Proximity MVRV explained: Glassnode's cost basis twist

Glassnode published a research note on October 20, 2026 that rebuilds two well-known on-chain ratios, MVRV and NUPL, around the coins whose purchase price sits closest to today's price. The piece was written by Vicky Yang, co-founder at Fountainhead Digital. This article looks at how the new proximity metrics are calculated, what they add to the classic versions and what they still cannot tell you.

What the Glassnode research shows

The starting point is a complaint about averages. According to Glassnode, market-wide cost basis measures are dominated by large pools of coins that rarely move. Those pools can drown out smaller groups of coins bought near the current price, which are the ones most likely to trade soon.

Short-term holder metrics, which only track recently moved coins, react faster. But the note argues they can miss large blocks held by longer-term investors when price approaches those blocks. The proposed fix keeps the detail of a full distribution while turning it into a single line on a chart.

The author summed up the goal this way: "These proximity signals focus on the supply most likely to transact next, not the supply that merely dominates the historical average."

What is MVRV, and what changes in the proximity version?

MVRV compares the market value of a coin with its realized value, which prices every coin at the moment it last moved on-chain. When the ratio is high, holders on average sit on large paper gains; when it is low, many sit near or below their cost. NUPL expresses the same gap as a share of market value.

The proximity versions swap the market-wide realized price for a new reference called the Proximity Realized Price. It is a weighted average of realized prices across slices of supply, where slices near spot count more. Glassnode then forms pMVRV and pNUPL from that reference, and adds a third gauge, Proximity Premium, which measures the gap between spot and the proximity realized price in units of recent volatility.

The metric: how the weighting is built

Everything rests on a Cost Basis Distribution. Instead of one average cost, it maps the price levels at which coins last moved. Glassnode notes that this full view is hard to compare from one day to the next, because it is usually read as a heatmap with many layers.

To compress it, the method cuts supply into 100 cost-basis quantiles, or equal slices ranked by purchase price. Each slice gets a weight that shrinks as its realized price moves further from spot. The distance is scaled by the 30-day rolling standard deviation of price, so the same gap counts for less in a volatile month.

Two indexes come out of this. The Proximity Support Index, PSI, is the weighted share of supply bought below spot. The Proximity Resistance Index, PRI, is the share bought above it. Both range from 0 to 1. The note cites two past episodes. In 2023, PSI (ema7) moved from 35.3% on August 24 to 52.8% on October 12 while price barely changed, from about 26.1k to about 26.7k. Later, between December 18 and January 25, the same smoothed index slid to 48.7%, down from 57.5%, with BTC near 85.5k and then 86.6k.

Glassnode also ran regressions for BTC and ETH. They compared a model using only past returns with one that added a single on-chain metric, such as pMVRV, Proximity Premium or STH-MVRV. According to the piece, the proximity metrics added more explanatory power than short-term holder metrics, and Proximity Premium passed a 1% significance test for both assets.

Limits: where the signal can mislead

A cleaner line is not the same as a cleaner truth. Several blind spots carry over from the raw data, and some are new.

• Cost basis is inferred from when a coin last moved on-chain. Coins that change hands on an exchange's internal books, or move between a holder's own wallets, can make that price wrong.
• The weighting choices, including the number of slices and the volatility scaling, are design decisions. Different settings would draw a different line from the same chain data.
• The author notes that derivatives markets add synthetic BTC exposure that never touches the chain. That demand can dilute what spot holders do, and cost basis data cannot see it.
• Support and resistance here are labels for where coins were bought, not walls. The note itself treats them as potential levels only, and the historical examples were chosen by the author to illustrate the method.

Regression results describe the past. A metric that explained returns over a test period can lose that link when market structure shifts, and the note points readers to Glassnode's own AlphaLab tool rather than to independent testing.

What to watch next

For readers, the useful habit is to ask which cost basis a chart is built on. A classic MVRV reading and a proximity MVRV reading can disagree because they average different coins, not because one is broken.

Watch whether other analysts start citing PSI or pMVRV, and whether they publish their weighting settings. A metric that only one provider can calculate is harder to check. Pair any reading with simpler data, such as realized price and holder age bands, before drawing conclusions.

Sources

• Glassnode, 20 Oct 2026

Drafted by AI from the research and on-chain sources listed, then published automatically once checks confirmed the figures against those sources and found no copied text. Not reviewed by a person. Metrics are explained, never turned into price calls.