On-chain — Glassnode, Nansen and blockchain data

Introduction

A price chart shows the effect. On-chain data shows the cause — who bought, who sold, how long they held and with what profit or loss they closed the position. In crypto every transaction is public. A trader’s edge does not come from having access to the data — Glassnode, Nansen, Arkham, Dune and CryptoQuant are available — but from the ability to read them correctly and connect them to market context.

The goal of this lesson is not to divine the future from MVRV. It is to build a system of filters that lets you assess whether the market is in a phase of distribution, accumulation or capitulation, and where investors’ cost zones sit. Without that you trade price alone, ignoring positioning.

What on-chain analysis is

On-chain analysis turns raw transactions from the blockchain (Bitcoin, Ethereum) into behavioural indicators. Unlike technical analysis, which analyses price and exchange volume, on-chain analysis analyses blockchain volume, coin age, purchase cost and flows between wallets.

We divide the metrics into 4 buckets:

The key rule: a single metric gives no signal. A confluence of 3 metrics from different buckets increases the credibility of the picture.

Fundamental metrics — how to read them without myths

MVRV (Market Value to Realized Value)

MVRV = Market Cap / Realized Cap
Realized Cap = sum (price at the moment of the last movement of each coin * quantity)
MVRV > 1 = the market is in aggregate profit
MVRV Z-Score = (Market Cap - Realized Cap) / stdev(Market Cap)

A MVRV Z-Score > 7-9 has historically coincided with Bitcoin overheating. MVRV < 1 is the zone where the aggregate investor is underwater — observed at bottoms, but it is not an entry trigger. The drop below 1 alone can last for months.

SOPR (Spent Output Profit Ratio)

SOPR = Realized value / Value at creation
SOPR > 1 = coins being sold at a profit
aSOPR = SOPR excluding transactions < 1h (filters out the noise)

An aSOPR resetting to 1 in an uptrend indicates that sellers have stopped taking profit — continuation. An aSOPR < 1 that holds for weeks is capitulation.

HODL Waves and LTH/STH

LTH — Long Term Holder: coins untouched for >155 days. STH — Short Term Holder: <155 days. A rise in LTH supply while price consolidates = accumulation by strong hands. A fall in LTH supply while price rises = distribution to STHs.

Glassnode — the working toolkit

Glassnode is the standard for Bitcoin and Ethereum. Most advanced metrics require the Studio plan, but the free tier is enough to start.

The weekly working procedure:

1. MVRV Z-Score and NUPL — where are we in the valuation cycle?
2. Exchange Net Position Change (30d MA) — are coins flowing onto exchanges (+ selling pressure) or off them (- pressure)?
3. STH Cost Basis — the average purchase cost of short-term holders. In a bull market it acts as dynamic support.
4. Stablecoin Supply Ratio (SSR) — a low SSR means large stablecoin purchasing power relative to BTC — fuel for a rally.
5. Fees vs Price — rising fees with a flat price can foreshadow an upcoming move.

Accumulation filter (template):
MVRV < 1.5 AND NUPL < 0.25 AND STH-SOPR < 1 for >14 days AND Exchange Net Position Change < 0 (outflow)

Distribution filter (template):
MVRV Z-Score > 6 AND LTH Supply falling for 30 days AND Exchange Inflow > 90-day average by 2 stdev AND SOPR > 1.15 holding

These are not buy/sell signals. They are regime filters.

Nansen, Arkham, Dune — who does what

Nansen

Wallet labelling: funds, smart money, whales. You do not guess who stands behind an address. Nansen shows Smart Money flows. Use it to:

• Check whether a token is being accumulated by the top 10-20 smart wallets
• Analyse where team unlocks end up
• Detect whether a project’s “TVL growth” is not 3 founder wallets

Arkham Intel

Similar to Nansen, but with better tracing of links between exchanges and entities. Useful for verifying project claims: “we have backing from X” — check whether X actually transferred the funds.

Dune Analytics

SQL on blockchain data. Ready-made dashboards: you can compare bridge user counts, DEX volume, L2 fees. You can build a query yourself without coding, thanks to the wizard.

How not to be manipulated by on-chain data

On-chain data has noise too. Exchange inflow can rise because an exchange is reorganising cold wallets, not because anyone is selling. A rise in active addresses can be bots doing airdrop farming.

Hygiene rules:

1. Check 7d and 30d averages, not a single daily candle of the metric.
2. Compare against 2-3 years of history, not only the last 30 days.
3. Mark internal events on the metric chart (halving, unlocks, an update).
4. Verify the source — does the Dune dashboard filter out wash trading on NFTs?

When the plan says do not trade

1. When your 3 key metrics give conflicting signals. MVRV overheated, but stablecoins are flowing in and exchanges are seeing outflows. That is consolidation. No directional edge.
2. When you look only at Bitcoin on-chain but you trade a small L2 altcoin. BTC on-chain says nothing about your altcoin’s unlock in 2 days. Focus on the on-chain of the network itself.
3. When a metric “looks cheap” but the narrative and flows are negative. MVRV <1 during an active regulatory gap (an SEC lawsuit) can fall lower. Cheap can get cheaper.
4. When you build a thesis on a single dashboard from Twitter. With no verification of the dashboard’s methodology. Dashboards often count TVL twice.
5. When on-chain data is sold to you as a certainty. “SOPR says there will be a rally”. No. On-chain data speaks about positioning, not about the future.

The mistakes that cost the most

1. Treating MVRV as an entry oscillator. MVRV >7 does not mean an automatic short. In 2020-2021 BTC held in the >1.8 zone for months. It is a regime gauge, not a trigger.
2. Ignoring STH Cost Basis as dynamic resistance in a bear market. Many traders buy at every drop of MVRV below 1, catching one bear leg after another. STH cost basis keeps falling.
3. Analysing smart money wallets without the transaction context. You see that smart money sold. You did not check that it sold because it received an unlock and has to cover fund costs. That is not a bearish thesis.
4. Mixing Bitcoin and Ethereum metrics. Do not compare BTC MVRV with ETH MVRV 1:1. They have a different supply distribution and a different holder profile.
5. No on-chain journal. You review metrics with nothing written down. A week later you do not remember what you saw. Without measurement there is no process.

Measurable exercise

A 21-day task:

1. Open free accounts: Glassnode Studio, Nansen (trial), Dune.
2. Every day at a fixed hour, write the following into a spreadsheet:

3. In the column next to it, write down what the BTC price did in the 3 days after the entry (in %).

Metrics to track:
Regime accuracy = (Number of days where conclusion = Accumulation and price +3d >0 OR conclusion = Distribution and price -3d <0) / 21
Time: 10 min/day
Goal: not to predict, but to calibrate your eye to when the metrics lie

4. At the end, choose one false signal and describe why the metric worked badly (e.g. a reorganisation of exchange wallets).

Check yourself

1. What does MVRV <1 say about the aggregate position of investors?
2. Explain the difference between SOPR and aSOPR and why the 1h filter matters.
3. What does a rising share of LTH supply with a flat price mean?
4. Why do you need Exchange Net Position Change when making a risk decision?
5. Why can a single on-chain metric not be a trading signal?

Frequently asked questions

Is on-chain data leading relative to price? Partly. Exchange flows and SOPR can show capitulation before price turns, but they do not give day-to-day timing. They are regime indicators, not oscillators. They work best as a filter: do not trade long into obvious LTH distribution.

Do I have to pay for Glassnode for this to make sense? Not to start with. The free tier shows MVRV, SOPR, Active Addresses, Fees and Exchange Flows on a daily basis. That is enough to build the habit. The paid tier pays off when you want granularity of LTH/STH cohorts and the SSR Oscillator.

How does Realized Cap differ from Market Cap? Market Cap = price * supply. Realized Cap = the sum of the prices at which each coin last moved. If someone bought BTC at 16k and holds to 60k, it is counted at 16k in Realized Cap. Hence Realized Cap shows an anchored cost, not a speculative valuation.

Nansen shows that funds are buying. Is that a guarantee of a rally? No. Funds buy OTC outside DEXs and often hedge the position with perpetuals. You see only part of the picture. Treat it as an element of confluence, not as a standalone signal. The question is: how many independent smart money wallets are buying, and have they been doing it for 2 weeks already?

How often should I check on-chain data? For a swing trader, 2-3 times a week is enough. On-chain data has inertia. Refreshing it daily generates noise and overtrading. Build a weekly checklist, not an alarm dashboard.

---