How our AI scoring works
Every score on CryptoInfo is generated from live data by transparent rules. Here is exactly what goes in, what comes out — and what it can’t do.
AI Crypto Picks (score 1–100)
Every 30 minutes we score the top 50 coins by market cap on four factors:
- Momentum — 7-day price change (strongest weight) and 30-day trend.
- Volume vs. market cap — real trading activity relative to size; healthy turnover scores up, dead volume scores down.
- Volatility — measured from hourly prices over 7 days; wild swings reduce the score.
- Size — very large caps get a small stability premium.
The four highest scores become the day’s picks. The one-line explanation is written by an AI model (Claude) from those exact metrics — it never invents information beyond the data.
Risk labels
Low: very large cap with calm volatility. Medium: established ($5B+) but swingier. High: everything else — including all gems.
Gem Finder
We scan coins ranked roughly 251–500 by market cap and surface the three with the strongest combination of 7-day momentum and genuine trading volume (minimum $1M/day). Small caps can move fast in both directions — every gem is labelled high risk and shows its market cap so you can judge the size yourself.
Scam Detector
The scanner reads live on-chain security data: a sell simulation (honeypot test), contract verification, buy/sell taxes, owner privileges (mint, pause, blacklist, balance edits), holder concentration, liquidity depth and LP locks. Each red flag adds to the risk score; each pass is shown too, so you see the full picture. Try it here.
AI News Summary
We pull headlines from established crypto news feeds. An AI model rewrites each in plain English and tags it bullish, bearish or neutral with a one-line market read. It summarises — it does not predict.
What none of this can do
No model — ours or anyone’s — reliably predicts crypto prices. Scores are data-driven opinions that can be wrong, data sources can fail, and past momentum says nothing certain about tomorrow. Treat everything here as a starting point for your own research, never as a recommendation to buy or sell. We may earn affiliate commissions on some product links; that never influences scores, which are computed purely from the data above.