FAQ and methodology
How the score works
No black box. Here is exactly what goes into the number, where the data comes from, and, just as important, what it cannot do.
What is the Onchain Intelligence score?
A single 0 to 100 reading of Bitcoin or Ethereum's onchain and derivatives conditions. 60 or above leans bullish, 40 or below bearish, around 50 neutral. It is the weighted blend of 13 metrics, re expressed as deviations from their own recent history, then standardized so the score uses the full range.
Which metrics go into it?
Exchange netflow, netflow to reserve ratio, exchange reserve, MVRV, realized cap HODL momentum, SOPR (proxy), active addresses, hashrate, miner pressure (MPI proxy), TVL, stablecoin supply, perp funding rate and open interest. Contrarian metrics such as MVRV, funding and exchange inflow push the score down when they run hot, which is mean reversion by design.
Where does the data come from?
Coin Metrics for onchain data such as MVRV, exchange flows, reserves, active addresses, hashrate and issuance, Binance Futures for funding and open interest, and DefiLlama for TVL and stablecoins. Only SOPR and Miner Pressure carry a
proxy label because their true versions need paid UTXO level data.What is a z score and why use it?
A z score says how many standard deviations a metric sits from its own 90 day baseline. “Exchange outflow is 2 sigma below normal” is a real signal, while “down 1.5 percent today” is just noise. It makes very different metrics such as hashrate, funding and TVL comparable on one scale.
Does onchain data actually predict the next 7 to 30 days?
Only weakly, and the tool is honest about it. Across roughly six years of out of sample, walk forward testing, the correlation between score and forward return is small but positive, and higher score bands do post higher average forward returns. Shorter horizons of 7 to 14 days tend to be cleaner than 30. Bottom line, it is a monitored signal, not a trade trigger.
What is walk forward validation?
Weights are fit only on past data, then scored on unseen future days, an expanding window, with no lookahead. The backtest card shows out of sample correlation next to in sample, and the gap between them is the overfitting penalty. The out of sample number is the one to trust.
Why does the score sometimes move opposite to price?
By design. Valuation and leverage metrics such as MVRV, funding and exchange inflows are contrarian. When price pumps, they heat up and pull the score down, flagging over extension. A rising score into a falling price is a possible bullish divergence, which the tool surfaces as a signal chip.
How often does it update, and does it store my data?
It refreshes on load and every few minutes. It stores nothing about you on a server. Score snapshots live only in your browser to draw your session history, and data responses are briefly cached in the browser. No accounts, no tracking.
Is this financial advice?
No. It is an educational research tool. It does not tell you to buy or sell, and it can be wrong. Do your own research and manage risk.