The metric that keeps pulling me back in is deceptively simple. An accumulation address is a wallet that has only ever received coins and never sent a single satoshi out. No spends, no exchange deposits, no shuffling to a new address. Money goes in and stays in. Screen the whole chain for wallets that fit that pattern, sum what they hold, and you get a rough proxy for supply that has moved into hands showing no intention of selling. When the total held by these addresses is climbing, the working theory is that conviction is rising and float is quietly leaving the market.
I like it because it is measurable and honest about what it measures. It does not try to read minds. It just counts a specific on-chain behavior. But the same thing that makes it clean makes it fragile, and the failure modes are more interesting than the signal itself.
What the metric is actually counting
Two related numbers matter here. The first is the accumulation address balance I just described. The second, and the one I lean on more, is the illiquid supply ratio. Instead of a hard binary of spent versus never spent, this scores every entity on the chain by its historical ratio of outflows to inflows. Wallets that receive a lot and send almost nothing get tagged illiquid. Wallets that churn, mostly exchanges and active traders, get tagged liquid. The illiquid supply is the sum held by that first bucket, and the ratio is that figure over circulating supply.
The reason this framing is useful is that it turns a fuzzy narrative about diamond hands into something you can chart. When illiquid supply rises while price is flat or drifting, you are watching coins migrate from wallets that trade into wallets that historically do not. That is float shrinkage. The available-to-sell pool is getting smaller even if nothing dramatic shows up on the price chart yet.
The setup people care about is the squeeze. If illiquid supply keeps climbing through a period of steady or growing demand, the pool of coins actually available on exchanges thins out. At some point a demand shock, an ETF inflow day, a large spot buyer, whatever, hits a book that is thinner than it looks, and price moves more than the flow alone would suggest. The metric does not predict the shock. It tells you the powder is dry and the room is small.
The false positives that quietly wreck it
Here is where I have to slow anyone down who gets excited about a rising line. The raw metric cannot tell the difference between three very different kinds of wallet, and two of them are not bullish at all.
- Genuine holders. Someone bought, moved to cold storage, and is sitting on it. This is the case the metric is designed to catch, and it is the bullish one.
- Custodian and ETF addresses. When a spot ETF or a large custodian holds coins on behalf of thousands of clients, those coins sit in a small number of addresses that receive constantly and send rarely. To the raw metric that looks exactly like the deepest-conviction whale on the chain. It is not conviction. It is custody. And custodied supply can be redeemed and sold the moment shares are created and destroyed, so it is arguably more liquid than the metric thinks, not less.
- Lost keys. Early coins where the private key is gone, wallets whose owners died, funds sent to burn or typo addresses. These never move because they cannot move. They inflate the illiquid figure permanently and they will never come back to affect price in either direction.
Lost supply and custodian supply push the number in opposite directions from a usefulness standpoint. Lost coins are a static overstatement you can mostly ignore once you know it is baked in. Custodian coins are the dangerous one, because they grow over time as the ETF and custody world absorbs more of the chain, and they masquerade as the exact signal you are hunting. A chart of rising illiquid supply that is really just a custodian ingesting client deposits will look identical to a chart of genuine accumulation. Same slope, opposite meaning.
How I actually read it
I stopped treating the absolute level as the signal a while ago. The level is contaminated by lost keys and custody, and nobody agrees on how to clean it. What I trust more is the rate of change and, crucially, what is driving it.
A rough workflow that has kept me out of trouble:
- Look at the change in illiquid supply, not the level. A rising level tells you almost nothing on its own given how much dead and custodied supply is folded in.
- Ask which entities are driving the change. If a handful of known custodian clusters account for most of the rise, discount it hard. That is custody flow wearing a conviction costume.
- Cross-check against exchange balances. Genuine accumulation usually shows up as coins leaving exchange wallets at roughly the same time illiquid supply rises. If exchange balances are flat or growing while illiquid supply climbs, be suspicious that you are looking at reclassification rather than real withdrawal.
- Give it time. This is a slow-moving structural signal, not an entry trigger. A one-day spike is noise. A grind over weeks and months is the thing worth respecting.
- Never size a trade on this alone. It tells you the float is thinning. It says nothing about when demand arrives to exploit that.
The most common way I have seen people misuse it is treating a rising line as a buy signal by itself. Illiquid supply can rise for a long time while price does nothing, because a small float only matters when something pushes against it. You can be completely right about the supply setup and still sit dead money for months waiting for the demand side to show up. The metric is context, not a catalyst.
Where it earns its keep
For all the caveats, I keep it on the dashboard because it answers a question price alone cannot. Price tells you what the marginal buyer and seller agreed on today. Illiquid supply tells you how much of the total is even in a position to be that seller. When most of the supply has migrated into wallets that historically do not sell, the market is more reflexive. Good news moves it further, and so does bad news, because there is less inventory to absorb either.
The practical version is to pair it with something that captures demand pressure so you are not staring at half the picture. On Blockcircle I tend to overlay illiquid supply against exchange netflows and whale wallet movement on the same timeline, which makes the custody-versus-conviction distinction a lot easier to eyeball. When the supply is thinning for real and demand is turning up at the same time, that is the window the metric was built to flag. When only the supply line is moving, I note it and wait.
None of this makes the metric a crystal ball. It makes it a decent estimate of how tight the room is. Just remember that some of the people in that room are custodians who will leave the second their clients ask, and some are ghosts holding keys nobody will ever find again. Count them if you must, but do not mistake either one for conviction.