How we calculate diminished value

The principle

Diminished value is the difference between what your vehicle would sell for with a clean history and what it will sell for now that a repaired collision appears on its record. Everything below is an attempt to estimate that difference from facts you can document.

Three constraints shaped the design.

1. Every term has to be explainable. The output goes in front of a claims adjuster who has to be able to follow it and argue with it on the merits. That rules out a regression with opaque coefficients, and it is why each factor in the report carries a written rationale rather than just a multiplier.
2. The curve has to saturate.Loss as a share of value cannot rise without bound. Past a point, a vehicle is simply "an accident car" and further repair spending stops moving the resale discount. A linear percentage-of-repair-cost rule, which several free calculators use, produces implausible numbers on expensive repairs.
3. It has to beat 17c on the merits, not by inflation. The insurer's formula is attackable because it caps loss at 10% regardless of the vehicle and zeroes out at 100,000 miles. Answering it with an inflated number would be just as indefensible, which is why there is a hard ceiling and an explicit conservative figure.

The calculation

The whole model is one multiplicative chain:

DV = pre-loss value
× base loss rate (from repair-cost-to-value ratio)
× damage-type factor (severity, structural, airbag, area)
× age factor
× mileage factors
× segment factor
× prior-history factor
× repair-quality factor
× disclosure factor
× pre-loss condition factor

The result is then blended with any market evidence you supply, capped, and rounded to a granularity that matches its magnitude, so the output reads as a considered figure rather than a raw float.

The base loss rate

The strongest single predictor is repair cost as a share of the vehicle's pre-accident value. It is also the signal a prospective buyer reads directly off a vehicle history report, which is what makes it defensible rather than merely convenient.

The curve is:

base rate = 0.27 × (1 − e^(−4.4 × ratio))

It saturates at 27%, approached asymptotically. Values sampled directly from the running engine:

The factors

Damage type

Severity multiplier, cosmetic 0.62, moderate 0.88, major 1.06, severe 1.18, compounded with 1.30 where structural or unibody repair was required, 1.12 where airbags deployed, and an area adjustment (1.06 for roof or undercarriage, 1.02 for side impact, 0.97 for rear-only). The product is clamped to the range 0.55–1.85 so compounding cannot run away.

Structural repair carries the largest single multiplier because it is the most heavily discounted item on any history report. Buyers and inspectors treat unibody work as permanent regardless of what the shop achieved.

Age

1.06 × e^(−0.052 × age), clamped to 0.38–1.06. Newer vehicles lose proportionally more because their buyers have the widest choice of clean alternatives and are paying a premium for as-new condition.

Mileage

Measured against expectation rather than absolutely: 1.0 + 0.16 × (1 − actual ÷ expected), where expected is 12,000 miles per year, clamped to 0.76–1.14. A five-year-old car with 30,000 miles scores ×1.080; the same car with 90,000 miles scores ×0.920.

A separate absolute-odometer step applies 0.95 past 100,000 miles and 0.88 past 150,000, because the used market reacts to those thresholds independently of age.

Market segment

Exotic 1.22, luxury 1.12, strong-retention mainstream 1.06, mainstream 1.00, high-supply value 0.92. Assigned by make, with model-level overrides where a nameplate behaves differently from its brand (a Corvette is not a Chevrolet Malibu). Premium and enthusiast buyers cross-shop many clean examples and screen harder on history.

Prior accident history

None 1.00, one prior 0.74, two or more 0.56. Only the incremental loss caused by this accident is claimable. Claiming the full step from clean on a car that already had a record overstates the loss and is trivially easy for an adjuster to defeat.

Parts and repair quality

Parts: OEM 0.96, unknown 1.00, aftermarket 1.07, used 1.09. Finished quality: flawless 0.97, minor imperfections 1.06, visible defects 1.16. The product is clamped to 0.90–1.28.

Disclosure

Visible on a history report 1.00, unconfirmed 0.94, not showing 0.80. Diminished value depends on buyers being able to see the accident. Until it appears, an adjuster has a real argument that no market discount exists, and they are right, which is why this factor is substantial.

Pre-loss condition

Excellent 1.04, good 1.00, fair 0.93. A car at the top of its market band had the furthest to fall.

Comparable listings

Where you supply listings, each is adjusted to your vehicle before being used:

• Mileage: (comp mileage − your mileage) × rate, where the rate is pre-loss value × 0.0000085 clamped to $0.04–$0.35 per mile. A mile costs more on a $70,000 car than a $9,000 one.
• Model year: (your year − comp year) × 9% of pre-loss value.

The clean-history mean and the accident-branded mean are then compared and the spread is read as a market-observed diminished value. It is only used when there are at least two of each, a single pair is one seller's pricing decision, not a market signal.

When usable, the market figure is blended at 40% against 60% for the model. Claimant-supplied comps are a small, self-selected sample; the model is stable but generic. That weighting lets real evidence move the number without letting one unusual listing swing it.

Caps and ranges

• Hard ceiling: 33% of pre-loss value. Applied regardless of what the factors produce, so the stated loss stays inside the range the retail market supports for a repaired, non-salvage vehicle.
• Conservative figure: 78% of the supported figure. What you should be able to settle at without argument.
• Upper figure: 124%, also subject to the ceiling. Your opening ask.
• Viability floor: $400. Below this we tell you the claim is probably not worth your afternoon, rather than selling you a packet for it.

Confidence scoring

Every estimate carries a score out of 100, because the same method applied to poorly documented inputs deserves less weight:

• 30 points, pre-loss value from a published guide rather than your own estimate
• 25 points, a documented repair cost
• 25 points, usable comparable listings (8 partial)
• 20 points, the accident confirmed on a history report (8 if unchecked)

The calculator tells you which points you are missing and how to get them, rather than presenting a precise-looking number built on guesses.

What this method cannot do

Stated plainly, because a method that overclaims is easy to discredit.

• It is not a licensed appraisal. We do not inspect the vehicle and we are not licensed appraisers. Where your insurer or your state requires an appraisal, we tell you and point you to one.
• It depends entirely on your inputs. Condition, mileage, repair cost and damage particulars are as you report them.
• It is an estimate of market loss, not a payout. No insurer is obliged to accept it. Settlements vary with liability, jurisdiction and negotiation.
• It does not model local market conditions. Regional supply and demand genuinely move used prices; the method does not capture that, and supplying local comparables is how you correct for it.
• It does not use licensed valuation data.The pre-loss value comes from you, sourced from a published guide. We do not resell anyone's book values.

Versioning

The method is at revision 1.0.0 and the state rules table at revision 1.0.0, compiled as of 2026-07-28. Both are printed in the footer of every report.

Stored estimates keep the full engine output from the moment they were calculated. If a coefficient changes later, a document you already sent to an insurer does not silently change underneath you, which matters if an adjuster comes back to it six weeks after you sent it.