NVT and settled-value ratios, and how they are known to fail

Network value to transactions divides a network's market capitalization by the value of transfers settled on it over a period. It was introduced as a rough analogue to a price-to-earnings ratio for networks that had no earnings, and it fails in a specific, well-documented way: the denominator counts ledger movement rather than economic activity, and the two diverge for reasons that have nothing to do with adoption.

The metric is published here as NVT, with a narrower construction at market cap to settled volume.

The construction and its original argument

The formula is market cap ÷ on-chain transfer value over a period. The reasoning was that a monetary network's usefulness lies in moving value, so the quantity of value it moves is the closest thing available to output. Dividing valuation by output produces a multiple. That framing was more careful than the ratio it inspired: it was offered as a descriptive tool for a young asset class, not as a fair-value model with a target level.

The analogy to an earnings multiple breaks at the first step. Earnings accrue to shareholders. Transaction volume accrues to nobody. A dollar moved across a network is not a dollar the network or its holders received, and moving the same dollar a second time creates nothing. The denominator is throughput, and throughput has no owner.

What the raw denominator actually contains

• Change outputs. On ledgers that track unspent outputs, spending part of a balance sends the remainder back to the sender. A naive sum counts that remainder as transferred value, which can inflate the total by a large multiple.
• Internal custodial movement. Custodians shuffle balances between hot and cold storage on operational schedules. These are ledger entries with no counterparty and no economic event behind them.
• Token transfers on the same chain. A stablecoin transfer moves value across a network, and that value is a claim on the stablecoin issuer, not on the network's native asset. Including it credits the chain's asset with someone else's economic activity.
• Batching and consolidation. Operators periodically merge many small balances into fewer large ones, producing enormous transfer totals on days when no user did anything new.
• Automated flows. Bots, MEV strategies and arbitrage loops move value repeatedly through the same capital, sometimes many times within a single block.

The standard response is an adjusted series, which strips change outputs, known exchange-internal addresses and identified self-transfers. Adjustment improves the number and introduces a dependency: the result now rests on an address-labeling database maintained by a data provider, whose coverage is incomplete, whose heuristics are usually undisclosed, and whose historical labels change as new clusters are identified. Two adjusted series for the same asset can differ materially and both be defensible.

Settlement moving off the base layer

The structural failure is the same one that affects address counts. As transactions move to a layer 2 or are settled internally by exchanges, the base layer's transfer value falls while the economic activity it supports does not. A rollup posting compressed data to a base chain generates fees and settlement demand while contributing almost nothing to the base chain's transfer-value total.

An NVT series that spans such a transition is comparing two different measurement regimes and presenting the difference as a change in valuation. Nothing in the formula flags the break, and nothing in the chart shows where it occurred. This is the main reason the ratio is presented here alongside settled volume and fee series rather than on its own.

The non-stationarity problem

Ratios are only usable against a reference. Equity multiples have a reference in the cost of capital and in a long record of realized outcomes across many companies and cycles. NVT has neither. Its level has drifted substantially across eras as the composition of on-chain activity changed, as custody centralized, as stablecoins came to dominate transfer value on several networks, and as scaling layers absorbed everyday transactions.

A band derived from one era therefore says little about another. Variants that smooth the denominator with a moving average, or that generate signals from crossings of such an average, address the noise and leave the drift untouched. Where such a variant appears to have worked historically, the fitting was done on a short and non-repeating sample, and survivorship bias in which assets remain available to test on makes the record flattering in a way that cannot be corrected after the fact.

What survives

Three narrow uses hold up. As a description, the ratio states how much market capitalization sits above each dollar of settled value, which is a fact about a network's current composition. As a comparison across networks with genuinely similar settlement behavior, it can separate systems whose valuations rest on transfer activity from those whose valuations rest on something else. As a divergence flag, a large move in the ratio invites the question of which side moved and why, which is a research prompt rather than a finding.

Adjacent measures often do the job better. Market cap to fees uses a denominator that someone actually paid, which is harder to inflate than value merely moved, because inflating it costs money. Realized capitalization reprices supply at the value it last moved at, describing aggregate cost basis rather than throughput. Turnover compares traded volume to market cap and describes market activity rather than settlement, which is a different question again.

Where all three are read together, the pattern of agreement and disagreement between them is more informative than any single level. A ratio that has moved while the others have not is usually revealing something about measurement rather than about the network.

The network activity pages carry the settled-value series and state which adjustments were applied, and the data sources pages identify who produced each labeling set.