabilization attempt Tape

Sep 14 trades near $79.78, +3.3% off the Sunday close, though still −7.6% on the 7-day window (CoinPaprika). The session's candle ran $76.91–80.29 — the low sits just above the week's $76.59 print, and the recovery from the Thursday close ($78.52 → $79.80) has covered roughly a quarter of the Sep 10 break. It is printed as the most recent data point, not as a forecast.

Our take: one session proves little, and we print it because it is the most recent data point, not because it predicts anything. The seven-day number is the honest one: +3.3% off a Sunday close is inside normal daily noise for this asset. If next week's review needs this story, we will have been wrong about the week in a way the fee series (story 5) would have shown first.

Three stories that mattered most

Sixteen items make a long list, so here is where we would spend attention if we could only follow three:

1. The fee series (items 5–7). A 48% one-month drawdown in the revenue run-rate is the fundamental event of the week; the price round trip is its reflection. Everything else on this list is either cause, cushion or context.
2. What did not break (items 8, 10, 11). Flat OI, neutral funding and range-bound TVL during a −12% week are the evidence that separates a positioning unwind from a confidence event. So far, the data reads as the former.
3. The revenue mix shift (items 12–14). A counter-cyclical yield stream, near-total revenue pass-through, and 28.5% of open interest in non-crypto markets are the structural facts that will outlast this week's candle. They are also the parts of the story that move slowest.

2. The fee story is the story

Price got the headlines, but the fee series is what repriced this week. Four consecutive weekly declines — $24.6M, $21.2M, $16.5M, $12.9M — is a 48% drawdown in the protocol's gross revenue run-rate in one month, and it coincides with the round trip in HYPE from the record. The mechanics are visible in the daily prints: the week's weakest days (Saturday $0.73M, Friday $1.44M) followed Thursday's −6.0% price break, consistent with retreat rather than fresh aggression.

The offset is structural: virtually all protocol revenue still flows to the Assistance Fund's buyback-burn, and the USDC aligned-quote-asset arrangement adds an estimated ~$160M/yr of reserve-yield-funded repurchases on top. The fee curve sets the cycle; the buyback plumbing sets the floor's mechanics.

3. What did not break

• Open interest: $10.32B with the HYPE book at $1.67B — the drawdown came on closing flows, not a liquidation cascade; funding at 0.010%/8h shows no crowded-leverage signature.
• TVL: a $6.28–6.44B range with the weekly close inside it — deposits did not flee the venue during a −12% token week.
• Breadth: 178 perp markets live, with ~$4B of OI in HIP-3 builder-deployed non-crypto markets — the venue's diversification story continued independently of price.

The week in seven sessions. The daily decomposition backs the week's central claim: six of seven sessions moved within a few points of BTC and ETH, and the damage came from one session — Thursday's −6.0% against BTC's −2.2% — with the fee peak ($2.81M) printing on that same day. The Friday that followed sharpens it: the majors bounced (ETH +3.2%, BTC +0.8%), HYPE barely joined (+1.1%). Session by session: Sep 7 $85.18 (−3.1%), Sep 8 $85.12 (−0.1%), Sep 9 $83.52 (−1.9%), Sep 10 $78.52 (−6.0%), Sep 11 $79.39 (+1.1%), Sep 12 $79.64 (+0.3%), Sep 13 $77.26 (−3.0%). The seven daily fees — $1.83M, $2.35M, $2.32M, $2.81M, $1.44M, $0.73M, $1.36M — sum to $12.8M, consistent with the weekly $12.9M print at rounding.

Measured after the week (Sep 14). Two structural checks were run after the Sunday close and are dated as such. First, the derivatives book: platform open interest measured $14.44B — $10.32B on the core DEX plus $4.12B across HIP-3 builder markets — so the fade ran against a stock that had not broken. Second, the builder-market census: 10 active builder DEXs, with xyz running 104 of the markets and ~98% of builder OI, and HyENA showing zero active markets — the first full deployer exit on the platform. These are archive measurements taken Sep 14, not week-of prints, and they are labeled accordingly.

4. Sources & method

Prices, candles and funding: Hyperliquid official API. Fees and TVL: DefiLlama. Supply, rank and 7-day change: CoinPaprika. Structural claims (AF revenue share, USDC yield share) are attributed to Galaxy Research's Hyperliquid research and Hyperliquid's own documentation, and are labeled as estimates where they are estimates; HIP-3 open interest is measured directly from the API's perpDexs endpoints. Raw API pulls for this week are archived in this site's build directory. Where a figure cannot be independently verified, it is not printed.

Frequently asked questions

Related reading

• Hyperliquid Weekly Data Report — Week of September 7–13, 2026
• Weekly Intelligence — Week of September 7–13, 2026
• The Pullback: Anatomy of HYPE's First Decline From the Record
• HYPE Market Intelligence — Week of September 7–13, 2026