Executive summary
CarCast's Q3 2026 forecast projects that 10 tracked used-vehicle segments will lose meaningful value over the next 8 weeks as the market enters fall. The signals cluster around three patterns: older full-size trucks and body-on-frame SUVs giving back the last of their post-pandemic premium, a late-model electric crossover absorbing new-model-year price cuts on comparable current-year vehicles, and a mid-luxury sedan continuing to lose walk-in share to luxury SUVs.
The steepest projected drop is the 2022 Ford Mustang Mach-E Select at -4.92% over the next 8 weeks — roughly $1,366 off the current median. All ten segments in this report are projected to decline between 1.35% and 4.92%.
Against that softening, five segments are moving the opposite way. The strongest projected gainer is the 2019 Mazda CX-5 Sport at +5.31% — a reminder that "used market" is not one market. Three of the five gainers are mid-size pickups, consistent with continued category-demand strength for body-on-frame utility.
Forecasts here use CarCast's 8-week horizon with P10 / P50 / P90 confidence bands and reflect data as of June 30, 2026.
Methodology sidebar
- Data sources. Real dealer listings via MarketCheck (segment-level medians), Cox Automotive's Manheim Used Vehicle Value Index (MUVVI, wholesale), and Bureau of Labor Statistics used-vehicle CPI.
- Forecast horizon. 8 weeks (P50 median with P10 / P90 confidence band).
- Model. Proprietary AI forecasting model retrained weekly on trailing history plus wholesale and macro signals.
- Confidence. Reported per segment; derived from the P10–P90 band width relative to the P50.
- Disclosed accuracy. 3-5% mean absolute error across tracked segments. See /accuracy for the current audit.
- Data-quality bar (Q3 2026). At least 12 weeks of continuous tracking history and at least 100 average weekly listings. Q3 2026 is CarCast's first quarterly report; the standing spec (26 weeks + 100 listings) becomes the bar starting Q4 2026 when more segments will have accumulated the full history.
The 10 used vehicles most likely to depreciate this fall
Selected as the 10 tracked segments with the largest projected 8-week percentage decline that also meet CarCast's Q3 2026 data-quality bar.
#1. 2022 Ford Mustang Mach-E Select
2022 was the Mach-E's peak inventory year at Ford, and Ford has cut MSRP on newer Select trims more than once since — used values track new pricing on a lag. Tesla Model Y price cuts continue to compress the small-electric-crossover segment, and the P50 lands below the 12-week actual because that pressure hasn't fully filtered through the medians yet.
See the full forecast for the 2022 Ford Mustang Mach-E Select→
#2. 2020 Chevrolet Silverado 1500 LT
2020 was the last model year of the pre-refresh Silverado interior, and half-ton trucks past the five-year mark typically lose their fleet-buyer premium first. The P10/P90 band stays tight around the P50 — the model reads this as a controlled post-pandemic normalization rather than a demand break, consistent with a 12-week actual movement of barely 1%.
See the full forecast for the 2020 Chevrolet Silverado 1500 LT→
#3. 2021 Chevrolet Suburban LS
2021 was the launch year of the redesigned 12th-generation Suburban, so early-run buyers are trading in during the first three-to-four-year window when unresolved first-year quirks weigh most on resale value. Large-body SUVs also carry the top-10's highest fuel-cost sensitivity, and the projected P50 sits meaningfully below where the 12-week actual has held.
#4. 2024 Ram Ram 2500 Pickup
2024 was the last model year before Ram's 2025 HD refresh, so off-lease returns and trade-ins for the outgoing look are hitting supply now. Heavy-duty pickups also react to commercial and fleet cycles more sharply than half-tons — the 12-week actual has already dropped -5.65% ahead of the projected P50 pullback.
#5. 2025 Toyota Sequoia TRD Pro
Very early first-year depreciation is normal on the current-generation Sequoia — 2025 is the segment's current model year and buyers at the ~$80K trim level shop with financing rates in mind. Macro sensitivity is heavy in the model's feature weights here; the next MUVVI print is the fastest read on whether the projection widens or holds.
#6. 2022 BMW 5 series
2022 is the final year of the outgoing G30 generation before the 2024 G60 redesign, so off-lease supply of the previous platform is landing in the market right now. European mid-luxury sedans also continue to lose walk-in share to comparably-priced luxury SUVs — a durable body-style rotation, not a one-quarter dip.
#7. 2019 Honda Pilot EX
2019 sits mid-cycle in the 3rd-generation Pilot — well-understood by both buyers and dealers, which typically compresses trade-in premiums. At the ~$21K price band it competes directly with slightly-newer Kia Telluride and Toyota Highlander used inventory, and the model reads the projected P50 as normal supply-side pressure rather than a demand fade.
#8. 2019 Toyota 4Runner SR5
The 5th-generation 4Runner ran through 2024, and the all-new 6th-generation platform launched for the 2025 model year. New-generation availability slowly compresses prior-gen resale even when the outgoing platform retains its enthusiast following — the projected 8-week P50 drop is the first read of that dynamic against a 12-week actual that had held nearly flat.
#9. 2025 Ford Mustang Dark Horse
Dark Horse launched for 2024 with dealer markups stacked on top of $60K+ MSRP, and 2025 is where that early over-sticker premium has fully unwound — the 12-week actual is down 9.01%, and the model projects a further 1.4% pullback. Performance coupes above $70K carry the highest rate-sensitivity in the depreciator list.
#10. 2019 Subaru Crosstrek Base
The 2019 Crosstrek is the previous-generation platform before the 2024 redesign, and sub-$18K used AWD crossovers have been one of the strongest gainer buckets year-to-date. The projected P50 pullback is fall seasonality against a 12-week actual that was still up 3.16%, not a demand break — the tight P10/P90 band around the forecast confirms low uncertainty.
The 5 counter-examples: segments projected to rise
The used market is not one market. Even when broader indices soften, specific segments hold or gain value — often because of constrained supply, model-cycle timing, or category-level demand shifts. These five segments are CarCast's largest projected gainers over the same 60-day window, using the same data-quality bar.
#1. 2019 Mazda CX-5 Sport
2019 is a mid-cycle year for the KF-generation CX-5 with the well-regarded naturally-aspirated 2.5L engine. Six-year-old compact crossovers under $20K sit squarely in the current sweet spot for first-time buyers and third-vehicle household demand — and constrained off-lease returns keep supply thin against that demand.
#2. 2023 Chevrolet Colorado LT
2023 was the launch year of the redesigned 3rd-generation Colorado on the turbo-only powertrain, and early-build units still command a premium while newer off-lease inventory remains sparse. Mid-size truck category demand also continues to pull pricing power from car segments — the 12-week actual was already trending up modestly before this projection.
#3. 2019 Kia Sorento LX
2019 is a mid-cycle 3rd-generation Sorento year with the proven V6 drivetrain. Sub-$15K three-row family SUVs are extremely thin on the used market right now — this trim sits exactly at the ceiling of first-time family-vehicle demand, and the model prices that supply-demand gap forward across the 8-week horizon.
#4. 2020 Nissan Frontier SV
2020 was the last model year of the long-running D40 Frontier — a platform with a durable enthusiast and budget-truck following. The low-$20Ks price band on a mid-size truck matches the sweet spot for tradesman and weekend-utility buyers, and the projected P50 gain reflects the model reading persistent demand against thin supply.
#5. 2019 GMC Canyon SLE
The 2019 Canyon is the previous-generation platform, and the six-year mark hits the sweet spot for dealers stocking trade-in-friendly mid-size trucks that still feel modern. Category demand and price-tier fit stack in this segment's favor — the 12-week actual was already up 5.01% before the projected 3.41% follow-through.
Methodology deep-dive
CarCast produces a fresh 8-week forecast for every tracked vehicle segment on a weekly cadence. A segment is defined by year, make, model, and trim. As of publication we track 808 forecastable segments.
Inputs. Each segment's forecast draws on four categories of data. First, trailing weekly medians from MarketCheck-sourced real dealer listings — this is the level and short-term shape of the price series. Second, the Manheim Used Vehicle Value Index (MUVVI), which tracks wholesale auction values across body types and gives the model a leading indicator for retail movement. Third, US Bureau of Labor Statistics used-vehicle CPI, both the national series and regional splits, which captures broader demand-side pressure. Fourth, per-segment inventory counts, which the model uses to sense supply-side moves before they translate into price.
Output. The model produces a P50 (median) forecast for each week over the next 8 weeks along with a P10 / P90 confidence band. A wide band signals real uncertainty — small sample size, volatile inputs, or a segment in transition. A narrow band signals the model has strong prior evidence for the projection. In this report the P50 is the headline number; the P10 / P90 is the honest range around it.
Retraining. Models retrain weekly on Sunday evening US time. Every published forecast is generated by that week's retrained model — nothing in this report is more than seven days stale relative to the latest fit.
Disclosed accuracy. Across all tracked segments, CarCast's disclosed mean absolute error is 3-5%. Segment-level accuracy varies; the accuracy page publishes the current per-segment audit.
What the model does not do. It does not predict individual VINs, mileage-adjusted transaction prices, or condition-adjusted retail. It predicts the segment-median trajectory. Dealers should read this the way a wholesaler reads MUVVI — as a directional signal, not a per-car quote.
For more on how CarCast's forecasts are built and audited, see /methodology and /accuracy.
Caveats and honest limitations
- This is a segment forecast, not a per-vehicle quote. A specific 2022 BMW 5 Series with 18,000 miles in Chicago is priced differently from the segment median. Use this report to spot direction, not to price a specific unit.
- Confidence bands are real. The P10 / P90 range around each projection reflects the model's actual uncertainty. Where the band is wide, treat the P50 as a rough centerline.
- Macro assumptions carry through. The model bakes in the current MUVVI trajectory and current CPI print. A large shock to either — a sudden tariff change, a fuel-price move, a credit-market break — invalidates the forecast horizon.
- We forecast; we do not sell. CarCast has no inventory position in any segment listed here. Nothing in this report is a recommendation to buy, sell, or hold any specific vehicle.
- Disclosed error. Historical mean absolute error is 3-5%. A forecast within that band on the disclosed accuracy page counts as accurate in our internal audit.
About CarCast
CarCast is a used-vehicle price forecasting platform for independent dealers and vehicle flippers. Where legacy tools like MMR report last week's auction values, CarCast projects the next 60 days of segment-level median prices with disclosed confidence bands. The platform tracks 808 segments, updates weekly, and provides free access to individual buyers.
Independent dealers use CarCast to price trade-ins, spot depreciation before it hits the lot, and defend margin against fast-moving inventory. Vehicle flippers use it to time buy-and-sell windows on specific model-years.
This report is CarCast's first quarterly flagship forecast release. Q4 2026 publishes early October.