Home Price Index Methodology
Bigfella Auto Transport Price Index
Methodology
How the index is built, what it includes, what it leaves out, and how corrections are handled. Method version bpi-1.0.
What is measured
The index measures what carriers are paid to move one vehicle on a corridor between two metro areas, in U.S. dollars. Payments are converted to a sedan equivalent (SUVs ÷ 1.10, pickups and luxury ÷ 1.15, the same factors the quote engine uses) so lanes with different vehicle mixes compare fairly. Open and enclosed transport are reported separately.
Data sources and classes
Every record carries a source class. Only class A moves the index.
- A. Completed carrier payment: a delivered Bigfella shipment with the carrier payment recorded, or a carrier payment recorded by Bigfella staff.
- B. Accepted carrier offer: a carrier accepted the load at that pay. Published as a separate, clearly labelled series; never mixed into the index.
- C. Submitted carrier offer, D. pricing-engine estimate, E. customer quote, F. booked customer price: counted for context only, never averaged into the index.
Windows and publication rule
Lane averages use a rolling 13-week window, with a 4-week average shown beside them. A lane is published when it has at least 10 qualifying completed payments in 13 weeks from at least 3 different carriers, with no single carrier above half of them. With 5–9 payments it is labelled limited data. Below that we show “Building historical data for this route.” Missing lanes are never estimated, back-filled or set to zero.
Index and national basket
Each lane’s index starts at 100 on its first published 13-week average and moves with it after that. The national index uses a fixed basket of ten corridors, three of which touch Houston, and is reported two ways: a simple mean of the published basket lanes and a volume-weighted mean (weighted by payments). Because Bigfella’s volume is concentrated around Houston, the two can differ; both are shown so the weighting is never hidden. Week-to-week movement is reported only for the national index; weekly lane numbers on thin data would be noise.
Cleaning
Each payment is counted once (a staff-recorded payment that repeats a delivered shipment is removed). With five or more payments on a lane, values outside 1.5× the interquartile range are excluded as outliers. Rows without a payment, date, or a recognised metro at both ends are left out and counted in the release’s source counts.
Carrier confidentiality
Only aggregated lane figures are published. The three-carrier minimum and the 50% cap exist so that no single carrier’s pay can be worked out from the numbers. Shipment-level data is never public.
Releases, sign-off and corrections
A draft release is computed every week from the database. It is published only after sign-off, with a named methodology owner and a named corrections-log owner recorded on the release. Published numbers never change: releases are append-only and each carries a data fingerprint so it can be reproduced. If a figure was wrong, a correction is added to the public corrections log with the reason and date.
Other indexes
Third-party price indexes are not used in these averages. If an external benchmark is ever shown, it will be labelled separately with attribution and only where its license allows republication.