Skip to content
Doefin

Doefin / Research

Evaluate a difficulty forecast.

How close was a forecast to the outcome? Learn how to compare forecasts and calculate their errors.

Observed network research · v1.0.0

25 recorded difficulty adjustments.

From 2025-11-12 to 2026-10-16, the dataset records 11 increases and 14 decreases. Mean absolute difficulty change: 3.75%. These are historical network changes, not forecast errors.

Source: mempool.space, captured 2026-10-22. This is a fixed one-year snapshot, not a live feed. No independent check against a Bitcoin node is documented.

Forecast archive: no published predictions

The archive contains no captured provider predictions or independent timestamps proving publication before an outcome or market listing. Dated forecast examples elsewhere on this site use made-up values.

Inspect the archive status →

Forecast comparison guide

Published

Comparing difficulty forecasts

Compare forecasts for the same adjustment and lead time, then assess their point errors and ranges. The linked calculation uses made-up values.

Read the comparison guide

What the report covers

  • Matching targets and forecast times
  • Error and uncertainty as separate measures
  • Records needed to repeat the calculation

The calculation

Calculate forecast error.

For each of the 5 made-up pairs, find how far the forecast was from the outcome. Average those distances, then download the CSV.

Work through the calculation

One synthetic pair

|3.2% − 2.0%| = 1.2 pp

The error is the size of the gap between the forecast and example outcome. pp means percentage points.

Sample prepared .

Understand the method and data.

All resources