Method & data status

The short version

These are community registry reports, not official police totals. Counts are influenced by who reports, how popular each motorcycle is, source coverage and collection continuity. TheftScore describes observed patterns; it does not predict whether an individual bike will be stolen.

Report volume

Registry entries are normalised into a make, model, segment, report date and source geography where those fields can be identified. Rankings count records by make and model. Monthly charts group dated records by calendar month and explicitly show missing months as zero. Current partial months are excluded from default comparisons.

Great Britain map counts use a conservative rollup to 11 regions. They are raw report concentration, without regional population or motorcycle-stock denominators, so they must not be read as geographic risk.

Relative representation signal

The representation signal divides a model’s all-history community report count by the latest currently licensed motorcycle count in DVLA VEH0120 vehicle licensing statistics, then multiplies by 10,000. Make matching is case-insensitive and exact; model matching removes non-alphanumeric characters and checks DVLA’s Model and GenModel, with a small set of data-verified aliases. An unmatched model receives no denominator or signal.

A model currently needs at least 250 licensed motorcycles and 10 registry reports. Within each classified motorcycle segment, its result is compared with the median qualifying result: below 0.6× is shown as “below typical,” 0.6–1.4× as “typical range,” 1.4–2.5× as “above typical,” and 2.5× or above as “strongly above.” Unclassified models receive no band.

This is not an annual incidence rate. It combines reports accumulated across the registry archive with one current DVLA stock snapshot. Older or declining model populations can therefore appear more represented, and small qualifying samples can be unstable. Treat it as a comparison prompt, not a risk verdict.

Current coverage

The dataset contains 4,392 reports across 816 observed make/model labels and 120 makes. 2.5% of records are unclassified because their source label and description do not identify a segment safely; they remain visible rather than being guessed. The latest indexed report date is 2026-07-19. That date is not a guarantee of ingestion freshness.

Sources & terms

Responsible use

Use these summaries alongside official local data, security practice, storage conditions and insurance advice. A registry count is evidence of reports in this source—not proof of total crime, a forecast, or a reason to label a place or motorcycle “safe.”

Questions & corrections

If you spot a classification issue, a broken source or something that needs clearer wording, email [email protected]. Please include the page URL and the specific figure or label where possible.