Report archive · 25.08.2026
Case No. 20260825 · 25.08.2026
Case № 20260825 · 25.08.2026 08:20 UTC · BTC $79,968 / ETH $2,487
analysis: Fable 5 by Anthropic on our own archive since February
Futures (Hyperliquid). The BTC short frame isn't just intact — it is getting deeper: the short is 2.6 times the long, $573.4M against $224.8M.
Over the week the book added $34M into short — and most of that step fell on the last day. Adding to a short while price rises is either a view or hedge management; the data can't tell them apart, the fact stands either way.
The ETH book moved the opposite way: the short is still 1.7 times deeper, but the weekly flow ran $104M toward long, and the last day added the same direction. Two books diverging by direction — that doesn't happen often.
There is less fuel in the system: the combined position size of overleveraged wallets is $673.7M, and 64% of it is longs.
Options (Deribit). Two separate positions. Quiet accumulation: in BTC, puts at one strike under the market and calls at two strikes above it — up to a very distant one — keep growing quietly; in ETH the biggest pile of the book is September puts at a single strike, with calls just above spot next to them. Blocks over the day: in BTC — a complex multi-leg with puts bought under the market and a call above it, and twice in a row the same financed entry upward — the structure of the day, strikes and size in the full version. In ETH — a combo leaning on the upside and a double call spread for September. A separate signature in the tape: someone methodically sold a very distant BTC call — dozens of trades, all one way. The rest of the ETH tape is the crowd: 4,340 small trades, no whale visible in them.
Money (the Hyperliquid border, 7 days, $250k threshold). Another negative week: $230.85M came in, $362.61M went out — a net $131.76M left the exchange. The cash through the bridge is mostly rotation: the lion's share of the withdrawn USDC settled on exchange deposit addresses. Coins through Unit move differently: more BTC came in than left, and what left disperses mostly to players' own wallets, past the exchanges. Tranches and addresses — in the full version.
This is positioning, not a direction forecast.
What the options market says. This is where you see what insurance against a price move costs. It costs less than the market actually moves: the Deribit option premium over realized movement is negative — −2.2 points in BTC and −4.0 in ETH. And the tilt of the price of protection stands on the calls' side: the market pays more willingly for taking part in a rise than for protection against a fall. This contradicts the whales: option demand looks up; nobody is scooping the cheap downside fear.
What the futures say. When there are more longs, they pay the shorts — that payment is called funding. The payment is positive and slowly rising: the BTC daily snapshot shows +0.0052% against +0.0047% yesterday, and in 8-hour equivalent Hyperliquid and Binance print the same +0.010%. This contradicts the whales: the long side of the perps is still paying for its position — the flow stands against their short.
What the crowd is doing. These are retail accounts on Bybit perps — the camp opposite the top Hyperliquid wallets. In BTC retail is nearly neutral: 53% long, near the middle of the scale; ETH is warmer at 65% long. The gap with the whales remains structural, and today's cut shows no sharp move in it. This camp is silent: without a move in the gap, the gap itself isn't news.
Any signs of stress. We check whether anyone is being forced out. Among the top Hyperliquid whales the day brought one forced closure of $2.4M, mostly long; on OKX swaps the day produced only 2% of the week's liquidation volume — an even background, no acceleration. This camp is silent: a single case is a statistic, not a cascade.
Cross-check in sum: two camps testify against the whales' short — the option tilt and the longs' payment — and two are silent. This is positioning, not a direction forecast.
The classic index prints 34.1 out of 100 (90-day window, 217 alts): deep bitcoin-season zone, unchanged on the day.
Our barometer (Hyperliquid perps) shows something livelier: alts hold 44.9% of open interest against 43.9% yesterday and 38.4% on 08.08 — the share has been growing for a third week. Hot funding sits in 77.1% of perps against 84.7% yesterday — the payment has cooled but remains mass-scale.
Retail in the top-10 alts is 65.7% long against 67.0% yesterday — hottest in DOGE (76.8%) and XRP (75.8%).
Whales hold 29.9% of their $1.29bn in alts as longs — practically unchanged on the day. Whales are structurally short in alts at all times — the signal lives in the CHANGE of the gap, and today it stands still. Full table: indiciadesk.com/en/altseason
Continuation ⟳. The corridor regime holds — the code confirmed the base scenario yesterday — but the filling has changed: the whales' short frame deepened into the rise, the bulk of the fuel moved under the market, and insurance costs less than the market actually moves. The structure is the same — the weight inside it has been redistributed.
Forecasts over the last seven sets (the code counts): 17 closed — 5 confirmed · 7 not confirmed · 5 cancelled at the boundary; four open.
What this does NOT mean: neither a turn down nor that the whales know the direction. Adding to a short into a rise is just as consistent with margin management and hedging as with a view.
🔒 In today's full RADAR: one fund holds the four biggest shorts of both books at a paper loss of tens of millions — who it is and where its close-out boundaries sit · the structure of the day — a financed BTC entry upward: strikes, expiries and size inside.
ANALYST access → indiciadesk.com/en/agent
💎DEEP
Portraits of the day (Hyperliquid, snapshot 08:16 UTC; weekly base — 18.08). The top of both books is still held by shorts with eight-figure paper losses — and none of them is cutting.
Two BTC shorts at 5×. The first — $125.7M from $69,878: uPnL −$15.90M, liquidation $132,545 (+66% from spot). The second — $98.3M from $65,312: uPnL −$18.04M, boundary $123,458 (+54%). Both have stood the same side for at least a week and added over that week. Low leverage, distant boundaries — the loss is simply being carried.
The turncoat at 10×. A $70.4M short from $78,107: uPnL −$1.67M, liquidation $96,390 — only +20% from spot. A week ago it stood the other side; over the day it added 70.8 BTC; the account, $16.0M, is a quarter of the position (self-signed "VBVIT" — not identification).
A long at 40×. A $56.0M position from $80,359 with a close-out boundary of $76,180 — just −5% from spot, the nearest boundary of the entire top. A week ago it didn't exist; 450 BTC added over the day on a $3.6M account (self-signed "Keisan2" — also not a name). The most aggressive structure in the book: a five-percent pullback closes it by force.
Two more BTC portraits. A $45.1M short at 20× from $74,045 (uPnL −$3.36M, boundary +92%) — flipped from long within the week and cut 52.6 BTC over the day. A $44.9M long at 40× from $79,262 (+$0.41M, boundary −20%) — new this week, cut 323 BTC over the day.
ETH's poles. The deepest loss of the entire book — a $122.5M short at 5× from $2,084: uPnL −$19.84M, liquidation $4,145 (+67%). Next to it a $101.2M short with the same signature (−$15.30M) and a $67.8M short at 15× (−$13.78M, boundary +61%). On the other side — two longs that didn't exist a week ago: $62.7M at 20× with +$10.49M and $49.7M at 4× with +$11.02M — the biggest gain of both books, its close-out boundary a full −59% from spot. A $52.2M long at 25× cut 8,100 ETH over the day.
Blocks over the day (our registry; venue — Deribit). BTC: a complex multi-leg of 2,500 contracts across three expiries (25SEP/28AUG/4SEP) — bought 500× call $80,000, 500× put $68,000 and 500× put $67,000, sold 500× put $72,000: upside participation plus deep downside insurance, paid for with the nearer put. And twice in the tape the same financed entry of 2,425 contracts (11SEP/25DEC/25SEP): bought 950× calls $82,000 and 125× calls $80,000, paid for with sold 950× calls $100,000 and 400× puts $62,000 — the structure of the day. ETH: a 6,000-contract combo (25SEP/28AUG) — bought 3,000× call $3,000 against sold 1,500× calls and 1,500× puts $2,400; and twice a 25SEP call spread — bought 1,000× $3,000, sold 1,000× $3,500. A separate signature: the BTC $175,000 call on 25SEP was methodically sold — 900 contracts across 23 trades, 100% one way. Market-maker anchor strength ≈ $203.4M (the sign is an estimate, and it shifts); state — "damping the move — holding as an anchor". The aggregate intent across the structures the code marks ⚪ no clear direction.
Cases in play (our registry, Deribit).
Case №546 · strangle ETH $1,000/$4,000 on 7,500 contracts to 25.06.2027. A far put and a far call bought — a position for a big move in either direction; entered at spot $1,806, day 51 in play today.
BTC $70,000 puts +7,627 (10 expiries) $82,000 calls +6,879 (10 expiries) $100,000 calls +5,926 (6 expiries) ETH $2,100 puts +35,695 (76% on 25SEP26) $2,600 calls +22,837 (10 expiries) $2,500 calls +22,197 (11 expiries)The biggest pile is still the ETH September $2,100 puts. In BTC the more interesting pair: the $70,000 puts and the $82,000 calls are growing at the same time — the book is preparing for a range of move, not for a side.
The flow by events (Hyperliquid, one day): the top-wallet tape gave 583 events — 172 adds against 114 cuts and only 8 reversals. Adding outweighs cutting one and a half to one: the book isn't fleeing, the book is being restocked.
Changing of the guard. Abraxas added to all four shorts over the week — 710.6 and 200.2 BTC into the first two, 15.5k and 19.5k ETH into the other two. Fasanara flipped into a BTC short from a long over the same week; "VBVIT" did the same. And the long side of ETH is being collected by fresh addresses: both of the book's most profitable longs appeared within the week. Old money deepens the short — new money enters long: the changing of the guard runs through the border of the book, not inside it.
The price of vol in the term structure (Deribit, our hourly history since February). The short end is tense, the long end is asleep: BTC 2-day IV 51.7 against monthly 43.3 — an inversion; ETH the same, 67.4 against 58.5. Inside our fear and greed index it shows even starker: the BTC momentum component sits in the 99th percentile of our archive since February while the risk-premium component sits at only the 18th. The move is there — the payment for the move isn't rising.
Max pain and the MM corridor (Deribit). The market makers' break-even corridor in BTC is $76,500–81,500: spot $79,968 sits in the upper half, max pain $79,000 — just under price. By our study an exact match of price and max pain happens only about a third of the time — a marker, not a prophecy (indiciadesk.com/en/docket).
Tomorrow's 26.08 BTC expiry. Max pain $79,000 — the Deribit and Bybit books agree; the day's walls — put $77,000 (127 contracts) and call $84,000 (149). September 25.09 is pulled to a single point: max pain $70,000 (Bybit sees $72,000), and both the put wall and the call wall stand on that same $70,000 strike (5,073 and 11,035 contracts). Across the whole BTC book: put wall $77,000 (1,516) — support, call wall $84,000 (6,765) — resistance.
ETH: the corridor is $2,420–2,600 at spot $2,487, max pain $2,520 — just above price. On 26.08 max pain is $2,500 (Deribit $2,520 · Bybit $2,490), the walls — put $2,460 (1,454) and call $2,500 (872). September: put wall $2,100 (30,911) — the very strike being quietly stacked — and call wall $3,000 (34,366). Across the whole book, right above spot stands a $2,500 call wall of 108,502 contracts — the densest level of both books.
PCR in time (Deribit). BTC put/call: month 0.43, week 0.56, now 0.60 — downside protection has been building without pause for a month. ETH is calmer: 0.50 a month ago, 0.51 a week ago, 0.54 now — the same side, weaker pace. The Bybit options book (tenors up to 30 days) is noticeably heavier in puts than Deribit: BTC 1.23 against 0.77, ETH 1.23 against 0.82 — whose book it is we don't know, so we hang no labels on it.
The priced move and the stress test (Deribit, our build_gex calculation). Options price BTC at ±2.3% per day and ±6.1% per week; ETH ±3.1% and ±8.1%. At −2% ($79,159) BTC is still inside positive MM gamma — a pullback gets bought back; at +2% ($82,390) price passes the $80,000 call wall — resistance thins. ETH at −2% ($2,460) is in positive gamma too; at +2% ($2,561) it is beyond the $2,500 call wall.
Gamma and fuel. The vacuum above BTC — $82,000–84,000 (2.4% from spot): little MM hedge there, the move runs faster; below, a thin zone at $73,000–74,000. The cascade map (Hyperliquid, snapshot 08:16 UTC): $94.8M of long fuel in the $75,992–78,392 band — right above the $77,000 put wall; $43.2M of short fuel in $81,592–83,992 — under the $84,000 call wall. There is twice as much fuel under the market as above it: a bigger RANGE of move for the same push down, not a higher PROBABILITY. ETH is nearly empty below ($9.9M) and entirely empty above. This is a relief map of risk, not a forecast — liquidations lag price. And the day's main asymmetry: hundreds of millions of fuel in the system while insurance costs less than the market actually moves — the cascade is underpriced.
🆓TAIL
Pendulum · Wheel — standing position: +$682 on the day · +$6,567 in total · counted since 11.05.2026.
signal trades win rate last total since Dust Strategy V2 37 5% −$1,125 +$92,969 21.03 Dust Strategy V1 31 10% −$1,125 +$37,031 20.03 Skew 2.0 V2 31 42% +$1,034 +$13,878 26.05 Volatility Convergence V3 16 44% −$634 +$7,614 26.05 Skew 2.0 V1 34 38% −$448 +$5,486 26.05Deepest in the red: Wheel Trail V8 −$7,088, then V6 −$6,081 and V7 −$4,060 — this family catches reversals, and the market has been walking in trends for weeks: signals against the current lose systematically. The mirror fact at the top of the table: the most profitable signal, Dust Strategy V2, wins only 5% of its 37 trades — rare big payoffs cover dozens of small losses. Win rate isn't money.
The code closed the set from 22.08 and the boundary scenarios of 23–24.08:
• "The corridor holds under the upper edge" (from 22.08, 45%): CONFIRMED — in range 100% of the time (close 78,485).
• "Burn-up through the short fuel" (from 22.08, 30%): NOT confirmed — high 79,256 against the 82,000 level.
• "Drop into the $71–72k vacuum" (from 22.08, 25%): NOT confirmed — low 76,080 against the 72,000 level.
• "Drop below the put wall into the fuel zone" (from 23.08, 25%): CANCELLED at the boundary — high 80,775 against the 80,000 level.
• "Drop into the long-fuel band" (from 24.08, 25%): CANCELLED at the boundary — high 80,775 against the 80,000 level.
Open: "The expiry magnet holds the corridor" (45%) and "Burn-up through the short fuel" (30%) — to 25.08, 15:11 UTC; "The expiry pin holds the corridor" (45%) and "Burn-up through the short fuel" (30%) — to 26.08, 15:21 UTC.
From the conclusions scored against spot history: yesterday's cut showed "no clear tilt", and what happened was BTC +3.7% — the neutrality didn't hold, the move came anyway. A week ago the cut showed a "risk tilt up", and what happened was BTC +24.7% — the direction was confirmed.
Over the last seven sets 17 scenarios were closed: 5 confirmed · 7 not confirmed · 5 cancelled at the boundary. Cancellations are nearly a third: a scenario dies not of time but of a level being touched — which is exactly why we publish the boundary together with the scenario, not after the fact.
We also count what doesn't work (our registry, 218 checked forecasts since February): the type range_BTC came true in only 37% of 152 cases — worse than a coin toss, so we don't use it on its own; tilt_BTC gives the same 37% on 138 cases. For contrast: the type relative holds 69% across all 218.
Not investment advice and not a recommendation to trade. Derivatives trading carries a high risk of loss; INDICIA DESK is a research publisher, not a licensed adviser under MAS, the SEC or the FCA — decisions and their consequences are yours.
Radar is assembled automatically from live Deribit / Hyperliquid / OKX data · time-stamped snapshot. A journal of the system's decisions, not investment advice. © 2026 INDICIA DESK.