What we tested about token unlocks, and what didn't work

Most popular unlock ideas didn't survive our tests. We publish the misses because they matter as much as the one rule that held.

Every idea below was written down before the numbers were computed, tested on 305 unlocks from 2023–2026 (more for some tests) against Bitcoin, and corrected for the number of ideas we tried. “Doesn't work” means the effect was too small or too unstable to rely on, not that it can never happen.

• Strong unlocks (≥ 10.4% of circulating) are weaker than Bitcoin in the week before — Works
  68 of 100 fell vs BTC, average -5.1%. Held in both halves of the sample and after removing the biggest moves. This is the one rule the site is built on.
• Small unlocks of 1–2% of supply — Doesn't work
  552 events: the move before the date was indistinguishable from a random week. Calendars that list every small unlock as news are listing noise.
• Who receives the coins (team vs investors vs treasury) — Doesn't matter
  Splitting the archive by recipient type gave no reliable difference. The size against circulating supply explains far more than who gets the coins.
• Unlock value against daily trading volume — Doesn't work
  Unlocks worth many days of trading did not behave differently from those worth less than a day (p 0.35). We still show it on token pages as context, not as a score.
• A bounce after the unlock — Doesn't work
  Neither the 3 days nor the 2 weeks after a strong unlock showed a reliable rebound. Coins that fell before the date did not snap back on average.
• The drop continuing after the date — Not confirmed
  Some further weakness in the 3 days after the unlock, but below our significance bar. We watch it in a forward test instead of claiming it.
• Starting two weeks early instead of one — Worse
  Measuring from 14 days before captured less of the move than from 7 days. The pressure concentrates in the last week.
• The last day before the unlock — Noise
  The final 24 hours carried no reliable signal; the move is mostly done by then. Three to four days before still showed most of the effect.
• Exchange inflows on unlock day — Doesn't work
  Tokens moving to exchanges on the day (197 unlocks) did not predict the next days' move.
• Funding-rate skew between exchanges — Doesn't work
  The idea that a crowded short on one venue signals a bigger fall did not hold (correlation −0.004).
• Volume spikes and buying pressure just before — Doesn't work
  Unusual volume in the days before, or a high share of aggressive buys, added nothing to the size of the unlock.

Details of the method: how we score unlocks. Questions about a specific result: ask in the Telegram channel.