How to Use Glassnode and CoinGlass Without Misleading Yourself

What Glassnode and CoinGlass actually are (and aren't)

Glassnode and CoinGlass are two of the most commonly screenshotted analytics platforms in crypto. Both make BTC and ETH market structure look measurable. Both are also routinely misunderstood, because a dashboard is not the same thing as a strategy.

Glassnode sits on the on-chain side. It indexes blockchain transactions, clusters wallets into labelled entities (exchanges, miners, funds), and turns that data into indicators like exchange netflow, SOPR, MVRV, and the percentage of supply in profit. The product is research-oriented, the metrics are designed for macro context, and a lot of Glassnode's work shows up as paid research reports that retail traders re-share as if they were free.

CoinGlass sits on the derivatives side. It aggregates futures data from many exchanges: open interest, funding rates, long and short liquidations, options max pain, and the now-famous liquidation heatmap. It is fast, visual, and built for people trying to time volatility. The heatmap in particular has become a meme: a red tower that "predicts" a flush, followed by a real flush, followed by a hundred screenshots.

The honest framing: these are signals, not systems. A single metric on either platform is one input. Used alone it is easy to fool yourself. Used together, with cross-checks, they become a useful lens on where leveraged positions sit and how coins are moving on and off exchanges.

Why on-chain and derivatives data need each other

On-chain data answers a question about where coins are. Derivatives data answers a question about where leverage is. A complete picture needs both, because they fail in different ways.

On-chain data looks authoritative because it is drawn from a public ledger, but a wallet's label is only as good as the labelling engine behind it. A transfer out of an exchange's known hot wallet may be a customer withdrawal, an internal treasury move, an OTC desk settlement, or simply a wallet-rotation security procedure. The label says "exchange outflow"; the meaning can be benign.

Derivatives data looks authoritative because it shows real liquidations and real funding, but it is also a snapshot of where leverage was placed in the recent past. Crowded short liquidity below the market does not pull price into it magnetically; it just tells you that if price gets there fast enough, the engine will trigger stops. Whether price gets there depends on spot flows and macro context, which is exactly what derivatives data does not show.

The two together let you ask sharper questions. If exchange netflow is heavily negative (coins leaving exchanges, which usually suggests holders are moving to cold storage) and the liquidation heatmap shows a dense cluster of longs above current price, a trader can form a hypothesis: a flush up into liquidations looks more likely than a flush down, because the path of less resistance runs through that leverage, not through cold-stacked coins. That is a working frame, not a trade.

Reading exchange netflow the right way

Exchange netflow is the single most quoted and most misused Glassnode metric. The raw chart is simple: net deposits minus withdrawals over time. When the line goes down, more coins are leaving than entering, and the crypto Twitter shorthand is "exchange outflow = bullish." That shorthand is half right and half dangerous.

The half right part: when coins leave an exchange's known wallet cluster, they are usually moving to self-custody or to a destination where immediate sale is harder. That reduces the immediate sellable supply and is, all else equal, mildly bullish.

The dangerous part is treating raw outflow as the signal. A large raw outflow on its own can be:

• A genuine withdrawal to cold storage by a long-term holder, which is the bullish read.
• An internal rebalance between an exchange's hot and cold wallets, which means nothing about demand.
• A transfer to an OTC desk or prime broker, where the same coins may be sold over the following days in a way the on-chain metric cannot see.
• A redemption or wrap operation (for example, coins moving to mint or burn a wrapped or liquid-staking token), which again is not demand information.

The cleaner reading uses netflow, the difference, over a longer window (7-day or 30-day moving average), and filters out known internal wallets when you can. A sustained negative netflow across multiple exchanges is a stronger signal than a single-day spike on one venue. And the most disciplined version pairs the chart with an entity-adjusted breakdown (miners vs exchanges vs funds) so you know which cohort is actually moving.

On the flip side, positive netflow (coins flooding into exchanges) is the bearish shorthand, and it has the same problems. Deposits into an exchange can be customers prepping to buy stablecoins, a miner sending rewards to be sold, or an exchange topping up a hot wallet because withdrawals were heavy the prior week. The label alone cannot tell you which.

SOPR, aSOPR, and their real caveats

SOPR (Spent Output Profit Ratio) is the ratio of the USD value at the time a coin is spent to the USD value at the time it was last moved. In plain English, it asks: are coins being spent at a profit or a loss, and by how much?

SOPR above 1 means coins moved on-chain are, on aggregate, being sold for more than their cost basis. SOPR below 1 means they are being moved at a loss. The mechanic that makes SOPR useful is loss aversion: many holders refuse to sell at a loss, so SOPR tends to oscillate around 1, acting as a kind of realized-cost gravity well.

aSOPR (adjusted SOPR) strips out the very long-term, often-stale coin movements that can dominate the raw SOPR signal. If a coin last moved in 2013 and moves today, that single transfer can spike or crater the metric in a way that has nothing to do with current market participants. aSOPR smooths those out so the ratio reflects more recent cohort behavior.

The caveats are real, and they are the difference between using SOPR as a useful context and getting burned by it:

• Exchange internal transfers pollute the signal. When an exchange consolidates customer balances across hot and cold wallets, every consolidation looks like a sell event at some price, and that distorts SOPR. Glassnode tries to filter these out, but the filter is imperfect and you should assume any sharp SOPR move has been touched by exchange plumbing.
• The metric is lagging on shocks. SOPR reacts to moves that have already happened. It is a confirmation tool, not a leading indicator. Treating a SOPR bounce as "bottom is in" without other confirmation is a classic mistake.
• Free tier coverage is limited. On Glassnode's free plan, aSOPR is one of the metrics behind a paywall, and historical depth on SOPR itself is shorter. If you are relying on a one-year chart of SOPR from a free account, you are missing the long-base-rate context that makes SOPR useful in the first place.
• Stablecoin and wrapped-asset transfers are noise. Wrapped BTC and stablecoin movements can look like spend events with weird cost bases and skew the ratio. Cross-check with a known-clean address filter when possible.

The honest way to use SOPR: as a confluence check, not a trigger. If price is testing a known resistance level, SOPR below 1 suggests holders are underwater and reluctant to sell (which can be supportive). SOPR spiking well above 1 during a rally suggests profit-taking pressure is building. Neither is a verdict.

Liquidation heatmaps: how they actually work

CoinGlass's liquidation heatmap looks magical because it often seems to mark the exact price where a wick ends. It is not magic. It is arithmetic, and the arithmetic has a half-life.

Every leveraged position on a tracked exchange carries a liquidation price. That price is a function of the entry, the leverage, the maintenance margin, and the funding that has accrued. CoinGlass aggregates these liquidation prices by bucket (every $50 or $100 on BTC, every $0.50 or $1 on ETH at typical settings) and shades the buckets by total notional liquidated at each price. The taller the tower, the more leveraged notional is concentrated at that price.

Mechanically, this matters because:

• Cascades trigger at clustered prices. When price moves into a dense liquidation band, forced market orders hit the book, and those orders can push price further into the band, triggering more. That is the cascade dynamic that gives heatmaps their predictive reputation.
• Funding flips the bias. A dense cluster of longs above current price with heavily positive funding means those longs are paying shorts to stay open. That is a self-reinforcing pressure to flush them, which is why short-term tops often line up with crowded long leverage.
• Whale players see it too. Large market participants watch the same heatmaps. A dense short tower below price can be a magnet for stop-hunts designed to scoop liquidity before a real move. The heatmap can be a coordination device as much as a measurement.

What the heatmap does not do is predict intent. It predicts mechanical behavior if price reaches a level. The big caveat is time decay. A liquidation cluster is a snapshot of leverage placed hours or days ago. If traders have closed those positions, reduced leverage, or shifted entries since, the cluster is no longer live. CoinGlass updates continuously, but the visual "tower" you see on a chart shared on X may be from hours earlier. Always check the timestamp and, if possible, cross-reference with the live open-interest chart.

Practical use: treat the heatmap as a "where could a fast move accelerate" map, not as a price target. If price is drifting toward a tower and funding is heavily one-sided, the risk of a fast wick into that tower is real. If price is far from any tower and funding is flat, the heatmap is mostly background noise.

Free vs Pro: what you actually lose on the paywall

Both platforms have free tiers that are useful for getting oriented and paid tiers that meaningfully expand what you can do. Knowing the gap matters because a lot of social-media screenshots come from Pro accounts and quietly hide limitations a free user would hit.

On Glassnode, the free tier gives you a curated set of on-chain metrics on BTC and a smaller set on ETH, with limited historical depth (often a few years rather than the full chain history). aSOPR, full MVRV breakdowns by cohort, and several entity-adjusted flows sit behind the Standard or Pro subscription. Free users can still see the headline exchange netflow and SOPR charts, but the resolution, the historical context, and the most useful cohort splits are paywalled. That is the practical reason so many "Glassnode charts" online are second-hand rather than primary.

On CoinGlass, free access is generous on the derivatives side. The liquidation heatmap, funding rates, and open interest are visible without a subscription, with a reasonable set of exchanges included. The paid tier adds deeper historical data, more exchange feeds (including some that are excluded from the free view), options-chain detail, and faster updates on liquidations during fast moves. For most retail users, free CoinGlass is enough for daily context; the gap shows up when you want to backtest a heatmap-based idea across multiple cycles.

The honest rule: if you are about to make a decision on a metric, confirm the tier you are looking at. A free-tier SOPR chart that omits exchange filtering is not the same dataset as a Pro chart. Treat screenshots that do not specify the source as second-hand and demand the original link.

Cross-checking with CryptoQuant and Dune

No single analytics platform is complete. Glassnode is strong on chain-native metrics, CoinGlass is strong on derivatives, and each has known blind spots. A reader who wants real confidence cross-checks.

CryptoQuant is the closest direct peer to Glassnode. It also indexes BTC and ETH on-chain data, but it differs in two practical ways. First, its exchange reserve and exchange flow metrics use different wallet-clustering heuristics, so when Glassnode and CryptoQuant agree on direction, that is meaningful; when they disagree, the disagreement is itself information (often a labelling difference). Second, CryptoQuant is more aggressive about surfacing miner flows and stablecoin supply, which can be useful when you want to know whether fresh dollars are entering the system. For SOPR specifically, CryptoQuant's version sometimes uses different filtering, so a SOPR value of 1.02 on one platform can map to 0.98 on the other. Always compare like with like, and prefer relative moves (up or down) over absolute values when cross-checking.

Dune is a different animal. It is a query platform over raw blockchain data, and the dashboards on it are built by community analysts. The upside is transparency: you can read the SQL. The downside is variance: a popular Dune dashboard may be maintained by one person who can change the methodology, and methodology changes are easy to miss. Dune is the place to go when you want to verify a specific claim (for example, "what fraction of ETH exchange outflows last week went to known staking deposit contracts?") rather than to consume a finished chart.

A reasonable daily workflow: use CoinGlass for derivatives context (heatmaps, funding, OI), use Glassnode free or Pro for the headline on-chain reads (netflow, SOPR), and keep CryptoQuant open in a second tab to spot-check the same metric with a different clustering. Reach for Dune when you want to interrogate a specific data point rather than read a chart.

How to read these tools without fooling yourself

The trap with analytics platforms is not that the data is wrong. The data is mostly fine. The trap is the cognitive loop they create: you look at a chart, you form a story, you look at another chart that confirms the story, you post the story, and the story becomes a narrative that the next ten people repeat. A few habits help break that loop.

Treat every metric as one vote. A heatmap tower and a netflow dip and a SOPR bounce do not, together, "prove" anything. They are three votes from three different methodologies. Two out of three agreeing is a hypothesis. Three out of three is a stronger hypothesis. None of it is a trade.

Anchor on the base rate. Before reacting to a metric, ask what its normal range is. SOPR spends most of its time between 0.95 and 1.10. A reading of 1.04 is not "elevated"; it is Tuesday. Liquidation heatmap towers are normal. Without base rates, every move looks like a signal.

Decide your action before you look at the chart. This sounds backwards and it is the most reliable defense against narrative bias. Decide in advance what conditions would change your view. Then look. If those conditions are not met, the chart is not a trade.

Watch for regime change, not level. A metric that has been trending one way for months and then flips is more informative than a metric at an absolute level. A netflow that has been negative for six weeks and just printed its first positive day is worth attention. A netflow that bounces around zero every week is not.

Track your own hit rate. If you act on these signals, log the metric, the action, and the outcome. After fifty trades you will know whether the signal actually has edge for you. Most people skip this step, which is why most people cannot tell whether the platform helped them or hurt them.

Following on-chain and derivatives data the smart way

BTC and ETH market structure moves fast, and so does the narrative around it. Tracking exchange flows, SOPR, and liquidation heatmaps manually across Glassnode, CoinGlass, CryptoQuant, and Dune is a losing game; the dashboards update constantly, the noise is high, and the temptation to read a single chart as a signal is constant. Zippfeed surfaces BTC and ETH on-chain and derivatives headlines with sentiment scoring (bullish, neutral, or bearish) and an importance rating, so you can separate real structural shifts from chart-of-the-day noise and react with context instead of impulse.