90% of On-Chain Metrics Are Just Noise. Here’s What Institutions Look For Instead

Up to 99% of on-chain metrics add nothing but noise, according to Charles Edwards, founder of Capriole Investments. The data only helps investors who know which signals matter and how each one is built.

That warning lands as institutions move deeper into crypto data. Julio Moreno, Head of Research at CryptoQuant, says professional desks now verify on-chain numbers against the traditional data they already trust.

On-Chain Data Goes Institutional

On-chain analytics began as a retail trading edge. Now banks, funds, and asset managers want the same data, and they ask harder questions about it.

In a YouTube podcast from BeInCrypto’s Market Intelligence experts council, both analysts described a clear shift. Large players no longer accept a number simply because a dashboard displays it.

Moreno said institutions have changed how they treat the figures.

“Now they don’t take the data just as face value but they want to really contrast that with their traditional data.”

Same Metric, Different Dashboard

A metric’s name reveals little about how a platform calculates it. Two providers can use the same label and still produce different results.

Edwards said the gap usually comes from the method behind the label:

“People will call a metric something but then the question is how is it calculated in the back end and that’s up to the platform.”

The differences come from data sources, normalization choices, and how each provider handles gaps. Most leading data platforms publish thousands of metrics, yet few explain these methods clearly.

What the FTX Collapse Revealed

The sharpest example came from the FTX collapse in November 2022. The same metric told different stories across providers.

CryptoQuant’s exchange reserves data showed FTX holding 20,177 BTC on November 6. By November 8, that figure read 0.64 BTC.

The reserves drained before the price broke down. On-chain data flagged the stress days before the market reacted.

Moreno said the disagreement between providers came from how each one groups addresses:

“You can be more aggressive in clustering or you can be more… careful.”

Some dashboards still showed FTX holding reserves at the same time. They linked addresses more aggressively, while CryptoQuant stayed conservative.

CryptoQuant data shows FTX’s Bitcoin reserves draining to near zero around November 7, 2022, before the price collapse that followed days later.

Above, we see the scale of the drain. According to CryptoQuant, FTX held 20,177 BTC on November 6, 2022, and just 0.64 BTC by November 8.

Why Most On-Chain Metrics Are Noise

Edwards argued that the volume of metrics is not the same as value. Investors who pour every available number into a model gain little.

“90 95 99% are not really going to add anything except noise.”

His firm builds models from a small, tested set of signals. Capriole’s Macro Index combines more than 60 on-chain, macro, and equities metrics into one model.

Capriole’s Macro Index blends more than 60 metrics into a single oscillator. It is an example of the curation Edwards argues separates real signal from noise.

For institutions weighing data providers, the takeaway is direct. Study the method behind a metric, not only the number it shows. The same focus on transparency now shapes industry awards for digital asset firms. As more large players enter, that scrutiny becomes the real edge.