On-Chain Bitcoin Valuation: Realized Cap, MVRV, and Beyond
On-chain valuation models like realized cap and MVRV ratio offer data-driven Bitcoin analysis. Here's how to read them.
On-chain Bitcoin valuation models give analysts something traditional markets rarely offer: a transparent, real-time view of every participant's cost basis. Metrics like realized cap, MVRV ratio, and NVT ratio transform raw blockchain data into signals about market psychology, miner health, and cycle positioning. Understanding how these models work, where they have succeeded, and where they break down is essential for anyone attempting data-driven Bitcoin analysis.
This guide covers the major on-chain valuation frameworks: what each one measures, the formula behind it, its historical track record at cycle extremes, and the limitations that can mislead analysts who rely on any single metric in isolation.
Realized Cap: The Network's Aggregate Cost Basis
Traditional market capitalization prices every bitcoin at today's spot rate. That approach treats coins last moved in 2010 the same as coins purchased yesterday. Realized capitalization solves this by valuing each UTXO at the price when it last moved on-chain, then summing those values across the entire supply.
Nic Carter and Antoine Le Calvez of Coin Metrics introduced the concept at the Baltic Honeybadger conference in September 2018. Carter described it as an on-chain volume-weighted average price: a measure of the aggregate capital that has flowed into Bitcoin, weighted by when each coin last changed hands.
How It Works
The formula is straightforward. For every unspent transaction output in the Bitcoin UTXO set, multiply the amount of BTC by the USD price at the time that UTXO was created. Sum those products across all UTXOs, and the result is the realized cap.
Lost coins matter: Coins that last moved in 2009 or 2010 are valued at fractions of a cent. Satoshi's estimated 1 million BTC contributes almost nothing to realized cap, even though it represents billions in market cap at current prices. This is a feature, not a bug: it filters out economically dormant supply.
What It Reveals
Realized cap behaves differently depending on market conditions. During bull markets, coins move at progressively higher prices, driving the metric steeply upward. During bear markets, selling occurs at losses, causing the realized cap to plateau or decline. By April 2025, Bitcoin's realized cap reached an all-time high of approximately $882 billion, according to CryptoQuant data. During the 2022 bear market, realized cap fell roughly 18.8% from its prior peak, reflecting a net capital outflow of approximately $88 billion.
This makes realized cap useful as a proxy for conviction. A rising realized cap during a price rally means new capital is entering at higher prices. A falling realized cap during a decline means holders are capitulating and realizing losses.
MVRV Ratio: Market Value vs. Realized Value
The MVRV ratio divides market cap by realized cap. Murad Mahmudov and David Puell introduced it in October 2018, building directly on Carter and Le Calvez's realized cap concept. The result is a single number that expresses whether the average bitcoin holder is sitting on unrealized profit or loss.
Reading the Signal
- MVRV above 1.0: the market values the network above its aggregate cost basis, meaning the average holder is in profit
- MVRV below 1.0: the market values the network below aggregate cost basis, meaning the average holder is underwater
- MVRV above 3.5: historically associated with overheated conditions and cycle tops
- MVRV approaching 0.8 or below: historically associated with capitulation and cycle bottoms
The logic is intuitive. When most holders are deep in profit, selling pressure increases as participants lock in gains. When most holders are at a loss, the remaining sellers are exhausted and accumulation begins.
Historical Performance at Extremes
| Date | Event | MVRV Level | Outcome |
|---|---|---|---|
| December 2017 | Cycle top | ~4.0 | 84% drawdown followed |
| December 2018 | Cycle bottom | ~0.75 | Marked the beginning of recovery |
| April 2021 | Local top | ~3.8 | 50% correction before second leg |
| November 2021 | Cycle top | ~3.0 | 77% drawdown followed |
| November 2022 | FTX bottom | ~0.82 | Marked the cycle low |
| October 2025 | Local top | ~2.5 | No classic top signal fired |
The October 2025 peak is noteworthy. MVRV topped at roughly 2.5, well below the 3.5+ readings that preceded prior cycle tops. As discussed in our analysis of whether the Bitcoin four-year cycle is dead, changing market structure from ETF inflows and institutional participation may be compressing the amplitude of on-chain signals.
The MVRV Z-Score
The MVRV Z-Score normalizes the gap between market cap and realized cap by dividing it by the historical standard deviation of market cap. The formula is: (Market Cap - Realized Cap) / StdDev(Market Cap). This standardization helps compare readings across cycles with very different absolute price levels.
Historically, Z-Score peaks have trended downward across cycles: roughly 9.3 in late 2013, about 6.3 in April 2021, and approximately 2.9 at the November 2021 top. This declining pattern has led some analysts to argue that Bitcoin's cycle tops are becoming less extreme as the market matures and liquidity deepens.
NVT Ratio: The PE Ratio of Bitcoin
Willy Woo introduced the NVT ratio in February 2017, framing it as a way to apply traditional valuation logic to a network with no corporate earnings. Instead of earnings, Woo used on-chain transaction volume as a proxy for the economic utility the network produces.
Formula and Variants
The basic NVT ratio divides Bitcoin's market capitalization by the total daily on-chain transaction volume denominated in USD. A high NVT suggests the network's valuation is running ahead of its actual usage, analogous to a stock trading at a high PE multiple.
The raw metric is noisy because daily transaction volume fluctuates heavily. Glassnode's NVT Signal addresses this by using a 90-day moving average of daily transaction volume in the denominator. Woo later proposed an NVT Price model in 2021 that uses a two-year median of the NVT ratio to derive a fair value estimate.
Interpretation
A rising NVT suggests speculative premium: price is climbing faster than on-chain usage justifies. A falling NVT suggests the network is undervalued relative to the economic activity flowing through it. However, Glassnode advises comparing NVT values only within cycles of similar maturity. Comparing readings from 2013 to those from 2021 is unlikely to produce meaningful signals because the network's base usage and participant mix have changed fundamentally.
Watch for distortions: Change outputs inflate volume figures because a transaction sending 1 BTC from a 10 BTC UTXO generates 9 BTC of change that registers as transaction volume. Batch transactions from exchanges also compress what would be many individual transfers into single on-chain events, deflating the count. These artifacts make raw NVT readings unreliable without adjustment.
Puell Multiple: Miner Revenue as a Cycle Signal
The Puell Multiple focuses on the supply side of the market. David Puell designed it to measure the profitability of bitcoin mining relative to its own history. The formula divides the daily USD value of newly issued BTC (block subsidies plus transaction fees) by the 365-day moving average of daily issuance revenue.
Miner Stress and Recovery
When the Puell Multiple is well below 1.0, miners are earning significantly less than their yearly average. This condition often coincides with miner capitulation: less efficient operations shut down, hashrate drops, and surviving miners sell reserves to cover operational costs. Historically, readings below 0.5 have aligned with deep bear market bottoms.
After the April 2024 halving cut the block subsidy from 6.25 to 3.125 BTC, the Puell Multiple declined steadily. It reached 0.64 in February 2026, reflecting acute revenue pressure on miners. By early October 2026, the metric climbed back above 1.0 (approximately 1.07), suggesting miners had begun earning above their trailing-year average again, according to CryptoQuant data reported by Crypto Briefing.
Strengths and Weaknesses
The Puell Multiple excels at identifying periods of extreme miner stress that correlate with market lows. However, it has limited utility for timing tops because elevated miner revenue can persist for months during sustained bull markets. It also becomes less predictable after halvings, when the baseline reset mechanically pushes readings lower regardless of price action.
Stock-to-Flow: A Cautionary Tale
The stock-to-flow model, popularized by the pseudonymous analyst PlanB in 2019, treats Bitcoin's scarcity as the primary driver of its price. The model divides Bitcoin's total existing supply (stock) by its annual production rate (flow). Because halvings cut the flow roughly every four years, the ratio doubles, and the model predicts a corresponding price increase.
Why It Gained Traction
Stock-to-flow's appeal was its simplicity and its apparent fit to historical data. From 2012 through 2021, Bitcoin's price roughly tracked the model's predictions across three halving cycles. The narrative was compelling: scarcity drives value, halvings reduce supply, and price follows mechanically.
Where It Broke Down
The model has faced substantial criticism, and its post-2024-halving performance has been poor. Vitalik Buterin argued in 2022 that the model gives a "false sense of certainty" and that the claim halvings cause price rises is unfalsifiable, since any price trajectory can be reinterpreted as confirmation. Nico Cordeiro of Strix Leviathan compared its predictive power to astrological models of financial markets.
The core problem is that stock-to-flow ignores demand entirely. It models only the supply schedule and assumes demand will scale to match. The 2022 crash, the post-2024 halving underperformance, and the changing role of institutional demand through ETFs have all exposed this limitation.
Comparing On-Chain Valuation Models
| Metric | What It Measures | Best Signal | Key Limitation |
|---|---|---|---|
| Realized Cap | Aggregate cost basis of all BTC | Capital flow direction | Does not generate buy/sell signals alone |
| MVRV Ratio | Profit/loss of average holder | Cycle tops and bottoms | Thresholds drift across cycles |
| MVRV Z-Score | Deviation from cost basis (normalized) | Extreme overvaluation/undervaluation | Requires sufficient history for calibration |
| NVT Ratio | Price relative to on-chain usage | Speculative bubbles | Change outputs and L2 distort volume |
| Puell Multiple | Miner revenue vs. yearly average | Miner capitulation (bottoms) | Mechanically reset by halvings |
| Stock-to-Flow | Scarcity (supply schedule only) | Long-term supply narrative | Ignores demand; poor recent accuracy |
Composite Indicators: Combining Metrics
No single on-chain metric captures the full picture. Institutional analysts increasingly build composite indicators that weight multiple metrics to reduce false signals. The general approach is to treat individual metrics as components that must confirm each other before triggering a signal.
Confluence-Based Analysis
A common framework checks for alignment across supply-side, demand-side, and behavioral signals:
- Supply-side: Puell Multiple, coin days destroyed, long-term holder supply changes
- Demand-side: NVT ratio, active address trends, exchange reserve outflows
- Behavioral: MVRV ratio, realized profit/loss ratio, SOPR (Spent Output Profit Ratio)
When multiple indicators across these categories align, the signal is stronger. For example, a Puell Multiple below 0.5 combined with MVRV below 1.0 and declining exchange reserves has historically preceded major recoveries. Conversely, elevated MVRV with rising NVT and accelerating exchange inflows has preceded corrections.
Institutional Dashboards
Firms like Glassnode, CryptoQuant, and CheckOnChain provide composite dashboards that aggregate multiple metrics into heat maps and scoring systems. VanEck publishes a monthly "ChainCheck" report that combines on-chain data with macro indicators for institutional clients. These tools have become standard for funds and trading desks managing Bitcoin exposure.
The Layer 2 Blind Spot
Every on-chain valuation model described above relies on data visible on Bitcoin's base layer. As more economic activity moves to Layer 2 networks, this creates a growing measurement gap.
What On-Chain Metrics Miss
Lightning Network transactions occur within payment channels and only appear on-chain at channel open and close. A channel could facilitate thousands of payments, but the base layer sees just two transactions. Similarly, protocols like Spark process transfers entirely off-chain through statechain key rotations: the Bitcoin doesn't move on-chain at all until a user exits to L1.
This creates specific distortions for each metric:
- NVT ratio is inflated because the denominator (on-chain transaction volume) undercounts actual economic activity happening on L2
- Realized cap understates activity because coins locked in Lightning channels or Spark leaves don't generate new UTXOs when they change hands
- Active address counts miss users transacting exclusively on L2 networks
- On-chain volume metrics cannot distinguish between organic decline in usage and migration to faster, cheaper layers
The ghost town effect: In mid-2025, Bitcoin processed an average of roughly 364,000 transactions per day, down 42% year-over-year according to VanEck's ChainCheck report. Glassnode described on-chain activity as a "ghost town." But simultaneously, ETF inflows were setting records and average transaction sizes grew significantly, suggesting large entities continued to use L1 for settlement while smaller transactions moved off-chain.
Adjusting for L2 Growth
Some analysts have begun experimenting with "L2-adjusted" versions of metrics like NVT that attempt to incorporate Lightning and sidechain volume data. However, this remains challenging because Lightning transaction volumes are estimated from gossip protocol data rather than directly observable. Spark's privacy model further complicates measurement, as transfers occur through cryptographic key rotations that leave no public trail until an on-chain exit.
The practical implication is that on-chain models may increasingly overstate Bitcoin's "valuation premium" as L2 adoption grows. An NVT reading that looked overheated in 2020 may simply reflect the fact that a growing share of Bitcoin's economic activity no longer registers on the base layer.
Known Pitfalls and Best Practices
Threshold Drift
The MVRV top zone has compressed from above 4.0 in 2017 to roughly 2.5 in 2025. Analysts who anchor to historical thresholds without adjusting for market maturation will mistime signals. The entry of ETFs, corporate treasuries, and institutional participants changes how far above (or below) cost basis the market is willing to extend before mean-reverting.
Change Outputs and UTXO Artifacts
Bitcoin's UTXO model means every transaction creates new outputs, including change sent back to the sender. These outputs are priced at the current market rate in realized cap calculations, which can inflate the aggregate cost basis during volatile periods. Metrics built on realized cap inherit this noise.
Entity Adjustment
Raw on-chain data counts addresses, not users. Exchange cold wallets holding customer funds appear as single addresses with massive balances. Clustering heuristics attempt to group addresses into entities, but these are imperfect. Glassnode and CryptoQuant both offer entity-adjusted variants of their core metrics that attempt to correct for this, and entity-adjusted MVRV tends to produce cleaner signals than the raw version.
Survivorship Bias
Most analysis of on-chain models focuses on the signals that worked. The stock-to-flow model's apparent success through 2021 attracted enormous following precisely because it fit historical data. Its subsequent failure is a reminder that in-sample accuracy does not guarantee out-of-sample predictive power.
Practical Applications
On-chain valuation models are tools, not oracles. They work best when used as one input alongside fundamental analysis, macro context, and an understanding of market structure changes. A few principles for applying them:
- Use multiple metrics in combination rather than relying on any single indicator
- Weight trend direction more heavily than absolute values, especially for NVT
- Adjust historical thresholds for market maturity: compare current MVRV readings to the most recent cycle, not to 2013
- Account for L2 activity when interpreting volume-based metrics like NVT
- Treat miner-focused metrics like Puell Multiple as lagging indicators that confirm rather than predict
For those exploring Bitcoin through self-custodial wallets, understanding on-chain metrics adds a layer of context to market conditions. Tools like Spark's stock-to-flow tracker let users visualize historical model accuracy alongside actual price data. Developers building on Bitcoin's Layer 2 infrastructure can explore the Spark SDK documentation to understand how off-chain transaction patterns interact with the on-chain data these models depend on.
As Bitcoin matures, the relationship between on-chain data and price is evolving. The metrics described here remain among the most rigorous tools available for assessing Bitcoin's position within a market cycle. Their value lies not in producing precise price targets but in providing a transparent, verifiable framework for understanding the behavioral dynamics of a network where every transaction is public.
This article is for educational purposes only. It does not constitute financial or investment advice. Bitcoin and Layer 2 protocols involve technical and financial risk. Always do your own research and understand the tradeoffs before using any protocol.