Miners are the one group of holders whose selling you can partly predict, because they have to sell. Everyone else on-chain is optional. A long-term holder can sit through a 60 percent drawdown and never touch the keys. A miner has a power bill denominated in fiat, hardware that depreciates whether it hashes or not, and in a lot of cases debt against those machines. So their coins carry a built-in exit motive, and that makes their wallets one of the more honest sources of forward sell pressure you can watch. The trick is that the honesty is easy to over-read. Miner flow is a real signal, but it is a small share of daily volume, and the way it interacts with price has changed as the market got deeper.
Reading miner-to-exchange flows without fooling yourself
The base observation is simple. Miners earn coins at the protocol level, they hold them in wallets that are reasonably well labeled by the on-chain data providers, and when they move those coins to an exchange deposit address, that is usually a prelude to selling. So a rising miner-to-exchange flow is a rough proxy for miners choosing to convert block rewards into cash rather than sit on them. When the flow spikes, you are watching treasuries drain.
Where people go wrong is treating every outflow as a sale. Miners move coins for a lot of reasons that have nothing to do with dumping on the market. They rebalance between custody providers, they post collateral for loans so they can pay bills without selling, they consolidate wallets, and large public miners route coins through OTC desks that never touch a visible exchange order book. So the on-chain print can look like aggressive selling when it is really a miner borrowing against a coin it fully intends to keep. The direction of the flow is a hint, not a confession.
A few habits keep this honest:
- Watch the trend in miner reserves, not single transactions. A single 5,000 coin move is noise. A reserve that grinds lower over weeks is a treasury being run down, and that is the part that matters.
- Separate the two regimes. Miners accumulating into a reserve while price is flat usually means they can afford to hold, which is a quiet bullish tell. Miners drawing reserves down hard into weakness is the stressed version, and it tends to cluster near local pain rather than tops.
- Normalize by issuance. After a halving, the daily coin subsidy drops by half, so the same dollar amount of selling shows up as a bigger fraction of new supply. Raw coin counts understate the pressure post-halving.
The Puell Multiple and revenue stress
The Puell Multiple gets at the same question from the income side instead of the flow side. You take the daily coin issuance valued in dollars, so block rewards times price, and you divide it by its own roughly one-year moving average. That ratio tells you whether miners are earning a lot or a little relative to their recent normal. High readings mean revenue is running hot, which historically shows up near euphoric price extremes when miners are flush and inclined to sell into strength. Low readings mean revenue has collapsed relative to trend, which is the stress zone where the marginal miner is underwater and either capitulating or about to.
The useful mental model is bands rather than a precise line. When the multiple sits in its lower band, miner income is compressed, weaker operators are shutting machines off, and the network is grinding through a revenue recession. Historically those low-band stretches have overlapped with major price bottoms. Not because the Puell Multiple causes the bottom, but because the same conditions that starve miners also mark the point where sellers are exhausted. When the multiple sits in its upper band, you are usually late in a run and miners have every incentive to distribute.
Two cautions I would not skip. First, the metric is mechanically tied to price, since price is in the numerator, so it is partly a slow moving valuation gauge wearing a mining costume. Do not treat it as independent confirmation of a price signal it is partly derived from. Second, the halving warps it on a schedule. Issuance halves overnight, so the numerator drops in a step function while the one-year average catches up slowly, and the ratio prints artificially low for a stretch after each halving. If you see the Puell Multiple crater the same month a halving lands, that is arithmetic, not capitulation. Read it a beat later once the moving average has digested the new subsidy.
Why capitulation marked bottoms, and what changed
The historical pattern that made these metrics popular is miner capitulation. It goes roughly like this. Price falls, revenue falls with it, the least efficient miners run at a loss, hash rate rolls over as machines go dark, and those stressed miners sell whatever reserves they have left to survive. That forced selling is the last leg of a bottoming process, because once the weak hands are flushed and the difficulty adjustment lowers the cost bar for survivors, the sell pressure that was structural becomes optional again. Low Puell readings and drained miner reserves tended to cluster right there, which is why they earned a reputation as bottom-finders.
What changed is depth. The daily flow miners can produce is a roughly fixed quantity of new coins, and after each halving it is a smaller quantity. Meanwhile the buy side got institutional. Spot ETFs, corporate treasuries, and structured products added a layer of demand that absorbs coin flow the old market could not. So the same amount of miner selling now lands into a book that is far thicker than it was a cycle or two ago. The signal did not disappear. Miners still sell, reserves still drain, and the Puell Multiple still ranges through the same stress bands. What muted is the price impact per coin. A miner capitulation that once moved the market several percent can now get quietly eaten by a single day of ETF inflows, so the on-chain print looks the same while the price reaction is smaller and slower.
The practical consequence is that these metrics are better as regime context than as triggers. They tell you which side of the cycle you are on and whether miners are in a position of comfort or stress. They no longer reliably time the turn on their own, because the thing that used to make miner selling matter, thin liquidity, is much less thin.
A workflow I actually trust
Here is how I fold this into a read rather than trading it blind. I start with the Puell Multiple to place the cycle, low band means stress and probable value, high band means distribution risk, and I mentally discount any reading in the first weeks after a halving. Then I check the miner reserve trend over weeks to see whether treasuries are building or draining, which tells me the direction of the pressure. Only then do I look at miner-to-exchange flows for confirmation, and I treat a spike as a question rather than an answer, asking whether it lines up with a stress regime or whether it is more likely a collateral or custody move. If all three agree, stress readings on Puell, draining reserves, and rising exchange flow, I weight the signal more, but I still expect the price reaction to be softer than the textbook because of ETF-era depth.
The failure mode to avoid is the isolated outflow alert. Someone sees a labeled miner wallet push a large sum to Binance, posts a chart, and calls a top. Nine times out of ten that transaction is a loan, a rebalance, or an OTC leg that never pressures the book. Wire your alerts to reserve trend and flow persistence, not single transactions, and you will skip most of the false alarms. In Blockcircle I keep miner reserves, exchange flow, and the Puell bands on the same view precisely so I am forced to read them together instead of reacting to one line. Miner behavior is still worth watching. It is just a slower, quieter tell than it used to be, and it rewards patience more than reflexes.