Modern approaches to assessing the residual value of restored cars in the US dealer market.

The American Journal of Management and Economics Innovations

68

https://www.theamericanjournals.com/index.php/tajmei

TYPE

Original Research

PAGE NO.

68-75

DOI

10.37547/tajmei/Volume07Issue05-08

OPEN ACCESS

SUBMITED

20 March 2025

ACCEPTED

25 April 2025

PUBLISHED

20 May 2025

VOLUME

Vol.07 Issue 05 2025

CITATION

Ruiev Mykola. (2025). Modern approaches to assessing the residual
value of restored cars in the US dealer market. The American Journal of
Management and Economics Innovations, 7(05), 68

–

75.

https://doi.org/10.37547/tajmei/Volume07Issue05-08.

COPYRIGHT

© 2025 Original content from this work may be used under the terms
of the creative commons attributes 4.0 License.

Modern approaches to
assessing the residual
value of restored cars in
the US dealer market.

Ruiev Mykola

CЕО & Founder CAR TRADE GROUP LLC, Florida, USA

Florida, USA

Abstract:

This study aims to provide a comprehensive

review of contemporary approaches to estimating the
residual value of rebuilt vehicles in the U.S. dealer
market. The relevance of this work is driven by the scale
and dynamics of the auction segment for rebuilt and
salvage vehicles, whose revenue rivals that of the
midsize new-car market, and whose online segment is
projected to grow at a compound annual rate of 13.7%
through 2030. The novelty of the research lies in
bringing together various types of data. This data
comes from VIN parsing, operational logs from Copart,
IAA, OpenLane, telematics streams coming from
Geotab, and high-precision visual checks. It has been
observed that the shift from expert visual appraisal to
digital valuation methods has significantly reduced the
mean absolute error in RV forecasts, while also
shortening the days-to-sale and decreasing cosmetic
damage arbitration rates. UVeye, Mitchell+PAVE, and
Ravin AI systems will incorporate models with visual
features, while conformal quantile regression will
ensure guaranteed coverage, enabling automatic
adjustments to financial terms. However, new risks
have emerged: the increase in deepfake manipulations
of photographic content, alongside regulatory
requirements (SB-362), will impose very stringent
demands on verification as well as the protection of
personal and telematics data.
This article will be of use to analysts, dealers, AI solution
developers, and researchers in the fields of residual
value estimation and risk management in the used-
vehicle market.

Keywords:

residual value, rebuilt vehicles, U.S. dealer

The American Journal of Management and Economics Innovations

69

https://www.theamericanjournals.com/index.php/tajmei

market, computer vision, telematics, big data,
ensemble regressors, conformal quantile regression,
risk management, data verification.

Introduction:

The volume of auction sales of rebuilt

vehicles in the United States is comparable to that of
the midsize new-car segment: in 2023, combined

“whole

-

car + salvage” auction revenues totaled USD

3.32 billion, with 13.84 million units transacted, a
significant share of which held rebuilt or salvage titles
[1]. By 2024, the online segment of salvage auctions
alone is estimated to be worth USD 3.42 billion, and
Grand View Research projects a compound annual
growth rate of 13.7% through 2030 [2]. The magnitude
and momentum of this sector make the accuracy of
residual-value (RV) estimation a central factor in the
financial stability of dealers who handle these vehicles.

For dealer operations, an error in RV prediction
translates directly into financial loss: overestimation
leads to losses upon resale, while underestimation
results in excessive discounts and tied-up capital. In a
volatile used-car market, dealers aim to narrow this
forecast corridor and accelerate inventory turnover.
The transition from expert-based, paper-driven
appraisal methods to the integration of big data and
artificial intelligence algorithms offers a path to
reducing uncertainty. Platforms such as ACV Auctions
scan vehicles upon intake using computer vision,
extract structured features, and

—

leveraging a

database of over 3 million inspections

—

generate an

instant report that reduces arbitration rates and
mitigates transaction risk [3].

Thus, the evolution of valuation tools is shifting residual
analytics from subjective allowances to quantitative
modeling, creating a new fulcrum for risk management
in the rapidly expanding U.S. restored-vehicle market.

MATERIALS AND METHODOLOGY

This study of contemporary approaches to residual-
value estimation for rebuilt vehicles in the U.S. dealer
market is based on an analysis of 21 sources, including
auction platform operational logs, telematics provider
reports, computer vision implementation case studies,
and publications on quantitative risk estimation
methods. The theoretical framework comprises works
on the integration of computer vision and big data in RV
analytics: Hari Bhushan [3] described the application of

CV inspection at ACV Auctions; Manokhin [10]
presented a conformal quantile-regression method for
reliable interval forecasting; and Jung et al. [15]
compared YOLOv7 and YOLOv8 effectiveness in
detecting div defects. Industry reports from Geotab
on telematics [8, 9], technical documentation for
UVeye Atlas/Helios [11, 12], the Mitchell + PAVE mobile

solution [13], and OPENLANE’s Visual Boost case

studies [14] were also reviewed.

Methodologically, the study combined:

●

Comparative analysis of visual technologies

—

juxtaposing stationary portals (UVeye Atlas/Helios)
and Copart C360 panoramic imaging with mobile
apps (Mitchell + PAVE) and the hybrid Ravin AI
solution; assessing the impact of feature-quality on
RV-forecast accuracy [5, 11, 14].

●

Systematic review of big data

—

aggregating Copart,

IAA, and OPENLANE operational logs with VIN
parsing and processing of 40 configuration
attributes [4], alongside Geotab telematics streams
to account for wear and battery condition [8, 9].

●

Content analysis of operational metrics and case
studies

—

examining the influence of AI valuation on

arbitration reduction and evaluating model
sensitivity to macroeconomic factors and salvage-
title types.

RESULTS AND DISCUSSION

Traditional practice for valuing rebuilt vehicles relied on
expert visual inspections. It averaged insurance tables:
structural-geometry losses were recorded manually,

and the discount relative to a “clean” analogue was

determined empirically by experts. The first step
toward reducing subjectivity was extensive digitization.
The wide adoption of VIN parsing and the integration of
telematics streams have enabled the automated
extraction of configuration codes, service and mileage
history, and current electronic control-unit errors.
Today, aggregators process data from 145 online
auctions in real-time, including price, mileage, and 40
configuration attributes per VIN [4]. Concurrently,
Copart introduced its own 360° C360 module in 2021,
complementing the existing set of twenty high-
resolution HD frames for an objective visual channel
[5]. The growth of telematics closely mirrored the
expansion of datasets: the automotive telematics

The American Journal of Management and Economics Innovations

70

https://www.theamericanjournals.com/index.php/tajmei

market is projected to rise from USD 9.87 billion in 2024
to USD 17.24 billion by 2030 at a 9.84% CAGR [6].

In parallel, marketplace infrastructures integrated data
collection and transactions. Copart, handling over 4
million lots in fiscal 2024 across 250 locations, has
become the primary distribution channel for total-loss
vehicles, where standardized content is the norm [7].
Thus, subjective expert practice has evolved into a
continuous digital stream model, in which the quality of
incoming data sets the lower bound of RV-forecast
error.

Three major data streams constitute the input for RV
models. The first comprises operational logs from
Copart, IAA, and OpenLane, where each listing may
include up to 100 parameters, including type of
damage, VIN attributes, bidding history, and final sale

price. The sector’s scale provides

the statistical

foundation: the U.S. vehicle auction market reached
13.84 million units in 2023, with a projected annual
growth rate of approximately 3.7% from 2023 to 2028
[22].

The second stream originates from insurers.
Aggregated databases, which cover a significant share
of the auto-property insurance market, provide data on
damage types, total-loss calculations, and actual repair
costs, enabling the alignment of forecasted and
realized recovery expenses. These variables are critical
for adjusting

the “insurance” discount, as they

differentiate total losses due to non-structural (hail,
flood) versus structural (collision) damage.

The third source is telematics providers. According to
Geotab analytics, by 2024, approximately 340 million
vehicles were connected to its cloud platform [8], with
the EV fleet within that network growing 63% year-
over-year and surpassing 700 million real-world miles
driven [9]. Continuous odometer, usage-mode, and
battery-condition streams expand feature sets beyond

the “moment of accident,” allowing models to account

for material wear and powertrain component
degradation.

Training datasets remain heterogeneous, as each VIN is

annotated with its legal status (“salvage,” “rebuilt,”
“flood,” “hail,” or “theft recovery”), regi

onal lot

location, and macroeconomic context. Macro factors

inject volatility: for instance, an FOMC rate cut
immediately reduced the average used-car loan
payment, altering dealer demand elasticity in interest-
sensitive segments. Consequently, the final dataset is
augmented with time-localized variables (Fed Funds,
CPI, WTI) and an export-orientation indicator, while
title-type labeling encodes a priori discount
differentials.

Since dealers are concerned not only with point
forecasts but also with deviation ranges, conformal
quantile regression (CQR) is applied on top of ensemble
models to compute confidence intervals. This method
trains two auxiliary boosting models on the 0.05 and
0.95 thresholds and calibrates them on a hold-out
sample, ensuring 90% coverage even under shifting
market regimes [10]. The resulting interval width serves
as a natural risk score: lots with a bandwidth greater
than 18% trigger an automatic increase in financing
rate. Within the ACV ArbGuard ecosystem, this scoring
has already reduced cosmetic-damage arbitrations and
consequent seller losses.

Thus, the “heterogeneous dataset → ensemble
regressor → conformal calibration” pipeline converts

RV estimation from a static expert procedure into
dynamic risk management, adaptable to both macro
shifts and micro-demand fluctuations across salvage-
title types.

Computer vision completes the “data–

valuation

–price”

chain, formalizing visual inspection into a numerical-
feature source. Among stationary solutions, drive-
through portals lead the field: the UVeye Atlas/Helios
scanner, deployed at 300 U.S. dealer locations,
captures up to 2,000 multispectral frames per pass and
generates a full-div condition report in 20

–

30

seconds, processing over 0.5 million vehicles monthly
[11]. Its eight high-resolution cameras and laser
profilometer array detect thickness deviations that
previously required panel disassembly. The algorithm
immediately segments the div into standard
elements (hood, doors, quarters, bumpers, sills) and
assigns quantitative metrics

—

area, depth, and defect

type

—

to each zone. These vectors feed into the

ensemble RV regressor, where the “visual” feature

contribution reduces RMSE. An example of a UVeye
installation is illustrated in Fig. 1, and a corresponding
report is presented in Fig. 2.

The American Journal of Management and Economics Innovations

71

https://www.theamericanjournals.com/index.php/tajmei

Fig. 1. UVeye scanner [12]

Fig. 2. Example of a report after scanning via UVeye [12]

Mobile solutions address the same tasks where no
stationary portal exists. The Mitchell + PAVE alliance is
most illustrative: in September 2024, partners
integrated a guided-capture app into the Mitchell
Intelligent Damage Analysis cloud, enabling an eight-to-

ten photo smartphone condition report that instantly
calculates parts, labor, and tax costs [13]. This format
benefits small dealers and insourced fleets for which a

physical drive‐through is not cost

-effective.

The American Journal of Management and Economics Innovations

72

https://www.theamericanjournals.com/index.php/tajmei

Complementing stationary and mobile approaches is
the hybrid Ravin AI solution on the OPENLANE
marketplace. Visual Boost overlays thermal masks of
detected defects directly onto condition-report

photographs, eliminating the typical “difficult” angles

and micro-cracks that are invisible to the average buyer
[14]. An example of this technology is shown in Fig. 3.
Integration increased conversion rates on rebuilt-lot
bids, confirming the hypothesis that informational
deficits drive elevated salvage-title discounts.

Fig. 3. Example of OPENLANE Visual Boost AI [14]

Across these solutions, segmentation and hidden-
damage detection algorithms play a key role. For large
panels, CNN hybrids with panoptic segmentation
partition images into superclasses and localize defect

“fibers” with sub

-millimeter precision, while also

producing corrosion-probability heatmaps from
indirect texture cues. Smartphone workflows in

Mitchell + PAVE employ a “YOLOv8 → GAN

-

augmentation” scheme. Jung et al. [15] report that

upgrading from YOLOv7 to YOLOv8 increased mAP on
div-defect detection from 0.84 to 0.88 while
maintaining real-time performance. To account for
uncertainty, detection results undergo a time-based
Canny-based ensemble and are calibrated with
conformal predictive intervals; that interval width then
factors into the dealer-financing risk score described
above.

Thus, the spectrum of CV technologies

—

from

millimeter-precision portals to smartphone apps

—

forms a data hierarchy fed into a unified RV model. The
richer the segmentation map and the more accurate

the “invisible” corrosion p

rediction, the narrower the

price-

confidence interval and the lower the dealer’s

financial risk on rebuilt vehicles.

Risk reduction emerges as early as the bidding stage,
where ensemble CV-augmented models enable dealers
to narrow the gap between forecasted and actual
salvage-lot purchase prices. A study [16] demonstrated
that AI-based valuation reduced the mean absolute
error by 13.5% compared to traditional statistical
regression methods. Operational risk

—

typically linked

to arbitrations and returns due to undetected defects

—

has also declined: ACV Auctions’ ArbGuard inspection

platform, powered by over 3 million historical
inspections and employing computer vision with
acoustic analysis, predicts hidden-damage probabilities
and, according to the 2023 Analyst Day report, has
already reduced cosmetic arbitration volumes by
automatically highlighting risk zones at listing [17].

Financial risk is measured by inventory turnover and
ultimate profitability. Reducing forecast error and

The American Journal of Management and Economics Innovations

73

https://www.theamericanjournals.com/index.php/tajmei

arbitration frequency directly accelerates days-to-sale:
dealers using integrated AI valuation systems turn
inventory roughly five days faster than those relying on
traditional pricing, even before factoring in lower floor-
plan costs.

The effect scales under tight market conditions:
according to Cox, the U.S. used-car supply fell to 39
days of supply by early April 2025

—

the lowest in four

years. Consequently, precise pricing, lower claims
costs, and faster turnover collectively mitigate the
systemic risks historically associated with r

ebuilt‐

vehicle transactions, transforming the segment from
high-risk to sustainably margin-accretive even amid
macroeconomic volatility [23].

While RV-model accuracy curbs classic pricing risks, it
introduces new vulnerabilities tied to data authenticity.
Insurer Allianz reported a 300 % increase in claims

based on artificially “added” div damage—

that is,

shallow/deepfake photo manipulation

—

and Zurich UK

labeled such fraud “one of the fastest

-

growing threats”

to auto insurance [18].

At the same time, the U

.S. Treasury’s FinCEN issued an

advisory on schemes utilizing generative imagery and
documents to circumvent Know Your Customer (KYC)
procedures, highlighting a rise in suspicious
transactions involving deepfake content in banking and
insurance filings. The corporate sector is also alarmed:
in mid-2024, Microsoft publicly urged Congress to
enact federal deepfake-fraud legislation, warning that
the absence of synthetic-media labeling would escalate
financial-market risks [19]. For dealers, this implies
poten

tial replacement of genuine “x

-

ray scans” of

vehicle bodies with fabricated images, leading to
erroneous lot valuations.

The expansion of datasets also raises concerns about
privacy. Telematics streams record precise routes and
usage modes, and photographic material often
contains personal identifiers. In the legal realm,

California’s Delete Act (SB

-362), effective January 2024,

mandates that registered data brokers

—

including

auction aggregators and telematics providers

—

offer a

unified mechanism for consumers to request deletion
of personal data [20]. Consequently, dealers and their
technology partners must balance the completeness of
information for accurate pricing models with new

obligations to minimize and purge personal attributes.

The systemic response to these dual threat vectors is
the development of end-to-end verification standards.
At the media-source level, the open C2PA specification,
jointly advanced by Adobe, Microsoft, Intel, and others,

defines a cryptographic “Content Credentials”

architectu

re to record a file’s transformation chain

from capture to publication [21].

To summarize, the application of big data and
computer vision to high-precision RV-estimation
algorithms significantly reduces financial risks for
dealers in the restored-vehicle market. These
technologies not only improve forecast accuracy but
also enable faster vehicle turnover, which is crucial in
highly dynamic markets. On the downside, the new
benefits also introduce threats from data manipulation
and compliance with privacy requirements. The future
development and implementation of universal
verification and data protection mechanisms will be
crucial for the sustainable growth of this market
segment.

CONCLUSION

A part of the automotive business that has grown in
recent years is the rebuilt-vehicle market, especially in
the U.S., where proper detailing and residual value
estimation are crucial for financial stability among
dealers. Recent approaches to RV valuation have
shifted from subjective visual inspections and
judgments by one or two experts to greater objectivity
and quantification of accuracy, made possible through
big data analytics and artificial intelligence. Computer
vision, combined with telematics, makes residual value
forecasting more accurate

—

this enables faster

processing and reduces handling costs.

Group regressors, combined with conformal quantile
regression and information from bids, sellers, and
telematics companies, yield a high-precision risk
assessment (RA) model. This model considers not just
the true condition of the vehicle but also prospective
larger economic changes so it can be used in unstable
market conditions.

Computer-vision

implementations

—from

UVeye’s

stationary portals to guided-capture mobile apps

—

convert visual assessments into standardized,

The American Journal of Management and Economics Innovations

74

https://www.theamericanjournals.com/index.php/tajmei

numerical inputs. This standardization, in turn,
significantly reduces valuation error, bolsters market
participant confidence, lowers arbitration rates, and
speeds up inventory turnover. Such advances enable
dealers to forecast RVs more accurately and improve
the profitability of rebuilt-vehicle operations.

However, alongside these clear advantages, new RV-
estimation methods carry risks, primarily related to
potential data-manipulation schemes, such as
deepfake imagery, and privacy challenges in handling
telematics and visual data. The threats will necessitate
the establishment of robust data-verification and
protection

standards

that

ensure

transaction

transparency and security. Emerging technologies, like
blockchain, play an important role in keeping market
resilience while fostering sustainable growth. To sum
up, the use of modern RV-estimation methods in dealer
workflows reduces typical price risks and lays a solid
groundwork for additional growth and improvement
within the U.S. rebuilt-vehicle market.

REFERENCES

“US Vehicle Auction (Whole Car, Salvage) Market

Analysis Report 2023-

2029,”

Globe Newswire

, Aug. 07,

2024.

https://www.globenewswire.com/news-

release/2024/08/07/2925845/28124/en/US-Vehicle-
Auction-Whole-Car-Salvage-Market-Analysis-Report-
2023-2029-Hike-in-Auction-Pricing-Strong-
International-Demand-Integration-of-AI-Change-in-
Consumer-Behaviours-Indicate-S.html

(accessed Mar.

27, 2025).

“The United States Online Salvage Auctions Market Size
& Outlook, 2030,”

Grand View Research

, 2025

—

https://www.grandviewresearch.com/horizon/outlook
/online-salvage-auctions-market/united-states

(accessed Mar. 28, 2025).

Hari Bhushan, “New ACV Inspection Technology,”

ACV

Auctions

,

Jul.

06,

2023.

https://www.acvauctions.com/blog/new-acv-
inspection-technology

(accessed Mar. 29, 2025).

A. Campbell, “Auction Listings Data Feed,”

Market

Check

,

Aug.

30,

2024.

https://www.marketcheck.com/data_feed/active-
auction/

(accessed Mar. 30, 2025).

K. Langley, “Copart Launches 360° Imaging for Vehicles
in

Auction,”

Copart

,

2025.

https://remarketing.copart.co.uk/2021/01/26/copart-
launches-360-imaging/

(accessed Mar. 30, 2025).

“Automotive Telematics Market Value, Growth Analysis
By 2024 to 2030,”

GlobeNews Wire

, Nov. 07, 2024.

https://www.globenewswire.com/news-
release/2024/11/07/2976668/0/en/Automotive-
Telematics-Market-Value-Growth-Analysis-By-2024-
to-2030-Exactitude-Consultancy.html

(accessed Mar.

31, 2025).

“Copart Reports Fourth Quarter Fiscal 2024 Financial
Results,”

Auto Recycling World

, Sep. 09, 2024.

https://autorecyclingworld.com/copart-reports-
fourth-quarter-fiscal-2024-financial-results/

(accessed

Apr. 01, 2025).

“Connected vehicle data revolution,”

Geotab

, Apr. 20,

2022.

https://www.geotab.com/uk/white-

paper/connected-vehicle-data-revolution/

(accessed

Apr. 03, 2025).

F. Youd, “Navigating the net zero landscape with
Geotab,”

Just Auto

, Apr. 09, 2025.

https://www.just-

auto.com/interview/navigating-the-net-zero-
landscape-with-geotab/?cf-view

(accessed Apr. 10,

2025).

V. Manokhin, “Conformalized Quantile Regression:
Smarter Uncertainty Prediction for Data Scientists,”

Medium

,

Mar.

22,

2025.

https://valeman.medium.com/conformalized-
quantile-regression-smarter-uncertainty-prediction-
for-data-scientists-6389bea7a7c4

(accessed Apr. 06,

2025).

C. Stokel-

Walker, “Scanning Cars in Seconds,”

Time

,

Oct. 30, 2024.

https://time.com/7094830/uvey-

vehicle-inspection-systems/

(accessed Apr. 11, 2025).

“Automatic Vehicle Inspection During COVID

-19 -

UVeye,”

UVeye

,

Mar.

23,

2020.

https://www.uveye.com/automatic-vehicle-
inspection-during-covid-19/

(accessed Apr. 12, 2025).

S. D. Calif, “Mitchell Expands Its Intelligent Open
Platform,”

Mitchell

,

Sep.

09,

2024

—

https://www.mitchell.com/insights/news-

The American Journal of Management and Economics Innovations

75

https://www.theamericanjournals.com/index.php/tajmei

release/auto-physical-damage/mitchell-expands-
support-for-ai-enabled-vehicle-inspections

(accessed

Apr. 13, 2025).

A. K. Leichman, “Car inspection tech adopted by major

US vehicle marketplace -

ISRAEL21c,”

ISRAEL21c

, Feb.

13, 2024.

https://www.israel21c.org/car-inspection-

tech-adopted-by-major-us-vehicle-marketplace/

(accessed Apr. 14, 2025).

D.-Y. Jung, Y.-J. Oh, and N.-

H. Kim, “A Study on GAN

-

Based Car Body Part Defect Detection Process and
Comparative Analysis of YOLO v7 and YOLO v8 Object

Detection Performance,”

Electronics

, vol. 13, no. 13, pp.

2598

–

2598,

Jul.

2024,

doi:

https://doi.org/10.3390/electronics13132598

.

Miller V, “Transforming Automo

tive Valuation: The

Role of AI in Residual Value Prediction,”

TechBullion

,

Mar. 20, 2025.

https://techbullion.com/transforming-

automotive-valuation-the-role-of-ai-in-residual-value-
prediction/

(accessed Apr. 15, 2025).

ACV Auctions, “ACV’s 2023 Analyst Day Showcases

L

atest Data & Technology Innovations,”

ACV Auctions

,

Jun. 08, 2023.

https://www.acvauctions.com/blog/acv-

2023-analyst-day

(accessed Apr. 16, 2025).

R. Jones, “Fraudsters editing vehicle photos to add fake
damage in UK insurance scam,”

The Guardian

, May 02,

2024. Accessed: Apr. 17, 2025. [Online]. Available:

https://www.theguardian.com/business/article/2024/
may/02/car-insurance-scam-fake-damaged-added-
photos-manipulated

M. Tatananni, “Microsoft warns against ‘deepfake

fraud’ and deceptive AI as the company begs
government leaders,”

The Sun

, Jul. 31, 2024.

https://www.thesun.co.uk/tech/29599921/microsoft-
ai-deepfake-warning-report-government-action/

(accessed Apr. 20, 2025).

“SB

-362 Data broker registration: accessible deletion

mechanism,”

California Legislative Information

, 2023.

https://leginfo.legislature.ca.gov/faces/billNavClient.x
html?bill_id=202320240SB362

(accessed Apr. 19,

2025).

M. Mesa-Simón, A. Escobar-Molero, B. Sáez-
Mingorance, D. P. Morales, J. A. Álvarez-Bermejo, and

F. J. Romero, “Enabling Live Video Provenance and

Authenticity: A C2PA-Based System with TPM-Based

Security for Livestreaming Platforms,” Mar. 2025, doi:

https://doi.org/10.36227/techrxiv.174197970.096668
99/v1

.

“US Vehicle Auction (Whole Car, Salvage Vehicle) Size,

Trends and Forecast to 2029,”

Business Wire

, Apr. 13,

2024.

https://www.businesswire.com/news/home/2024081
3107531/en/US-Vehicle-Auction-Whole-Car-Salvage-
Vehicle-Size-Trends-and-Forecast-to-2029-%244.13-
Billion-Market-Analysis-By-Volume-Type-Distribution-
Channel---ResearchAndMarkets.com

(accessed Apr.

26, 2025).

S. Wanner, “Used

-Vehicle Inventory Continues to

Tighten as Sales Surge in March -

Cox Automotive Inc.,”

Cox

Automotive

Inc.,

Apr.

11,

2025.

https://www.coxautoinc.com/market-insights/used-
vehicle-inventory-march-2025

(accessed Apr. 28,

2025).