A $323M stack that "caught the bottom": where the bit-linked whale bought, and the tell that says it's selling

A $323M stack that "caught the bottom": where the bit-linked whale bought, and the tell that says it's selling

Ten addresses. $323 million of ETH and BTC longs. $41.95 million of floating profit. That is the screenshot doing the rounds this week, sourced to the Chinese on-chain analyst ai_9684xtpa via Huoxing Finance, for a cluster its tracker tags "bit-affiliated". The headline writes itself — smart money caught the bottom again — and on current market data it belongs to a specific regime: Bitcoin up 23% in the last 20 days and 32% over 60, Ethereum up 58% over 60, and the crypto fear/greed index at 73. A rebound that size is exactly when "floating profit" posts circulate.

Here is the breakdown as reported: 65,977.97 ETH worth about $165 million with $15.08 million of unrealized gain, and 2,000 BTC worth about $158 million with $26.87 million of unrealized gain. Add the two and you get the $323 million and the $41.95 million — case-closed headline arithmetic.

The floating profit is not the interesting number. The number behind it is the entry, and it is reproducible on a napkin: value minus unrealized gain, divided by token count.

Asset	Tokens	Position value	Floating profit	Implied average entry	52-week low
ETH	65,977.97	~$165M	$15.08M	~$2,272	$1,507
BTC	2,000	~$158M	$26.87M	~$65,600	$57,770

($165M − $15.08M) ÷ 65,977.97 ≈ $2,272. ($158M − $26.87M) ÷ 2,000 ≈ $65,600. Both entries sit a few percent above the current tape's 52-week lows — $1,507 for ETH, $57,770 for BTC — and both are in the money now: ETH at roughly $2,487 is up about 9% on its entry, BTC at roughly $79,300 is up about 21%. The same read was cross-checkable on August 22, when the cluster closed 40,000 ETH near $2,513 for a realized ~$9.9 million — an entry of about $2,266 implied by that print.

This cluster is not new. On February 10, two addresses this tracker follows — 0x6c8…d84f6 and 0xa5b…01d41 — were the #1 and #2 largest ETH longs on the Hyperliquid perp venue, roughly 95,000 ETH combined, one tranche a 20x leveraged position opened around $2,044. It kept adding through spring, sat through the stretch when ETH open interest shed $2 billion in a week and funding went negative, and absorbed weeks of being underwater. On August 22 it did the thing it had not done in about four months: sold a tranche into strength, then immediately had another address in the cluster start re-adding, roughly 9,000 ETH and counting. Three addresses then held about 59,000 ETH with $8.73 million of floating profit. Two days later, the tracker consolidates the count to ten addresses and $41.95 million.

So which story is it? Two readings, and they split on one question.

Reading one: the low got absorbed. A bucket that bought at $2,272 / $65,600 while the tape bottomed at $1,507 / $57,770, then held through a 30–58% rally without distributing into either pop, and took its first profit as a partial that was immediately reloaded, is behaving like a position that believes the move. In a greedy, BTC-led tape, that is evidence the marginal seller ran out of inventory.

Reading two: the harvest ladder. Unrealized profit is hindsight, not a plan. The same cluster now holds $42 million of reasons to sell into the next strong session, and the August 22 close into strength was the first rung of a ladder — with the re-add acting as the buyback between rungs. Same wallet history, opposite forward options. Nothing on-chain has decided between them yet.

State the verification limits before the checklist, because they bind how you can use this. "bit-affiliated" is a tag on an analyst's tracker, not a filing; only two of the ten addresses appear in public write-ups, and both are truncated. Second, $323 million of "longs" is notional exposure. These are Hyperliquid perpetual positions, at least one of them 20x and margined by USDC, so paid-in capital is a fraction of the headline — and floating profit on a levered book is larger, and disappears faster, than the same dollars in spot. Label, sample, and basis all unverified. That downgrades this from "run tonight" to "watchlist," and the watchlist has four lines.

Watch the two addresses, not the press. Open the hyperbot trader pages for 0xa5b…01d41 and 0x6c8…d84f6. Note position size up or down, and distance to the liquidation line. Deleveraging without repurchase beats any headline.

Watch exchange balances, not news. Distribution prints as ETH and BTC flowing into known exchange hot wallets in size. The only flow series I can pull shows net outflows on most of the past week's sessions across Binance spot — coins drifting off exchanges — which is the mild accumulation read, not proof of it.

Watch funding on the perp venues. A book this size pushing funding persistently positive is paying to stay long into greedy tape; funding collapsing negative means shorts are the crowded side and a squeeze is live. Either extreme is information. Middle of the range is noise.

Watch the regime you would trade, not the whale. On current data, fear/greed reads 73, BTC dominance near 60%, altcoin season at 32. This is a BTC-led rebound; the ETH book in the headline is the lagging, more leveraged leg of the same position.

That is the observation. Here is why it stops being one. The "smart money caught the bottom" take expires the day price breaks the 20-day trend behind the current 23% (BTC) and 32% (ETH) 20-day rallies, the day the tracked addresses push material size to exchange hot wallets, or the day 0xa5b…01d41 starts closing the biggest ETH long on the venue instead of reloading it. Re-run the check before you trust the next screenshot: a float-profit headline with truncated addresses and an unverified label is the folklore, and the folklore is what costs money — the wallet is only evidence while you can actually open it.