The Resale Value
Methodology

How we calculate a fair price.

Every valuation number on this site is derived from real transaction data using the process described below. This page is intended to be complete enough that a statistically-trained reader could reproduce our approach.

Last updated: July 2026

1. Data source

Our exclusive data partner is Cars24, India's largest used-car platform. Every valuation on this site is grounded in Cars24's inspection and transaction dataset:

  • 10M+ vehicle inspections conducted by Cars24-trained inspectors across 200+ Indian cities
  • Actual transaction prices (not asking prices) at which vehicles changed hands
  • 140+ inspection parameters per vehicle: engine, body, tyres, service history, ownership, city
  • Rolling 24-month window to keep the model responsive to current market conditions

We disclose the data partnership openly on every page of this site and on our about page. Independence is enforced by contract and by our editorial policy.

2. The pricing model

Each vehicle's fair-market band is derived from a two-stage model:

Stage 1: Comparable transactions

We identify the 500–3,000 most similar recent transactions to your vehicle by make, model, year (±1), variant, kilometrage bucket, ownership count, and city cluster.

Stage 2: Regression adjustment

A gradient-boosted regression corrects for residual factors: exact kilometrage, month of sale, city-level demand, and inspection score. Output is a mean expected price plus a prediction interval.

The bounds shown on your result page represent the 10th and 90th percentiles of the prediction distribution. That means 80% of comparable transactions historically closed within this band.

3. Confidence bands

Confidence is a proxy for how densely populated your vehicle's segment is in our dataset:

High
1,000+ comparable transactions in the last 6 months. Typically popular, mainstream models under 8 years old.
Medium
200–1,000 comparables. Older mainstream models or newer niche variants.
Lower
Under 200 comparables. Rare models, very old vehicles, or luxury cars. The band widens accordingly.

4. Refresh cadence

Our model is retrained every Sunday on the previous 24 months of rolling data. Model version and training-set-size are visible in the site footer. Corrections and material model changes are logged publicly on this page.

5. What we don't do

  • We do not recommend where you should buy or sell your vehicle. We do not route you to dealers or marketplaces.
  • We do not charge for valuations. We do not collect personal information at the point of valuation.
  • We do not adjust prices in favour of any listing partner. Our data partnership with Cars24 does not influence output pricing, enforced by our independence agreement.
  • We do not offer guarantees on the price you will actually get. Real-world outcomes depend on your specific vehicle's condition and negotiation.

6. Known limitations (v1)

This site is in early access. Two limitations we are transparent about right now:

  1. Our current MVP uses a simplified pricing heuristic. Production integration with the full Cars24 model launches shortly.
  2. We cover cars first. Two-wheelers, insurance IDV, and loan-vs-resale tools are being built. See the homepage for the roadmap.

Found an error?

If you believe a valuation is materially wrong for a specific make or model, we want to know. Corrections are logged publicly on this page.

Report a correction →