SentalysisIn this guide
Key takeaways
- Sentiment is measured indirectly. Every indicator is a proxy for a mood, with its own blind spots.
- There are six broad families of sentiment data: price-derived, positioning, on-chain, social discussion, attention and surveys.
- Composite scores such as the Fear and Greed Index blend several families into one number, which is convenient and loses detail.
- Compare a reading with its own history, for example with a Z-score, rather than with the absolute level of a different series.
- Sentiment describes how people feel, not where prices go next. Treat it as context for research, not a forecast.
What market sentiment means
Every price is set by people deciding to buy or sell, and those decisions depend partly on how they feel: confident, anxious, bored or excited. Market sentiment is the name for that shared mood. It sits alongside fundamentals, the things that would matter even if nobody felt anything, such as supply, usage and interest rates.
Crypto is often described as a sentiment-heavy market, for structural reasons that are easy to see: it trades around the clock, a large share of participants are individuals rather than institutions, and much of the conversation about it happens in public, on social media, in real time. Narratives travel quickly, and so do changes of mood.
None of that makes sentiment easy to measure. You cannot poll every holder every hour, and the people who answer polls or post online are not a random sample. So sentiment analysis works with traces: things people do, or say, or search for, that move with their mood.
The six kinds of sentiment data
Most crypto sentiment indicators fall into six families. Each answers a slightly different question, and each can mislead in its own way.
| Family | What it measures | Watch out for |
|---|---|---|
| Price-derived | Volatility, momentum, drawdowns and trading volume compared with recent averages | It describes what price already did, so it can restate the price rather than explain it |
| Positioning | Derivatives data such as funding rates, open interest, long/short ratios and liquidations | Coverage differs by exchange, and crowded positioning can persist for a long time |
| On-chain | Exchange flows, active addresses and holder behavior recorded on a blockchain | Slow-moving, chain-specific, and its interpretation is often contested |
| Social discussion | The volume, tone and themes of posts about crypto | Bots, sarcasm and coordinated campaigns, and posters are not all holders |
| Attention | Public interest in words and topics, read against each topic's own history | Attention has no direction: excitement and alarm both raise it |
| Surveys | What participants say about their own expectations | Small, self-selected samples, and people do not always do what they say |
A useful habit is to name the family before you trust a number. A gauge that is mostly volatility will look fearful after a crash almost by definition; a gauge built from discussion can be swung by a handful of loud accounts.
Attention: the broadest trace
Attention data has one big advantage over discussion: it includes people who never post. Someone who is simply curious about “bitcoin” during a sell-off leaves a trace without joining any conversation. It also reaches beyond crypto words. Attention to recession, conflict or interest rates can be set beside a crypto price just as easily as attention to Bitcoin itself.
Its weakness is the flip side of its breadth. A spike in attention says that more people were interested, not whether they were hopeful or frightened. The chart below shows why that matters. Attention to the phrase “bull run” peaked in August 2025, close to the price high. Attention to Bitcoin itself peaked months later, in February 2026, during a sell-off.
The same word can draw attention for opposite reasons in different months. That is why attention is most useful as a measure of intensity, read alongside price and discussion, rather than as a measure of mood on its own. The turning-points study shows where attention spiked around Bitcoin's highs and lows.
Composite scores and what they hide
Because each family is partial, many tools blend several into one number, usually a 0–100 fear and greed score. The Sentalysis Fear & Greed gauge is one: it reads market mood from attention to fear-related and greed-related topics, and shows both sides next to the needle.
A composite is convenient: one glance, one number. It is also lossy. When its inputs disagree, the score can land in the middle and read as “neutral” while the market is anything but calm. That is why the Sentalysis gauge labels a middle reading Mixed when both sides are strong. And a composite built mostly from price behavior partly restates the price chart it is placed next to.
How to read sentiment without fooling yourself
Five habits make sentiment data far more useful, and far less likely to tell you what you already believed.
- Compare a series with its own past. A reading of 40 means nothing until you know whether 40 is normal for that series. A rolling Z-score expresses every reading as distance from its recent average.
- Check more than one family. If attention, discussion and positioning all moved together, the observation is sturdier than any one of them alone.
- Look for persistence, not one spike. A single extreme day can be news; a theme that stays elevated for weeks is a different kind of observation.
- Separate the topic from the direction. Knowing that people are paying attention does not tell you which way they are leaning.
- Write the question down first. Deciding what you are looking for before you look is the simplest defense against seeing patterns in hindsight.
What sentiment cannot tell you
Sentiment data describes a mood; it does not establish cause. When attention and price move together, either could be reacting to the other, or both to something else entirely. Extremes can last far longer than seems reasonable, and relationships that held in one period can vanish in the next.
Some research finds short-lived links between sentiment measures and crypto returns, but they vary by period, asset and method, and they tend to weaken once they are widely known. The honest use of sentiment is as context for your own research.
Where Sentalysis fits
Sentalysis focuses on two of the six families. The Attention Interest Oscillator measures attention to tracked keywords across crypto, markets, geopolitics and macroeconomics, and sentiment charts put it beside the price of seven crypto assets. HiveMind organizes social media discussion into themes and strands. Composite indexes and data alerts let you combine and follow the series you care about.
FAQ
Questions and answers
What is crypto market sentiment?
It is the shared mood of crypto market participants: how optimistic, fearful or indifferent they are about the market. Because a mood cannot be observed directly, it is measured through traces such as price behavior, positioning, on-chain activity, social discussion, attention and surveys.
How do you measure crypto sentiment?
With proxies from six families of data: price-derived measures like volatility and momentum, derivatives positioning, on-chain activity, social media discussion, public attention to topics, and surveys. Composite gauges, such as the Sentalysis Fear & Greed gauge, combine several signals into one score.
What is the best crypto sentiment indicator?
There is no single best one, because each family answers a different question. Attention is broad but has no direction, discussion has tone but is easy to manipulate, and price-derived measures describe what already happened. Comparing a few families, each against its own history, is more informative than relying on one number.
Does sentiment predict crypto prices?
Not reliably. Some studies find short-lived relationships between sentiment and returns, but they vary by period and asset and often fade once widely known. Sentiment is best used as context for research, not as a forecast.