KoinlyticsBitcoin on-chain flows produced two numbers this week that would normally signal a crisis. Over 32,000 BTC moved from short-term holders to exchanges at a loss on August 1, the largest loss-driven move by that cohort in 30 days. Total daily migration across all cohorts reached approximately 39,600 BTC, the biggest single-day movement since the FTX collapse in November 2022.
In 2022 those numbers would have meant holders fleeing exchanges for self-custody. This time the direction is reversed. Coins are moving toward exchanges, and the reason is a hardware wallet failure rather than an exchange failure.
What the Short-Term Holder Number Actually Measures
Short-term holder is a specific on-chain classification: coins that last moved within roughly the past 155 days. It approximates recent buyers rather than long-term accumulators, and it is the cohort most sensitive to drawdowns because their cost basis sits near current price.
When that group sends coins to exchanges while underwater, the standard interpretation is capitulation or defensive de-risking. Analysts weight loss-driven deposits more heavily than profit-driven ones, because selling at a loss requires a stronger motivation than taking profit does.
The important caveat, which gets dropped constantly in this kind of analysis: a deposit is not a sale. Coins arriving at an exchange address may be sold, may be used as collateral, may be moved between an investor's own accounts, or may simply be parked. Exchange inflow measures intent to have optionality, not executed selling.
The Coldcard Distortion
This is where the standard reading breaks down for the first week of August. The Coldcard firmware exploit, which by August 4 had confirmed losses of 1,596 BTC across more than 7,300 addresses, sent a wave of unaffected users to move funds off devices they could no longer independently verify.
Those users had two destinations: a different self-custody setup, or an exchange or regulated custodian. A meaningful share picked the second. That produces exchange inflow that looks identical to capitulation in the raw data and means something completely different.
The distinguishing signals worth checking:
- Whether inflows are matched by a corresponding rise in exchange sell-side depth, which selling requires and custody migration does not
- Whether the depositing addresses cluster around the known Coldcard address footprint
- Whether price actually broke on the flow, which it did not: Bitcoin held $60,000 and closed the week near $64,000
- Whether small-denomination transfers spiked, which they did, hitting a four-year high, consistent with individual users consolidating rather than institutions repositioning
The Cleanest Evidence Is the Price
32,000 BTC of genuine loss-driven selling into a market doing $22 billion to $26 billion of daily spot volume would leave a mark. It did not. Bitcoin traded $63,565 on August 4 and $64,037 on August 5, up on both sessions, while US spot ETFs took $170.3 million of net inflows.
Supply arriving at exchanges that does not move price means demand absorbed it. That is the definition of a market with a bid underneath, and it is a materially more constructive reading than the raw inflow number suggests in isolation.
Where the Genuine Weakness Is
None of this means the on-chain picture is clean. There is real short-term holder stress in the data. That cohort is underwater in aggregate, which is what makes a deposit at a loss possible in the first place. A cohort sitting on unrealised losses is a standing supply overhang that gets released on any sustained move higher, because breakeven is where recent buyers exit.
The practical shape of that: rallies into the short-term holder cost basis meet resistance from people getting out flat, not from anything technical. That level currently sits in the low-to-mid $60,000s, which is precisely the range that has capped every move since the start of the month.
The largest daily Bitcoin migration since FTX happened because people stopped trusting a device, not because they stopped trusting a market. Same data, opposite meaning, and the difference only shows up if you check what price did.
What to Watch
- Exchange reserve totals over the next two weeks, and whether the inflow reverses as Coldcard migrations complete
- Short-term holder realised price, currently the effective ceiling on rallies
- Small-transfer counts returning to normal, which would mark the end of the custody migration wave
- Whether ETF inflows continue absorbing the on-chain supply or step back after Friday's jobs report
On-chain data is genuinely useful and genuinely easy to misread. A number that historically meant one thing can mean the opposite when the surrounding event changes. Checking the flow against the price is the cheapest sanity test available, and it is the one most commentary skips.