moonboard.ioResearch & Data
Research

Where measuring crypto markets gets hard

Moonboard publishes numbers about assets. This section is about how much those numbers can carry - the areas where measurement is genuinely difficult, and where the platform's own approach has limits worth naming.

Most crypto research output falls into one of two categories: commentary with no data behind it, or backtests with no acknowledgement of how fragile they are. Neither is useful for very long.

What follows is organised differently. Each area starts from a measurement problem, sets out what can actually be observed, and is explicit about which claims the available data supports and which it does not. Where Moonboard's own metrics have a known weakness in an area, that weakness is stated here rather than in a footnote — the methodology pages carry the same information from the other direction.

Published areas

Social intelligence

What post counts, engagement and share of voice actually measure - and the gap between conversation volume and conversation that carries information.

Market sentiment

Classifying tone at scale: where NLP works on crypto text, where irony and jargon break it, and what happens to a sentiment score at the extremes.

Liquidity and market structure

Turnover as an observable, depth as an inference, and the point where widely used liquidity proxies stop describing what they claim to.

Ecosystem and developer activity

Repository activity, contributor counts and community growth as measures of project health over horizons longer than a trading day.

Narratives and sector rotation

How themes are defined, why sector classification is a modelling choice rather than a fact, and what rotation looks like in measured data.

AI agents and synthetic activity

Separating organic engagement from generated and coordinated activity - and why confident detection claims deserve scepticism.

Open areas

Named because they are part of the agenda, not because there is anything to read yet. These have no pages behind them, and will only get one when there is something substantive to say.

Market microstructure

Order flow, price impact and execution, where public data is thinnest.

not yet published

Anomaly detection

Identifying observations that break from an asset's own recent distribution.

not yet published

Market manipulation

Wash trading and coordinated promotion as measurement problems.

not yet published

AI agents in financial markets

Autonomous participants as both consumers and producers of market data.

not yet published

Community dynamics

Growth, retention and concentration in asset communities over time.

not yet published

Digital asset ecosystems

Cross-chain activity and how ecosystem boundaries are drawn.

not yet published

How this research is produced

Everything here is written by the Moonboard team from the data the platform collects and processes. Where a claim rests on Moonboard's own observations, it says so and describes the window and the universe. Where something is a hypothesis, it is labelled as one.

There are no citations to studies that were not read, no invented statistics, and no results presented as validated that have not been validated. When a research note eventually reports a measured result, it will come with its method, its sample and the biases that could have produced it — see research notes for the format.

Related