DRINKER LABS / ANALYTICS
On-chain report · 23 Sep 2026

From $BERT to a BTC short

On 22 September a Solana wallet sold its entire 4.59M $BERT position. About two and a half hours later it moved $139,022 in USDC onto Phoenix Eternal and opened a 3.2073 BTC short at about 2× leverage. As of this report the position is still open.

Position
Short 3.2073 BTC
Isolated BTC perp on Phoenix Eternal. Average entry $86,392, notional $277,086.
Leverage & risk
≈2.0× · liq. ≈$123.5–126k
$139.0k collateral. BTC would need to rise roughly 43–46% from entry to liquidate (estimate; see method).
Status
Open · ≈+$2.9k
No activity on the Phoenix accounts since 22 Sep 17:47 UTC. Unrealized P&L marked at Binance BTC $85,496, before funding.
01 · Position

One isolated short, filled in two clips

The wallet registered a Phoenix Eternal trader account and then created an isolated sub-account that can hold only one position. Each deposit and its market order ran in the same transaction, routed through Phoenix's Flight order proxy.

Time (UTC)DepositedSoldFill priceBinance 1m rangeNotionalFee
22 Sep 17:42:45$79,373.211.8326 BTC$86,426$86,457–86,525$158,384.75$79.19
22 Sep 17:47:25$59,649.001.3747 BTC$86,347$86,429–86,470$118,701.57$59.35
Total$139,022.213.2073 BTC$86,392 avg—$277,086.32$138.54
Collateral held
$138,786.65
read from the sub-account
Equity at mark
≈$141,759
BTC $85,496, before funding
Bankruptcy price
≈$129,695
collateral fully consumed
Liquidation (est.)
≈$123.5–125.9k
assumes 3–5% maintenance margin

Both fills printed below the low of the same minute on Binance spot, which fits a market sell working down the bid side. The taker fee on each fill was exactly 5 bps of notional.

02 · Flow of funds

Where the $139k came from

About 57% of the collateral came from that afternoon's token sales. The other 43% arrived from a second wallet, G9zppf…, less than two minutes before the second order.

The August buying was funded by $83,496 in five USDC transfers from another wallet, nQKJya…. Neither funding wallet matches our list of exchange hot wallets. Each has fewer than 100 transactions, so both look like private wallets rather than exchange withdrawals. We have not established who controls them.

03 · Timeline

A month in, one afternoon out

  1. The wallet is funded and buys 4,592,820.62 $BERT for $67,500 across 15 swaps. $83,496 USDC in from nQKJya…. Also $9,000 into WASABI, $6,000 into USDUC and $500 swapped to SOL.
  2. 4.5M $BERT and 20M WASABI move to a second wallet, BwM8qw…, in 19 transfers. That wallet's first-ever transaction is this transfer. All 84 transfers between the two wallets, in both directions, carry both wallets' signatures, so one party controls both. Each also has a third co-signer, trfb53…, which we have not identified.
  3. The $BERT comes back, then is sent out again, now all 4,592,820.62, in 23 smaller transfers.
  4. The tokens are pulled back in batches and sold between batches. By 15:14 the wallet holds no $BERT and no WASABI. $BERT sold in 11 swaps of 193k–712k tokens for $72,568.12. The second wallet's last token account is closed at 14:56, and it now holds nothing.
  5. Phoenix trader and isolated sub-account created. $79,373.21 deposited and 1.8326 BTC sold at market.
  6. $59,649.00 USDC arrives from G9zppf….
  7. $59,649.00 deposited into the same sub-account and 1.3747 BTC sold at market.
  8. Dust deposits of 0.0001–0.001 USDC arrive from look-alike addresses. Address poisoning: G9zWag…Hgyp and G9z1yy…Hgyp copy the first three and last four characters of the real funder G9zppf…Hgyp. This is unrelated to the trade.
  9. No further activity on either Phoenix account. The short is still open.
04 · The spot trades

A one-month round trip on $BERT, closed at +7.5%

The wallet sold exactly the $BERT it bought: 4,592,820.62 in and 4,592,820.62 out. The average sale price was about 7.5% above the average buy.

TokenBoughtCostAvg buyProceedsAvg sellResult
$BERT4,592,820.62$67,500.00$0.01470$72,568.12$0.01580+$5,068 (+7.5%)
WASABI (Wasabi Cheese)20,166,227$9,000.00$0.000446≈$7,149*$0.000354−$1,851 (−20.6%)

* $6,308.68 USDC plus 7.164 SOL, valued at the Binance SOL price at the time of the swap ($117.24). $BERT traded at $0.01519 on DexScreener at the time of writing, below this wallet's average sale price. After the sale the wallet still holds 12.38 SOL, about 1.12M USDUC and a balance in one unpriced token.

05 · Method

How we know it's a short

Phoenix Eternal logs each step in plain text (DepositFunds, Place Market Order, trade amount and fee), but its logs don't state whether an order was a buy or a sell. We read the direction from the account itself.

  1. We pulled every transaction for the wallet (141) and for its USDC token account (98) from public Solana RPC. We needed both lists: plain transfers into a token account don't appear in the owner wallet's own history, and that's how the $59,649 arrived.
  2. We read the isolated sub-account's current state. Its signed position field is −32,073 lots, and negative means short.
  3. Three checks back that reading up. First, 32,073 is exactly the sum of the two fills (18,326 + 13,747). Second, the collateral field in the same account ($138,786.652675) equals the figure Phoenix printed in its own log. Third, both fills printed below Binance's one-minute low, as market sells would.
  4. We confirmed the lot size (0.0001 BTC) because only that size puts the implied fill prices within 0.1% of BTC spot at the same minute.

Limitations. We decoded the account without Phoenix's official account layout (IDL), so the field offsets are inferred. The three checks above make an error unlikely, but it is possible. Binance spot stands in for Phoenix's mark price. Funding payments since entry are not included. The liquidation band depends on Phoenix's maintenance-margin parameter, which we did not verify. This is a snapshot, and the position can change at any time after 23 Sep 12:26 UTC.

06 · What the chain doesn't say

Open questions

  • Why. A 2× short funded partly by selling a memecoin could be a directional bet on BTC or a hedge against exposure held elsewhere. On-chain data can't tell the two apart.
  • Who. This report identifies wallets, not people. This wallet and BwM8qw… are provably under common control, because both signed every transfer between them. The two funding wallets are linked only by the transfers they sent.
  • What else. Positions on other venues, or in wallets that don't touch these, are out of view.