New Addresses

New Addresses plots the daily count of Bitcoin addresses that appear on the ledger for the first time. The thicker line is the 90-day simple moving average — the long-horizon trend signal that smooths through inscription waves and exchange operational rhythms. The lighter underlay is the raw daily count, useful for surfacing single-day events the SMA otherwise hides. Bitcoin’s USD price runs muted on the log left axis for cycle context.

Today’s daily count is 255,910; the 30-day average is 287,433; the 90-day average is 289,291; the year-over-year change on the 90-day SMA is −7.6%. The series carries 5,779 daily points from 17 Aug 2010 through to the most recent close on 13 Jun 2026.

For each daily close t:

NA(t) = | { addr : addr appears on chain for the first time on day t } |
sma30(t) = mean(NA(t−29) ... NA(t))
sma90(t) = mean(NA(t−89) ... NA(t))
yoy(t) = sma90(t) / sma90(t−365) − 1

First-touch means the address has never previously appeared in any confirmed transaction — either as input or output. The first daily print of an address counts toward NA on that day; every subsequent transaction contributes to active addresses but not to NA. The 90-day window is the primary trend smoother because the raw daily series has substantial weekly periodicity from exchange-side wallet rotations.

The regime bucket compares the direction of two 90-day SMAs — price and new addresses — over the trailing 60 days. Adoption divergence fires when price is rising but the on-boarding SMA is falling; Accelerating fires when both rise together, or when price falls while on-boarding keeps rising; Normal covers everything else (including any window where one SMA is roughly flat). The year-over-year change shown above is a separate read-out, not the regime threshold. Reconstruction details — including the receive-address-rotation caveat that affects every first-touch dashboard — live on the methodology page.

Three lenses, in increasing analytic value. Level reading. The 30-day average against the 200–500k range that has bracketed post-2018 prints — below 200k flags deep-bear or post-Runes settling; above 500k flags inscription-wave activity, not adoption. YoY reading. The headline regime bucket; useful as a slow momentum filter. Cycle-low lead reading. The dated table below makes the on-boarding floor reversal explicit — the signal that has historically preceded each cycle price low.

Reading every cycle anchor against the live series surfaces the on-boarding pattern. The 2017 cycle top printed sustained daily values above 500k; the 2021 cycle topped twice with daily prints in the 350–560k band (the April peak pushed past 550k, the November peak printed lower); the 2024 cycle decoupled materially from earlier patterns — the daily underlay shows the inscription-era volatility, while the 30-day SMA stayed in a narrower band. Prices are our own daily closes; counts are the daily and SMA values from the underlying series.

The 30-day SMA of new addresses has bottomed before each Bitcoin cycle price low on the daily-close record. The lead is the page’s distinctive read — the signal that on-boarding momentum has begun to recover ahead of the underlying price. The table below walks each cycle low, finds the local minimum of the 30-day SMA in the surrounding window, and reports the lead in days.

Three observations carry through the lead table. First, every cycle low on record was preceded by a 30-day SMA bottom in new addresses. The lead has averaged about 6.4 months across the four cycles in the table, with no cycle where the SMA bottomed after the price. The signal is not actionable on a daily horizon — the SMA bottom date itself is only knowable retrospectively, after enough subsequent prints have confirmed the trough — but it is a reproducible regime confirmation that the on-boarding floor has been set.

Second, the magnitudes are diverging from prior cycles. The 2015 SMA bottom printed near 79,730; the 2018 bottom near 255,460; the 2022 bottom near 381,175 — each cycle’s on-boarding floor has been higher in absolute terms but lower as a fraction of the prior cycle’s peak. The addressable-bound side of this argument is the same one driving the active-address decoupling — block weight is fixed, transaction shape is roughly fixed, address counts saturate around what each block can contain.

Third, the post-2023 environment introduces a different distortion. Inscription protocols — Casey Rodarmor’s January 2023 Ordinals release and the Runes launch at the April 2024 halving block — produce multi-day spikes of single-use receive addresses that inflate the daily count without representing first-time human on-boarding. The 30-day SMA absorbs most of the spike but not all of it; readings during inscription-active windows should be interpreted alongside the transaction count to separate genuine on-boarding from protocol activity.

A new address is not always a new user. HD-wallet software rotates a fresh receive address per incoming transaction by default; an exchange mints a fresh deposit address per customer; a custodian rotates addresses for operational hygiene. Each pattern inflates the count without adding a participant. The inflation is structural and persistent; cross-cycle comparisons lean on the SMA direction, not the absolute level.

Inscription protocols broke the level outright. The Ordinals software release on 21 January 2023 opened a wave of single-use receive addresses; the BRC-20 mania of April–May 2023 pushed daily prints sharply higher in a way that had nothing to do with adoption; and inscription activity later in 2023 (the autumn Ordinals and BRC-20 surge) drove the year’s peak daily prints. The Runes launch at the April 2024 halving block likewise inflated counts in its window. In each window the daily count is a protocol-activity gauge, not an on-boarding gauge. The 90-day SMA absorbs most of the spike but not all of it — read against transaction count and fee revenue to separate the components.

Layer-2 first-touch is invisible. Lightning Network channel opens are base-layer transactions and do print as new addresses; subsequent Lightning routing inside a channel does not. Sidechain (Liquid, Rootstock) activity does not appear here at all. As Layer-2 share grows, base-layer first-touch under-represents true network on-boarding by a widening gap.

The pre-2014 history is sparse. Daily first-touch counts in the 2010–2013 era reflect a much smaller absolute base; cross-era comparisons should weight the structural growth-rate side, not just the headline-number ratio. The cycle-lead table starts in 2014 for that reason.

If you accumulate on a schedule, a confirmed bottom in the 30-day SMA — followed by 30 to 60 days of monotone-rising prints — has historically marked the start of a multi-month cycle recovery in on-boarding. Treat the SMA-bottom signal as a tactical accumulation accelerator, not a trigger; the lead has averaged about 6.4 months ahead of the price low, not measured in days.

If you are timing the cycle, the adoption divergence regime — price rising while the 90-day SMA contracts year-over-year — has historically marked the late-stage of cycles. Pair with active addresses and ETF flows for the post-2024 institutional read, where new participants arrive through a custodian rather than a self-custody address and never touch this count at all. Pair with transactions and fees to decompose inscription activity from genuine on-boarding.