Skip to content
Doefin
All research

Observed network research

What changed across 25 observed difficulty adjustments?

25 consecutive difficulty changes from a fixed mempool.space snapshot, with the original source response, calculation method and downloads.

Observed network report · Published

25 adjustments: 11 rises and 14 falls

The dataset records 25 consecutive difficulty changes between 2025-11-12 and 2026-10-16: 11 increases and 14 decreases. The mean absolute change was 3.75%. The minimum change was -11.16% and the maximum was 14.73%.

How the changes were calculated

The source is the mempool.space one-year difficulty-adjustment API, captured on 22 October 2026. Sort its 26 adjustment blocks by height and check that consecutive blocks are 2,016 blocks apart. Exclude the earliest: its preceding difficulty is outside the snapshot. For each remaining adjustment, calculate 100 × (new difficulty / previous difficulty − 1). Do not use the provider’s rounded adjustment ratio.

The mean absolute change averages the size of the 25 percentage changes, ignoring whether each is positive or negative. Each change has equal weight. Calculations retain the provider’s full difficulty precision; rounding affects only the display.

Sources, version and citation

The downloads include the source response exactly as received, plus matching block details for the first and last retained observations. These checks use the same provider. They show consistency within that source, not independent verification against a Bitcoin node. The API’s rolling window determines which adjustments are included; the study does not select profitable periods.

All findings and downloads are available without email. Version 1.0.0 has a stable dataset identifier and a citation manifest. The manifest lists SHA-256 digests, or file checksums, for the original response and CSV/JSON downloads. They detect changes from this version, but prove neither an external publication time nor that hosted files cannot change.

No DOI, a persistent research identifier, has been registered or deposited. An update requires a new version.

What these observations cannot establish

The forecast archive has zero published records. Dated forecasts and accuracy examples elsewhere on this site use made-up values. A historical adjustment is not a prediction published before its outcome.

These findings measure neither forecast error nor the reliability of forecast ranges. They establish no forecasting advantage, mining profit or trading return. The comparison guide explains which original records are needed to score forecasts.

Proof of publication before an outcome or market listing needs independently verifiable records. This site cannot create that proof after the event.

Observed network research · v1.0.0

Download and cite the dataset.

Source: mempool.space, captured 2026-10-22.

Inspect the forecast archive statusInspect the archive status →

Preview a report and dataset email Demo

Preview the email content. No email is sent, no personal details are saved and no newsletter subscription is created.

Try the calculation

Calculate error for five example forecasts.

Find how far each forecast was from its outcome, then average those distances. The five pairs are synthetic, and the CSV contains every value.

Calculate forecast error