Forecasts
For every coin we price, we work out the range its price is likely to stay within over the next 24 hours and 7 days, from its own recent volatility on our own on-chain data. The range should hold 80% of the time. Each forecast is scored when its time is up, and the record below is published as it stands. A range says how far a price tends to move, not which way, and it is not financial advice.
Track record, last 90 days
Coverage is how often the price ended inside its range: 80% is right, lower means the ranges are too narrow, higher means too wide. The score adds a penalty for every miss to the width, so lower is better. A model appears on coin pages only with at least 200 scored forecasts, coverage between 72% and 88%, and a better score than the baseline. Method version 1.
| Horizon | Model | Scored | Coverage | Average width | Score | Status |
|---|---|---|---|---|---|---|
| 24 hours | Recent volatility (EWMA) | 0 | – | – | – | No scored forecasts yet |
| 24 hours | Baseline: plain volatility | 0 | – | – | – | No scored forecasts yet |
| 7 days | Recent volatility (EWMA) | 0 | – | – | – | No scored forecasts yet |
| 7 days | Baseline: plain volatility | 0 | – | – | – | No scored forecasts yet |
How it works: the model measures each coin’s hourly price swings, weighting the last few days most, and scales them to the horizon. The baseline weights the whole week equally. Crypto prices jump more often than this simple picture allows, which is exactly what the record is there to catch.