Everyone sees the whales are shortthe asymmetry today is that they took a half-step back and not one witness followed them.
Case #20260907 · 07.09.2026 08:56 UTC · BTC$79,427 / ETH$2,492 · analysis: Fable 5.1 by Anthropic
📅YESTERDAY
The code closed the three scenarios set on 04.09. "The corridor holds" — confirmed: BTC spent 100% of the time inside the band and finished at $79,904. "Burn-up through the short fuel" — not confirmed: price never reached the $83,000 trigger. "Slide under the gamma flip" — not confirmed: it didn't reach $77,300 either.
Six more forecasts remain in play: three from 05.09 run to term today at 10:00 UTC, three from 06.09 stand until tomorrow. Their verdicts come in the next issues, once the code has computed them.
⚡THE GIST
• The Hyperliquid perp whales stepped back from the short a little over the day: the BTC tilt is 2.6 times short-over-long against 3.1 yesterday, and the day's net moved toward long — against the week's drift into short.
• The Deribit options book keeps quietly stacking calls just above price: more than 10,000 contracts over the week at two strikes, $80,000 and $82,000, in small lots outside the blocks.
• Spot ETF funds didn't trade over the weekend; the last evidence is the week to 04.09, when $987M went into BTC funds against $924M the week before.
• Insurance is still cheap but has lifted off the floor: BTC two-day vol sits in the 29th percentile of its own hourly history since February against the 24th yesterday.
• Verdict — continuation: the structure is the same, whales short, with the funds, the options book and funding against them. The split narrowed over the day not because the witnesses crossed to the whales' side, but because the whales stepped back a little.
• 📊 Our Polygon — Pendulum: standing position, −$32 on the day, +$9,956 in total since 11.05.
• 📊 Our Polygon — signals: best — Dust Strategy V2 +$90,719 at a 5% win rate · worst — Wheel Trail V8 −$7,938. Table at the bottom.
🐋WHALES: WHAT CHANGED — three arenas
Arena one — Hyperliquid futures. The top whales in both assets still stand mostly short, but over the day that tilt got shallower, not deeper. In BTC the short is 2.6 times the long against 3.1 yesterday: both sides thickened over the day, the long side faster. Over the week the net is still shifted toward short; over the last day it moved back toward long — the direction changed inside the window. ETH tells the same story, only louder: the day's move toward long is ten times the whole week's drift into short. Overleveraged whales under watch: 61 against 51 yesterday, and 71% of them stand long. None is close to a margin call: there's more fuel under the market, nobody has lit it.
Arena two — Deribit options. Two separate whale positions here, and today they again don't line up. The first — quiet accumulation in small lots outside the blocks. In BTC the book stacks only the upside: calls at three strikes from the current price and higher. In ETH it's both sides at once — calls above the market and puts below, and the puts at two strikes together came to slightly more than the calls at one. The second — block trades. Over two days the snapshot recognised 14 structures in BTC and 6 in ETH; the most visible are a put spread that pays as long as BTC holds above its nearer strike, and an ETH condor expiring today, 7 September. Three new cases of the day, among them a structure rare for this book, with two expiries — legs and premiums are in the paid part.
Arena three — money at the exchange border. Cash is coming in, coins are going out — and cash outweighs. The "bridge" channel is USDC cash: over seven days $95.6M came in and $27.4M went out — three and a half times more entered than left. The "Unit" channel is the coins themselves: in BTC withdrawals exceeded deposits ($16.4M against $10.8M); in ETH and SOL a small inflow. Where the cash went: $19.5M of the $27.4M outflow settled on exchange deposit addresses, the rest went to players from the registry. Deposits came mostly from private addresses with no public label. Border snapshot 07.09, 08:29 UTC.
This is positioning, not a direction forecast.
🧭CROSS-CHECK — five camps of witnesses
What the funds are doing. Spot ETFs are the only camp that measures purchases of the coin itself, not a leveraged position; the flow figures come from SoSoValue. Funds trade on business days only, so there are no fresh numbers for the weekend: the last trading day is Friday 04.09, and the evidence hasn't changed since. Over the week 31.08–04.09 BTC funds took in $987M against $924M the week before: the flow is holding and thickening slightly. On Friday the buying came above all from BlackRock's IBIT and Fidelity's FBTC. ETH is the reverse — $218M against $824M a week earlier, a shrink of almost fourfold; in SOL the weekly flow compressed to a token sum, and Friday was negative. This contradicts the whales: institutions are buying the coin itself where the whales hold short; nothing new over the day.
What the options market says. This camp measures one thing: what insurance costs on the Deribit book. Our fear-and-greed index stands at 85 out of 100 — two points below yesterday, and still extreme greed. The greed comes from quiet, not from a rally. BTC two-day vol sits in the 29th percentile of its own hourly history since February against the 24th yesterday: cheaper than in seven hours of ten, but no longer right at the floor. This contradicts the whales: the whales hold short, and the book is barely paying for fear of a fall.
What the futures say. This camp measures who pays whom to hold a position. BTC funding on Hyperliquid is +0.007% in eight-hour terms against +0.010% yesterday: longs are still paying shorts, but less — the leveraged crowd cooled a little (what funding is). The BTC futures basis came up from negative yesterday to zero: the term price has levelled with spot. This contradicts the whales: longs keep paying to hold, so the leveraged crowd stands against the whales' position — if more weakly than a day ago.
What the crowd is doing. This camp measures how retail accounts stand against the whales. On Bybit perps 66% of ETH accounts stand long, 54% in BTC; both shares slipped a point over the day. The gap between the crowd and the whales is structural and almost always there; the event is its motion, and over three days it barely moved and sits in the bottom quarter of its own history. This camp is silent: it adds no new information over the day.
Any signs of stress. This camp measures whether anyone has already been forced out of the market. Among the top Hyperliquid whales, one forced closure over the day, a $2.9M long — yesterday there was none. On OKX swaps $437k was liquidated, mostly shorts; that's 9% of the week's volume — an even background, no acceleration. This camp is silent: one liquidation isn't a cascade.
Again: this is positioning, not a direction forecast.
🌡ALTSEASON — the classic and our barometer
The classic altseason index stands at 37.3 out of 100 — bitcoin-season territory; over the day it gave back a point from 38.2 after the jump the day before.
Our barometer watches money, not prices, and counts it from the Hyperliquid perp registry. Alts' share of open interest is 44.4% — $4.86B of the whole market's $10.95B.
Hot funding covers 88% of contracts against 82% yesterday — the heat is back.
Retail in the top-10 coins stands 66.9% long; the hottest are XRP and DOGE, both near 77%.
Whales in alts are structurally short: only 35.3% long across $1.43B of positions, the biggest of them in HYPE. That gap is permanent; the signal is in its CHANGE — and over the day it moved by less than a point.
Full barometer: indiciadesk.com/en/altseason
⚖️VERDICT OF THE DAY
continuation ⟳ — the structure is the same as yesterday: perp whales short, and three independent camps of witnesses — the funds, the options book and funding — against them. But the split narrowed over the day not because the witnesses crossed to the whales' side, but because the whales stepped back a little: the BTC tilt is shallower, and the day's net moved toward long.
Forecasts this week: 21 — 4 confirmed · 1 partially · 6 not · 4 cancelled at the boundary · 6 in play.
What this does NOT mean. It isn't a claim that the whales are giving up and BTC goes against them. One day of a shift toward long after a week into short is a change of pace, not of direction; the system logs it as a first weak sign, and nothing more.
💎SCENARIOS
The corridor holds — 40% · Burn-up through the short fuel — 33% · Slide into the long fuel — 27%
This is how we weigh the odds — not a promise; trigger and cancellation levels are in the full version.
🔒In today's full RADAR:
— one address holds the two biggest shorts in the market, and over the day a third big short added to its ETH position — with sizes, the distance to a margin call and what it's already costing the owners;
— case #1053: a structure rare for this book, BTC with two expiries — legs, net premium and payoff profile.
ANALYST access → indiciadesk.com/en/agent
💎DEEP
🔒WHALES IN DETAIL 💠
Portraits from the Hyperliquid registry, snapshot 08:16 UTC. Nicknames aren't treated as identification.
The market's biggest loss.ETH SHORT $207.1M, leverage 5×, entry $2,228, uPnL −$21.8M, liquidation at $3,687 — 48% above spot. Account $108.4M. Over the week the address added 28,073 ETH: the short is being built, not lifted. The same address in BTC. SHORT $171.1M, leverage 5×, entry $72,152, uPnL −$15.7M, liquidation at $126,056 (+59%); 447BTC added over the week. Together the two positions carry $37.4M of unrealised loss and are nowhere near a margin call. A second address with the same pattern.ETH SHORT $113.5M and BTC SHORT $98.8M, both at 5× leverage, uPnL −$15.6M and −$16.9M, liquidations at $3,875 and $131,093 on a $66.8M account. Together with the first address that's $69.9M of unrealised loss across the two; yesterday it was $71.5M, so the loss compressed slightly over the day. A third big short added.ETH SHORT $79.1M at 15× leverage, entry $2,169, uPnL −$10.2M, liquidation at $3,836 (+54%), account $49.6M — over the day the address added 1,893 ETH to the short. Its BTC SHORT of $24.5M at 20× has its liquidation all the way up at $220,307: nearly three times the buffer there. The market's biggest gain.ETH LONG $49.8M, leverage only 4×, entry $1,936, uPnL +$11.1M, liquidation at $1,020 — 59% below spot, account $12.4M. Held for at least a week. Next to it a long of the same breed: ETH$75.4M at 20×, entry $2,134, uPnL +$10.8M, liquidation at $1,755. The risky side.BTC LONG $42.3M at 40× leverage with liquidation at $69,516 (12% below spot) and ETH LONG $96.2M at 25× with liquidation at $2,352 (6% from price) — one wallet, account $8.2M. New this week.BTC LONG $39.7M at 40×, entry $79,914, liquidation at $72,752 (−8%), account $4.1M — a week ago the position didn't exist; 300BTC added over the day. The address carries a self-signed nickname; that's a nick, not an identification. Closest to triggering.BTC SHORT $23.8M at 30× leverage, entry $79,018, liquidation at $80,687 — only 2% above spot; an address with no public label, the position new this week. Yesterday's closest short had a 3% buffer.
Block structures in the snapshot — three new cases, all from the Deribit book. Case #1053 · BTC call ratio spread with two expiries: 50 calls at $86,000 sold for 25.09.2026, 30 calls at $88,000 bought for 25.12.2026 — 80 contracts, entered at $79,436. More sold than bought: the structure gains on moderate growth, while a sharp run past $88,000 before the September expiry turns into a loss. Net premium — about 1.04 BTC paid (≈$82k), because the far call costs more than the two nearer ones. Rare for this book.
Профіль виплат на експірацію 2026-09-25. Кит заплатив $83k премії. Беззбитковості в показаному діапазоні немає.
Case #1054 · BTC risk reversal: 25 puts at $50,000 bought, 25 calls at $125,000 sold, both legs for 25.12.2026 — 50 contracts, entered at $79,436. Protection against a deep fall paid for with a ceiling; net premium about 0.015 BTC received — the structure is nearly free for its owner.
Профіль виплат на експірацію 2026-12-25. Кит отримав $1k премії. Беззбитковість: $124 638.
Case #1055 · ETH condor at $2,440/$2,460/$2,500/$2,520 expiring 7 September — 600 contracts, entered at $2,490. Net premium about 0.42ETH received (≈$1k). It's the same corridor as case #1048, which the code has already closed.
Профіль виплат на експірацію 2026-09-07. Кит отримав $1k премії. Беззбитковість: $2 453 і $2 507.
Case #948 · ETH condor at $1,700/$1,900/$2,800/$3,000, size 11,500, entered at $2,462, expiry 25.09.2026 — thirteenth day in play. Net premium about 36.8ETH received (≈$91k): the whale is paid for price staying between $1,900 and $2,800.
Профіль виплат на експірацію 2026-09-25. Кит отримав $91k премії. Беззбитковість: $1 867 і $2 830.
Case #546 · ETH strangle at $1,000/$4,000, size 7,500, entered at $1,806, expiry 25.06.2027 — sixty-fourth day in play. Net premium about 293ETH paid (≈$530k) — a position on a big move either way.
Профіль виплат на експірацію 2027-06-25. Кит заплатив $530k премії. Беззбитковість: $859 і $4 141.
Closed over the day: case #1048, an ETH condor on the same $2,460–$2,500 corridor — the code scored it "confirmed" by the calculation at expiry: ETH finished at $2,486, the whale about +$1k. Case #1042, a BTC bear call spread — "not confirmed", price didn't move by expiry, the whale about −$2k. ΔOI over the week, both ways:BTC+5,428 calls at $82,000 (spread across 9 expiries), +5,194 at $80,000 (67% for 25.09) and +2,960 at $90,000. ETH: +22,663 calls at $2,800 (78% for 18.09) against +13,969 puts at $2,400 and +11,308 at $2,300.
🔬Who exactly
Both of the biggest shorts are Abraxas Capital Mgmt (Heka Funds): two addresses of one house, ETH and BTC on each. The house carries $69.9M of unrealised loss and increased the short in both assets over the week.
The third big ETH short, the one that added 1,893 ETH over the day, is Fasanara Capital: $79.1M at 15× leverage, uPnL −$10.2M on a $49.6M account.
Both of the riskiest longs — BTC at 40× and ETH at 25× — are held by Machi Big Brother: $138.5M combined on an $8.2M account.
The new 40× BTC long sits behind an address self-signed "Keisan2"; that's a nickname, not an identification.
Money by address. A whale from the registry topped up margin: over the week the Fasanara Capital address brought in $16.0M USDC in four tranches — two of $5.0M and two of $3.0M — and it's the same address building the ETH short. The biggest cash deposits were two tranches of $15.0M USDC each to an address with no public label. The cash outflow went to Coinbase deposit addresses — $9.90M and $7.50M. The coin channel sent $4.89M and $2.93M in BTC to Wintermute, while BTC worth $3.96M and $2.89M came back in via Binance.
🔒THE WEEK'S DYNAMICS 🔬
Perps.BTC: net $290M short (long $185M against short $476M); over 7 days an $11M shift toward short, over the last day $6M toward long — the direction changed inside the window. ETH: net $270M short (long $347M against $617M); $3M toward short over 7 days, $32M toward long over the day — here the pace of the turn is ten times the weekly drift.
Gamma flip.BTC$68,565 — up roughly 2,200 over the week; ETH$2,271 — up 45. When the flip and the walls crawl after price, the market makers are ceding the level; when they stand still, the level is real and the market makers hold it (what the flip is).
The price of fear. On the weekly window the premium is negative: IV38.6 against realised HV 39.7, and over 7 days it compressed by another 0.4. Options are cheaper than the market actually moves — a rare regime.
Calendar. The nearest expiry, 08.09, moved its max pain point by −1,000 over the week — the flow is dragging the consensus down, after spot.
Changing of the guard. Over the week the whales assembled 201 confirmed structures: 60 bullish, 48 bearish, 26 on volatility, 67 on range — against 207 yesterday, so a few old ones rolled off and almost none were added. The biggest — a call diagonal on 6,000 ETH. The lean is mixed: there's no single direction in them.
🔒CROSS-CHECK IN DETAIL 💠
BTC levels. Everything below comes from the Deribit options book. The market-maker break-even corridor is $78,500–$81,500 — the ceiling came down 500 over the day from $82,000; by our study across 368 expiries price stays inside such a corridor in roughly 80% of cases (what the corridor is). Below, a put wall at $79,000 on 2,324 contracts — yesterday the floor stood at $78,500 with only 246: over the day it thickened almost tenfold and moved up to price. Above, a call wall at $82,000 on 14,251 contracts — the ceiling is still heavier than the floor, but by six times now, not by tens. Max pain $80,000.
ETH levels. Corridor $2,420–$2,540, max pain$2,480 (what max pain is). Put wall $2,100 on 53,995 contracts, call wall $2,580 on 2,617 — here the floor is incomparably heavier than the ceiling, as yesterday.
The nearest expiry, 08.09.BTC: max pain$79,500 on the combined book of two exchanges (Deribit $80,000 · Bybit $79,500), put $79,000 on 151 contracts, call $82,000 on 317 — a thin expiry, its weight is token. ETH: max pain$2,490, put $2,100 on 3,151, call $2,580 on 2,120.
The quarterly, 25.09.BTC: max pain$72,000, and both put and call are densest at $70,000 (8,722 and 10,962 contracts) — this expiry holds 44% of all BTCOI, and its centre of gravity sits ten thousand under spot. ETH: max pain$2,150, put $2,100 on 37,104, call $3,000 on 41,782.
Fuel. Above the market $49.8M of short positions in the $81,026–83,409 band, below it $33.5M of longs in the $75,465–77,848 band. One and a half times more fuel above spot: that's about the RANGE of a move for the same push, not about a higher probability of that move.
Gamma vacuum. Empty above at $86,000–$88,000 and below at $72,000–$73,000: no market-maker hedge sits there, and a move through those bands runs faster.
Stress test ±2%. At −2% ($77,847) the dealers are still in positive gamma — a pullback gets damped and bought back. At +2% ($81,025) price is still under the call wall — the ceiling holds the move.
OI flow.BTC put/call: month 0.58 · week 0.56 · now 0.55 — the ratio is stable, the book isn't repositioning.
Two options books.BTC put/call 0.54 on Deribit against 1.21 on Bybit; ETH0.64 against 0.82. We don't know who owns the Bybit book, so this is neither "whales" nor "retail" — just a second book with a different lean.
IV by tenor.BTC 2d P33/C34 · 7d P35/C37 · 30d P37/C38, percentiles since February 29th, 27th and 32nd. ETH 2d P41/C43 · 7d P44/C47 · 30d P50/C52, percentiles 23rd, 18th and 29th. ETH two-day vol climbed over the day from the 14th percentile to the 23rd — the fastest move of any tenor.
💎SCENARIOS IN DETAIL 💠
The corridor holds — 40%. ⏱ 2 days, to 09.09. BTC stays inside the $78,500–$81,500 band. Cancellation level: a touch of $78,000.
Burn-up through the short fuel — 33%. ⏱ 2 days, to 09.09. Trigger: a touch of $81,100 — the bottom of the band holding the short positions. Cancellation level: a touch of $78,000.
Slide into the long fuel — 27%. ⏱ 2 days, to 09.09. Trigger: a touch of $77,800 — the top of the long-fuel band. Cancellation level: a touch of $81,100.
This is how we weigh the odds — not a promise; the code will compute the verdict from spot history.
🔭WHAT TO WATCH
• The $82,000 call wall on 14,251 contracts: while price is under it, the ceiling holds the move up (💠).
• The $79,000 put wall on 2,324 contracts — thickened almost tenfold over the day and stands half a percent under spot: the first level where dealers again have something to hedge from below (🔬).
• The BTC gamma flip at $68,565 — its break changes the regime first; we're watching whether it keeps crawling higher (🔬).
🆓TAIL
📊POLYGON
(our own signals at real prices, not a backtest)
⚙️Pendulum · Wheel(standing position, real prices)
−$32 on the day · +$9,956 in total · since 11.05.2026
⭐ Top-5 signals($1,000 per signal; "total" is the sum of all trades, not the return on a single thousand):
signal trades win last total since
Dust Strategy V2 395%−$1,125+$90,719 21.03
Dust Strategy V1 329%−$1,125+$35,906 20.03
Volatility Convergence V3 2245%−$507+$14,608 26.05
Skew 2.0 V2 3837%+$315+$11,762 26.05
Fear Flash V2 2232%−$418+$3,023 16.05
Deepest in the red: Wheel Trail V8 −$7,938 · Wheel Trail V6 −$6,908 · Against the Current · top V1 −$5,598 — all three hunt a reversal, and all three pay for every attempt while the trend continues. The most profitable signal — Dust Strategy V2 — wins only 5% of its 39 trades, and its last trade is in the red too. Win rate isn't money.
📅REVIEW OF PREVIOUS CONCLUSIONS
(verdicts are computed by code from spot history)
The scenarios from 04.09 are closed: "The corridor holds" confirmed — BTC spent 100% of the time inside the band and finished at $79,904; "Burn-up through the short fuel" not — the $81,267 high never reached $83,000; "Slide under the gamma flip" not — the $79,151 low never reached $77,300. So of the three, the one we weighted most (45%) came through, and both directional ones didn't.
Yesterday the system saw the risk tilted down — BTC made −0.5% over the day: price slid that way, but softly, under the confidence threshold. A week ago it also saw the risk tilted down, and BTC went +1.8% — the fear built into the positioning didn't materialise.
Weekly calibration: of 21 scenarios set since 31.08, the code closed 15 — 4 confirmed, 1 partially, 6 not, 4 cancelled at the boundary; 6 still in play. Under a third are full confirmations: the system calls it worse than we'd like, and we publish that.
The tally by name, whole history: "Burn-up through the short fuel" — 5 confirmed of 16, cancelled at the boundary three times. "The corridor holds" — 4 full of 12 plus 2 partial. "Slide under the gamma flip" — 4 of 11. "Squeeze up into the short fuel" — 4 of 10. "Slide into the long fuel" — 2 of 5. "The corridor holds under max pain" — 3 of 5 plus one partial.
The failing type in one line: "ETH slide into the long fuel" came through in neither of its two settings — one not confirmed, one cancelled at the boundary; it isn't in today's set, and the system doesn't use it. The corridor type holds at a third — worse than a coin toss, which is why its 40% today is the ceiling for us.
📏Scale calibrator
• What protection costs. Our fear-and-greed index — 85 out of 100. We compute it from the price of volatility and the option skew on Deribit, not from news, which is why it diverges from the well-known index.
• Cheap or expensive for this market itself.BTC two-day vol — 29th percentile of the hourly history since February, weekly — 27th, monthly — 32nd: cheaper than in most hours since February.
• Which way protection tilts.BTCskew is absent from today's snapshot — the source returned no number, so we're not calibrating the tilt today; on the last available snapshot calls were slightly pricier than puts. Terms: indiciadesk.com/en/glossary/
Not investment advice and not a recommendation to trade. Derivatives trading carries a high risk of loss; INDICIA DESK is a research publisher, not an adviser licensed by MAS, the SEC or the FCA — decisions and their consequences are yours.