Three weeks under water, then up 177%: how they held and took $8.7M on ether
In July someone bought 21,929 ether calls with the price at 1,950. On 22 September they sold the lot at 2,730.
$4.85M in, $13.56M out. Up $8.7M in 57 days.
And right away they sold about 4,000 more calls they did not hold. Not a bet on a fall. A bet that ether stays below 2,860 to 3,200 into March.
- Source: our hourly archive of every Deribit trade with the aggressor side tagged, since 11 June 2026
- Open interest: daily Deribit chain snapshots at 04:00 UTC. Archive volumes match the exchange's daily volumes to the contract
- Data frozen 23.09.2026, 18:30 UTC
Bought in blocks on the morning of 27 July, sold in blocks on the morning of 22 September on the same strikes. Volatility at entry and exit the same: price did all the work.
What happened
In two and a half hours, 05:57 to 08:29, 27 blocks of ether calls went to the market: March 2027 at 1,900, 2,300, 2,500, 2,800 and December 2026 at 2,000. Ether traded around 2,730 meanwhile.
All those strikes sit below price. So the calls being sold were deep in the money, not the cheap edge. That is not how you bet on a fall. That is how you take profit.
Across all ether calls that day 32,721 were sold net. Two thirds of that falls on five instruments bought from 27 July. Those five, below.
July: what was bought
At 06:16 a block of 1,000 March calls at 1,900. At 07:15 and 07:18 two blocks of 2,500 and 2,350 at 2,500. After noon three more, at 2,300, 2,500 and 2,800. Ether about 1,950, volatility 56.
Next morning open interest on those strikes was up by exactly what was bought, almost 11,700 in a day. A new position, not someone else's shorts being closed.
By 17 September almost 10,000 more were added. 21,929 contracts in total. A jump in open interest like that is what a strike beacon catches: it writes to Telegram the same day.
Bars: contracts in five instruments bought from 27 July and sold on 22 September, per our archive. Dashed line: the exchange's open interest on the same five strikes. Blue line: ether price.
57 days
For the first three weeks the position was under water: ether moved between 1,840 and 1,920 and the calls were worth less than was paid. They did not sell.
Then ether went from 1,950 to 2,730, up 40%. Calls deep in the money move almost like the coin. Volatility was the same at entry and exit, so it played no part.
The tally: $4.85M in, $13.56M out. Up $8.71M. In coin exactly double.
Green line: position value at the archive's trade prices, $M. Red dashed: premium paid. Between days without trades the last known price is carried.
September: sold and added
They sold more than they held: 21,929 was the position, and about 4,000 more went out. Whether truly from zero is what open interest says: if there are more contracts in the system after the sale, the seller opened new ones rather than handing over his own.
On 2,300 and the December 2,000 contracts went down: that was closing. On 1,900, 2,500 and 2,800 there were more contracts open after the sale than before it.
So the July long was closed in full, and on three strikes new short calls were left behind. The money was taken and the same upside sold to the market again.
What the new shorts bet on
For the new calls the seller collected about $2.5M of premium. They keep it if in March ether is below strike plus premium: 2,860 for 1,900, 3,050 for 2,500, 3,200 for 2,800.
That is not a bet on a fall: on a 10% drop these calls give back only part of their value, a put would pay many times more. It is a bet on a ceiling: ether may rise, but not above three thousand by spring.
The short delta of these calls is about $7.7M. If spot sits underneath, these are covered calls. If not, it is a naked short with risk above 3,200. A beacon on the March 2,500 and 2,800 shows when this position moves.
The next day
On 23 September, the day after the sale, ether fell 3.3%, from 2,753 to 2,661. There is no link to the sale: the seller was not betting on a fall. They sold calls in the money, taking profit rather than buying a cheap shot at a crash. In a hundred days of archive this is the fourth one-day drop of that size.
The big sale did not repeat. On 22 September sellers of ether calls took $19.4M more out of the market than buyers paid. On 23 September the figure was $0.58M, more than thirty times smaller. A one-day event, not the start of an exit.
The large trades of 23 September are different in kind: holders of calls already in the money move them to a higher strike ahead of Friday's expiry, on bitcoin from 70,000 to 75,000, on ether from 2,300 to 2,450. Part of the profit is taken, the bet on higher prices is kept. Housekeeping before expiry, not a bet on a fall.
The new shorts at March 1,900, 2,500 and 2,800 are open. If ether goes above 2,860 they start losing, and the same chain snapshots will show it. We return to these strikes when open interest on them moves again.
We see this every day
The radar goes out every morning: where open interest moved overnight, which blocks went through, and at which levels market maker behaviour breaks.
Open the radar in TelegramWatch a specific strike →Next
Evidence and limits
Flow is counted from the hourly archive of Deribit trades since 11 June 2026, every trade tagged with the aggressor side. For 22 September the archive volumes on all six instruments match the exchange's daily volumes to the contract.
Open interest comes from daily chain snapshots at 04:00 UTC. The change between 22 and 23 September covers the full trading day.
The exact ledger: 21,929 bought; on 22 September 27,063 sold and 1,144 bought back on the same five instruments, net 25,919; of that 21,860 closed the July long and 4,059 were sold beyond what was held; open interest confirms at least 3,793 of those as new shorts. These figures are not to be mixed.
Delta and breakeven of the new shorts come from the 23 September chain snapshot and the 22 September sale prices.
We do not know who this is: the exchange gives the side of a trade, not a name. That the July buyer and the September seller are one hand shows in fingerprint, strikes and open interest, not in an identifier. The 34,000 ether that arrived on Hyperliquid on 17 September is not linked to this: different venue.
A decision journal, not investment advice. Reconstructed from public exchange records.