The basic idea: coins remember when they last moved
On a transparent blockchain, every unspent output has a timestamp. The moment a coin sits in a wallet without being spent, the clock starts, and it keeps running until that coin moves again — sent, sold, swapped, whatever. This is not true of a bank balance, where the dollars in your account have no individual history. It is true of Bitcoin and most UTXO-based chains, and it is approximately true of account-based chains like Ethereum if you track balance changes instead of individual coins.
That timestamp is the raw material for an entire category of on-chain analysis. If you can tag every coin in circulation with "time since it last moved," you can bucket the entire supply by age and watch how the buckets change over time. That bucketing is the foundation for HODL waves and, from a different angle, coin days destroyed.
Neither metric tells you anything about price directly. Both tell you something about the psychology and positioning of the people holding the asset, which is a different and slower-moving signal.
HODL waves: supply sorted into age bands
A HODL wave chart takes the entire circulating supply and splits it into age brackets — coins that moved in the last 24 hours, coins last moved 1–3 months ago, 3–6 months, 6–12 months, 1–2 years, 2–3 years, 3–5 years, 5+ years, and so on. Each bracket is expressed as a percentage of total supply, and all the brackets are stacked into one chart that always sums to 100%.
Visually it looks like a river of colored bands stacked on top of each other, moving left to right through time. The bands expand and contract as coins age forward into older brackets or get "reset" back to the youngest bracket by moving.
A few mechanical points worth being precise about, because they trip people up:
- A band growing doesn't mean coins are flowing in. A coin doesn't have to be bought or sold to migrate from the "1–2 years" band to the "2–3 years" band — it just has to keep sitting still while a year passes.
- Any movement resets the clock. If a wallet that's held coins for four years sends them anywhere — including to another wallet it controls — the coin drops back into the youngest bracket. The chain has no way to know if that was a sale, a security migration, or a custodial shuffle.
- The bands are supply, not holders. A whale moving a large single-address balance can shift a band noticeably more than thousands of small retail wallets combined.
What the pattern looks like across a cycle
The textbook cycle pattern is fairly consistent across the assets liquid enough to track this way:
During strong bull markets, the older bands shrink and the young bands grow. This is long-term holders selling into strength — coins that had been dormant for years suddenly move, get sold or redistributed, and reset into the youngest bracket. At the same time new buyers are entering, trading actively, and adding fresh supply into the sub-3-month bands. Visually, the chart's bottom (old) bands compress and the top (young) bands swell.
During bear markets and the accumulation phases that follow, the reverse happens. Older and dormant bands grow, both because existing holders have little incentive to sell at depressed prices and because time simply keeps passing — coins that aged into the "6–12 month" band during the bear market's early stretch keep aging into "1–2 years" and beyond as the bear market drags on. Active trading volume drops, so the young bands thin out.
The pattern is genuinely useful as a rough proxy for where a cycle stands structurally: a chart dominated by young, actively-moving supply looks like a hot, distributive market; a chart dominated by thick old bands looks like a market that's been left alone for a while. Where it breaks down is in timing — HODL waves tell you about positioning, not about when positioning will change. A market can sit in "old and dormant" for a long stretch before anything happens.
Coin days destroyed: weighting movement by dormancy
HODL waves answer "how is the supply distributed by age." Coin days destroyed (CDD) answers a narrower question: "how much old-coin movement happened today, weighted by how old that coin was."
The calculation is simple. Every unspent coin accumulates "coin days" for every day it sits still — one coin held for 40 days has accumulated 40 coin-days. When that coin finally moves, it "destroys" all the coin-days it had built up. CDD for a given period is the sum of (amount moved × days since last moved) across every transaction in that period.
The effect of the weighting: a transaction moving 10,000 coins that had sat still for 5 years destroys far more coin-days than a transaction moving 500,000 coins that were bought yesterday and sold today. Raw transaction volume would treat the second transaction as the bigger event. CDD treats the first one as vastly more significant, because it represents genuinely old, previously-committed supply re-entering circulation.
This is what makes CDD useful as a spike detector. A single wallet that has been silent for years suddenly emptying out will barely register in transaction count or trading volume, but it produces a visible spike in CDD because the dormancy multiplier is so large. Analysts use these spikes as a flag — not a verdict — that unusually significant old supply just became active and is worth investigating further: is it a known exchange cold wallet, an early miner, an estate settlement, a lost-and-found key.
Interpretation traps, stated plainly
Both metrics are honest about what happened on-chain and unreliable about why it happened. A few specific failure modes:
- Passive aging looks identical to conviction. A growing 3–5 year band could mean holders are deliberately refusing to sell. It could also just mean that a large cohort of coins bought during a specific period is aging forward together, with zero active decision-making involved. The chart cannot distinguish "I am choosing to hold" from "nothing has happened to this coin in a while."
- Exchange and custodial reshuffling distorts both metrics the same way it distorts MVRV. When an exchange consolidates cold storage — moving coins between its own wallets for security or operational reasons — every one of those coins resets to age zero and, if the batch is large and the coins were old, produces a CDD spike. Nothing about actual market supply or holder sentiment changed. This is the same blind spot that makes MVRV's realized value calculation noisy around known custodial address activity.
- New supply issuance mechanically inflates the youngest band regardless of demand. For proof-of-work assets, newly minted coins always enter in the "moved in the last 24h" bracket. A period of low trading activity elsewhere in the market can still show a nonzero young band purely from issuance.
- Address clustering is imperfect. Chain analytics firms attribute addresses to entities using heuristics, not certainty. A CDD spike attributed to "a long-dormant whale" might actually be an exchange relabeling internal wallets, and that distinction usually isn't resolved until well after the fact, if ever.
None of this makes the metrics useless. It means every reading needs a second look at what specifically moved before it gets treated as a behavioral signal.
A grounded way to use this
Treat HODL waves and CDD as structural, cycle-scale instruments, not entry or exit tools. They move on the order of weeks and months, not candles. If you're building or reading a composite score that blends on-chain, price, and momentum inputs, this is the category that belongs in the slowest-moving component — a confirming layer on where the cycle sits, not a trigger.
Practically: check the HODL wave chart maybe once a month, not daily. Look for the balance point between the young bands (sub-6-month) and the old bands (2+ years) — a market with young bands under roughly 15–20% of supply and old bands dominant has historically looked more like late-accumulation than late-distribution, and the reverse configuration has looked more like a market getting long in the tooth. Treat any single CDD spike as a research prompt, not a conclusion — pull the transaction, check if the source address is a known exchange or custodian, and only then decide if it represents a real behavioral event or routine wallet housekeeping.