Why a 2-year rolling normalization matters
After 2017, exchange batching, SegWit, the Lightning Network and spot ETF custody allowed Bitcoin’s market value to expand without linear growth in daily Layer-1 address counts, which stabilized around 600K to 1M per day. An unadjusted static formula drifts upward over time, whereas normalizing over a rolling two-year window isolates genuine cycle overheating and undervaluation.
How NVM is computed here
- Take the 30-day simple average of Coin Metrics daily active BTC addresses (AdrActCnt) to smooth weekend dips.
- Compute the log Metcalfe ratio each day: ln(Market Cap ÷ 30D Active Addresses²).
- Standardize that value against its trailing 730-day mean and standard deviation to get the NVM Z-Score, and invert the mean to plot the implied Metcalfe baseline price.
Zones
- Below -1 (Undervalued): market cap has fallen much faster than active on-chain user activity.
- -1 to -0.3 (Below Trend): mildly cheap relative to two-year network usage.
- -0.3 to 1.2 (Fair Value): market valuation aligns with active address activity.
- 1.2 to 2.2 (Hot): price appreciation is outpacing user growth.
- Above 2.2 (Overheated): speculative excess; the 2013, late 2017 and spring 2021 tops all pushed above 2.5.
For on-chain fundamentals, I read NVM alongside MVRV Z-Score and NUPL: NVM tracks active user growth while MVRV and NUPL measure holder cost basis.
Coin Metrics counts unique Bitcoin mainnet addresses appearing as a sender or receiver in confirmed transactions on a given UTC day.
Yes. Temporary bursts of non-monetary transactions inflate active address counts and lift the implied Metcalfe baseline, so check whether address surges reflect sustained adoption.
MVRV Z-Score or NUPL: NVM tracks active network usage while MVRV tracks holder cost basis, covering both user fundamentals and positioning.
