Active Addresses

Active Addresses plots the daily count of unique Bitcoin addresses that transacted — either as the input or the output of a confirmed transaction — against the 30-day simple moving average that suppresses single-day operational noise. The thicker line is the SMA; the lighter underlay is the raw daily count, surfaced so single-day spikes from exchange reshuffles or inscription waves are visible rather than smoothed away. Bitcoin’s USD price runs muted on the log left axis for cycle context.

The unit is an address-set cardinality, not a transaction count and not a user count. An address transacting twenty times on the same UTC day adds one to the total; twenty users sharing one exchange hot wallet also add one. Today’s daily count is 557,329; the 30-day average is 632,120; the year-over-year change on that average is −16.6%. The series runs from 17 Aug 2010 through the most recent close on 13 Jun 2026.

For each daily close t:

AA(t) = | { addr : addr appears as input or output of a confirmed tx on day t } |
sma30(t) = mean(AA(t−29) ... AA(t))
yoy(t) = sma30(t) / sma30(t−365) − 1

An address transacting twenty times on the same UTC day counts once. Coinbase issuance rows with no sender address are excluded. The 30-day window is the conventional smoothing choice — long enough to suppress operational single-day spikes (an exchange consolidating sweeps, a wallet provider rotating its receive prefix), short enough to retain cycle-length resolution.

The regime bucket comes from the year-over-year change on the 30-day SMA: Expansion above +30%, Contracting below −10%, Stable in between. Reconstruction details, including the address-clustering caveats that affect every active-address dashboard, live on the methodology page.

Reading every cycle anchor against the live series surfaces three eras. Pre-2018 cycle tops printed dramatic SMA peaks; the 2021 cycle topped twice with the address set already flattening relative to 2017; the 2024 cycle decoupled fully, with the Runes launch event spike clearly visible in the daily underlay even though the SMA continued to drift. Prices are our own daily closes; counts are the raw daily and 30-day-SMA values from the underlying series.

The cleanest way to see the chart’s shifting meaning is the per-cycle peak 30-day SMA together with Bitcoin’s closing price on the day of that peak. Each cycle’s “USD-per-active-address” is the price divided by the SMA — the implied price growth per unit of address-set activity. The ratio rises by an order of magnitude across cycles; the address set itself does not.

First, the peak 30-day SMA stopped growing after 2021. The 2017 peak SMA, the 2021 dual peaks, and the 2024 pre-halving window all printed in a tight band of roughly 900k to 1.2M addresses per day — despite price tripling between cycles. The addressable bound is set by block weight and average transaction shape, not by adoption. There is a hard ceiling on how many distinct addresses can confirm in 144 blocks a day, and the network has been near it at every top since 2017.

Second, USD-per-active-address has risen by roughly an order of magnitude across cycles. The ratio is sub-dollar at every top — the address set is large relative to spot — but it climbs cycle over cycle: each top assigns several times more market value to the same unit of base-layer address activity. Read the rightmost column of the table directly above; the figure rises from a cent at the 2017 top to several cents at the most recent peak. The chart records this as price-axis-and-address-axis divergence; the table states it as one number per cycle.

Third, the post-2023 window introduced a different kind of distortion. Inscriptions — beginning with Casey Rodarmor’s Ordinals protocol, whose genesis inscription was minted in December 2022 before activity exploded in early 2023 — and Runes — the token protocol that launched at the April 2024 halving in block 840,000 — both consume base-layer block weight without growing the unique-address footprint. Many inscription transactions reuse a small handful of inscriber and marketplace addresses for thousands of mints. Address counts compress while transaction counts and fees rise; the active-address chart diverges from the transaction count and the fee revenue charts in ways no prior cycle exhibited. The chart still measures something — the address-set side of base-layer activity — but reading it as a unitary “adoption” signal in 2026 is a category error.

Every active-address dashboard inherits the same four failure modes. This series is no exception — what follows is each one with the specific window it most visibly distorts the chart.

Inscriptions broke the adoption read — Jan 2023 onward. The Ordinals genesis inscription was minted on 14 December 2022, but the protocol’s public release in January 2023 is when inscription activity exploded; the Runes protocol launched in the halving block 840,000 on 20 April 2024. Both run thousands of mints, transfers, and settlements through a small set of inscriber and marketplace addresses. Across 2023–2024 the transaction-count and fee charts spiked into record territory while the active-address SMA drifted sideways or fell — the largest count-vs-throughput divergence in the chain’s history. Any page that read the falling address count as “declining usage” in this window read it backward.

The exchange-batching floor — visible at every cycle low. A custodial venue settles thousands of customer withdrawals by sweeping into one hot-wallet address; the more flow migrates to custody, the more users hide behind a single counted address. This is why the 30-day SMA could trough near 600k at the 2018–2019 low and near 890k at the November 2022 post-FTX low while real holder counts were rising — the activity moved on-exchange, off the address ledger. The bias deepened again after the January 2024 spot-ETF launches concentrated buy-side flow into a handful of custodian addresses.

Layer-2 is structurally invisible — and growing. Lightning Network payments, Liquid and Rootstock sidechain transfers, and federated-bridge activity never touch the base layer, so none of it registers. Lightning alone routes thousands of BTC of economic flow annually that this chart cannot see by design. As retail payments migrate to those rails, the count will increasingly understate usage with no way to correct for it from on-chain data.

The pre-2014 history is not comparable — before ~2013. Address-set telemetry on the first three years of the chain reflects far smaller absolute participant counts and far thinner exchange infrastructure. The 2013 peak in the table sits below 200k per day; comparing it to a 2024 reading is comparing two different network topologies, not two points on one trend. Weigh structural growth, not headline ratios, across that boundary.

If you are sizing the cycle, read the YoY regime as a slow momentum filter, never as a destination. Expansion confirms participation is accelerating; Contracting is consistent with bear stretches and with the benign post-inscription decoupling, so it cannot stand alone. Cross-check against ETF flows and exchange flows for the institutional lane the address count is now blind to.

If you operate or build on the base layer, the level reading is the useful one: a sustained 30-day average near or above one million addresses per day means block weight is being fully bid for and fee pressure is likely. Pair the count with daily transactions and daily fees to separate capacity-bound regimes from demand-bound ones — when transactions and fees climb but addresses do not, you are watching inscriptions, not new users.