A $7.1M butterfly: someone paid for a 95,000 bitcoin bet with September calls

At 13:04 UTC on 21 September a three-leg block for 30 October went through on Deribit: 1,000 bitcoin calls bought at the 90,000 strike and 1,000 at 100,000, 2,000 calls sold at 95,000. The exchange itself tagged it as a ready-made butterfly.

Within two minutes four more identical blocks went through, five more from 14:13, the last one at 15:01. Eleven blocks in all, exactly 1,000 butterflies in each.

The buyer paid 83.2 BTC. Bitcoin stood at 85,200-85,960, so that is about $7.1M, $646 a butterfly. So in two hours one bet on 11,000 butterflies appeared on three strikes.

Six of the eleven blocks had one more leg. Five sold 800 calls at 85,000 expiring 25 September each, the sixth sold 500 calls at 86,000. That is 4,500 calls with four days left.

For them the butterfly buyer received 90.9 BTC, about $7.76M, more than he paid for all 11,000 butterflies. The legs sit in one block, so the same participant sold the calls and bought the butterflies.

On 25 September bitcoin settled at 83,931, below both strikes, and those calls expired worthless. So he paid for the butterflies with calls that four days later were worth nothing.

Each butterfly pays up to $5,000 if bitcoin settles at exactly 95,000 on 30 October. On 11,000 that is $55M, 7.7 times what was paid.

The further from 95,000, the less. Below 90,646 or above 99,354 the payout is less than the butterfly cost, and below 90,000 or above 100,000 it is zero.

So the buyer needs bitcoin to rise 11.5% from the first block's price in five weeks and stop near 95,000. On 7 October a 250-butterfly position from our previous investigation appeared on the same three strikes: this one is 44 times larger.

The butterflies could also have closed an old position. Open interest tells the two apart: we compared the chain snapshot at 02:00 UTC on 21 September, before the blocks, with 01:50 on 22 September.

On the 95,000 strike it rose from 1,366 to 23,316 contracts, up 21,950 against 22,000 sold in the blocks. On 100,000 it rose 10,814, on 90,000 12,542.

So this is a new position. On the September 85,000 strike open interest fell 2,967 over the same day: at least part of the calls sold were an exit from calls bought earlier.

Four identical blocks went through in 33 seconds. Each sold 1,500 butterflies on the same three strikes and bought 500 call spreads 90,000/95,000 expiring 6 November.

In all 6,000 butterflies went for 7.8 BTC, about $647K, $108 each. The 2,000 November spreads cost 9.8 BTC, about $813K.

There were no other trades on the 100,000 strike that day, and open interest fell by exactly 3,000. So the seller closed at least half of these butterflies out of his own position. Since 1 July our archive shows butterflies on these strikes bought by the thousand only on 21 September. If it is the same participant, he lost about $3.2M on the 6,000.

If the buyer still holds the other 5,000 butterflies, they pay up to $25M on 30 October, again only between 90,646 and 99,354. At the exchange mark on 22 October they are worth about $407K.

The November spreads work differently: they have no upper limit. On 6 November they profit above 90,407 and pay a maximum of $10M at 95,000 and above. Right now they are worth about $589K against $813K paid.

At 06:37 UTC on 22 October bitcoin stood at 82,478. So part of the bet on exactly 95,000 on 30 October was moved into a bet on anything above 90,000 on 6 November, a week longer.