Reading on-chain data without fooling yourself
Every transaction is public. That doesn't mean every chart tells the truth.
Analyst · 9 min · Updated 2026-11-08
💡 Tracking shipping containers
Imagine seeing every shipping container moving around the world — but not what's inside or who owns it. You can spot busy ports and big shipments, but one company moving boxes between its own warehouses can look like a trade boom. On-chain data is the same: incredible visibility, easy misinterpretation.
Popular metrics and their traps
| Metric | What it suggests | Common trap |
|---|---|---|
| Active addresses | Network usage | One person can use thousands of addresses; bots and spam inflate counts |
| Exchange inflows/outflows | Possible selling (in) or self-custody (out) | Exchanges reshuffle internal wallets; labels can be wrong |
| Fees / revenue | Willingness to pay for block space | One-off events (mints, airdrops) create spikes |
| Coin-age / long-term holders | Holder conviction | Old coins moving could be custody migrations, not selling |
| TVL (DeFi) | Capital deposited in protocols | Rises with token prices and can be double-counted |
Three rules of honest analysis
- Methodology first: know how the metric is defined and which addresses are labelled.
- Look for a mechanism: why would this metric cause price to change?
- Check the base rate: how often did this "signal" fail in the past?
Correlation is not a crystal ball
Many indicators look predictive when tested on the past only because they were chosen after the fact. That's overfitting. Robust research tests on out-of-sample periods and reports failures, not just wins.
🧮 Practical example: the "whale dump" that wasn't
Key takeaways
- On-chain data is transparent, not self-explanatory.
- Check definitions and labels.
- Demand a mechanism and test failure rates.
- Beware overfitting.
Education only — not financial advice.