Every term, in plain language.

Reliability grade

A–E, from the vehicle’s failure frequency relative to the market (A = most reliable). Built only from real MOT data — modelled signals never move it.

Relativity

How a vehicle’s failure hazard compares to the market average at the same age/mileage. 1.0 = market average; below 1.0 is better than average.

Burn cost

Expected claim cost over a period = frequency × severity. The warranty premium is the burn cost grossed up for a target loss ratio.

Frequency shape vs level

The shape of failure-by-age comes from real MOT inspection curves; the absolute level is anchored to FCA-regulated warranty claims disclosures. Two different, independently-sourced things.

Severity priors

Typical repair cost per component/segment — a documented research estimate from local workshop pricing, always flagged “estimated”, not audited claims data.

Confidence band

The ± range on a figure, with named drivers (thin cell, fallback level, estimated severity, projection distance). Wider means less certain — and it says why.

Coverage tier

mot_direct (real data for this model) · platform_proxy (scored via a documented mechanical twin) · component_proxy (only engine/transmission via a sibling) · insufficient_data (withheld, nothing fabricated).

Cross-market (xm) signal

Engine/transmission risk from US NHTSA complaints/recalls, since MOT never inspects those internals. Tiers: nhtsa_matched (real data) → engine_proxy (borrowed sibling) → segment_modelled (class average).

Chemistry modelled

An EV battery degradation curve derived from its documented battery chemistry class (LFP/NMC/NCA) using published class-level fade research — not per-model telemetry.

Fleet empirical

An EV battery degradation curve from a real per-model fleet or academic study (the strongest EV-battery tier).

Carry-forward / default relativity hold

Projecting a model past its own newest observed age by holding its market-relativity constant. “Researched generation” is a verified boundary; “default relativity hold” is the weaker, wider-band version with no generation research.

Segment percentile

How reliable this vehicle is versus same-segment peers of the same age. Mileage percentile is the analogous odometer-vs-age comparison.

Local cost index vs FX rate

Two separate market factors: the parts/labour cost differential vs Singapore, and the SGD→local currency conversion — kept distinct so you can see what drives a price difference.