The Intelligence Layer: From Raw Listings to Predictive Insights
How AutoBridge transforms 131,000+ raw listings into machine-generated price forecasts, feature valuation models, and survival curves — plus where the intelligence stack is heading next.
The Raw Material: 142,000+ Listings and Counting
The Georgian auto market is a high-entropy environment. Every hour, listings appear, update, and disappear across the major marketplaces. Prices shift. Cars sell. Others sit and age. Without structure, this torrent of data is worthless.
AutoBridge processes around 200 new listings daily across 357 tracked makes and 1,179 actively-forecasted model segments. Each entry is cross-referenced to eliminate duplicates, normalized for currency and mileage, and enriched with derived metrics: days on market, price deviation from segment median, and a quality score. The result is a clean dataset with over 7 months of historical depth — the foundation for everything that follows.
142k
Active & historical listings
1,179
Model segments forecasted
219
Days of price history
204
New listings per day
Market Temperature
Strong Buyer's Market
30-Day Price Intelligence: Where the Market Is Heading
For 1,179 distinct make/model/year combinations with sufficient price history, AutoBridge computes forward price forecasts using linear regression on a 90-day rolling window. Each forecast comes with an R² confidence metric — the closer to 1.0, the more predictable the underlying trend.
What the data reveals in late April 2026 is a market under compression. Inventory is elevated, inventory ratios are high, and the system registers a strong buyer's market signal. Sellers who expected foreign buyers are now competing with each other, forcing systematic markdowns across multiple segments.
The exceptions are instructive: Toyota Camry and Prius trend slightly upward, while budget models like the VW Jetta show steeper projected declines. SUVs — Highlander, Forester — display higher volatility, reflecting shifting demand in the post-excise period. R² values across the top-15 average around 0.37, indicating moderate trend consistency rather than random noise.
30-Day Price Intelligence: Where the Market Is Heading
For 1,179 distinct make/model/year combinations with sufficient price history, AutoBridge computes forward price forecasts using linear regression on a 90-day rolling window. Each forecast comes with an R² confidence metric — the closer to 1.0, the more predictable the underlying trend.
What the data reveals in late April 2026 is a market under compression. Inventory is elevated, inventory ratios are high, and the system registers a strong buyer's market signal. Sellers who expected foreign buyers are now competing with each other, forcing systematic markdowns across multiple segments.
The exceptions are instructive: Toyota Camry and Prius trend slightly upward, while budget models like the VW Jetta show steeper projected declines. SUVs — Highlander, Forester — display higher volatility, reflecting shifting demand in the post-excise period. R² values across the top-15 average around 0.37, indicating moderate trend consistency rather than random noise.
Feature Price Impact vs. Market Baseline (USD)
The DNA of Value: What Specifications Actually Cost
Not every specification carries equal weight. Right-hand drive, common in US-auction imports purchased for eastern re-export, loses value on Georgian left-hand roads now that the Kazakhstan route has tightened. A plug-in hybrid commands a substantial premium — but from a thin sample. AWD is the most robustly priced premium, supported by 3,200+ comparable pairs.
AutoBridge computes feature price impacts by comparing median prices of listings sharing a given specification against the global market baseline, controlling for listing volume. The clearest premiums: plug-in hybrid (+$14,000), AWD (+$3,500), electric (+$2,800). The sharpest penalties: right-hand drive (−$8,450) and manual transmission (−$6,300).
These figures are not theoretical — they represent what buyers actually pay or discount on today's Georgian market, derived from thousands of listings with complete specification data. They update monthly as new listings accumulate.
The DNA of Value: What Specifications Actually Cost
Not every specification carries equal weight. Right-hand drive, common in US-auction imports purchased for eastern re-export, loses value on Georgian left-hand roads now that the Kazakhstan route has tightened. A plug-in hybrid commands a substantial premium — but from a thin sample. AWD is the most robustly priced premium, supported by 3,200+ comparable pairs.
AutoBridge computes feature price impacts by comparing median prices of listings sharing a given specification against the global market baseline, controlling for listing volume. The clearest premiums: plug-in hybrid (+$14,000), AWD (+$3,500), electric (+$2,800). The sharpest penalties: right-hand drive (−$8,450) and manual transmission (−$6,300).
These figures are not theoretical — they represent what buyers actually pay or discount on today's Georgian market, derived from thousands of listings with complete specification data. They update monthly as new listings accumulate.
Feature Price Impact vs. Market Baseline (USD)
Average Days on Market by Body Type
Time Is the Hidden Variable: How Fast Each Body Type Sells
Price is a flow, not a stock. A listing priced correctly sells in days; one priced wrong becomes inventory overhead — capital tied up, opportunity cost accumulating. AutoBridge tracks every listing from creation to removal, computing the average days on market for each vehicle category.
Hatchbacks and minivans clear fastest at under 19 days on average. Sedans and SUVs — the dominant volume segments — take around 21 days. Wagons are slowest, reflecting narrow demand for that body type in the Georgian market context.
These averages are a pricing discipline tool. A car priced 20% above its segment median does not simply sell 20% slower — it falls out of the competitive clearing window entirely, eventually requiring a markdown that erases any pricing gain. Dealers who price to the median consistently clear inventory faster.
Time Is the Hidden Variable: How Fast Each Body Type Sells
Price is a flow, not a stock. A listing priced correctly sells in days; one priced wrong becomes inventory overhead — capital tied up, opportunity cost accumulating. AutoBridge tracks every listing from creation to removal, computing the average days on market for each vehicle category.
Hatchbacks and minivans clear fastest at under 19 days on average. Sedans and SUVs — the dominant volume segments — take around 21 days. Wagons are slowest, reflecting narrow demand for that body type in the Georgian market context.
These averages are a pricing discipline tool. A car priced 20% above its segment median does not simply sell 20% slower — it falls out of the competitive clearing window entirely, eventually requiring a markdown that erases any pricing gain. Dealers who price to the median consistently clear inventory faster.
Average Days on Market by Body Type
The Intelligence Stack: What's Coming
The current system computes what linear algebra and SQL can derive reliably from 142,000 listings. The roadmap replaces each approximation with a more rigorous method — and adds capabilities the current stack cannot support at all.
AutoARIMA Forecasting
The current OLS model treats price history as a straight line. AutoARIMA automatically selects seasonal patterns and detects structural breaks — like the April 1 excise shock — without manual intervention. Expected accuracy improvement: 15–25% on R² for volatile segments.
True Hedonic Regression
Today's feature impact figures compare medians without controlling for confounding variables. A proper OLS model with one-hot encoded covariates will isolate the EV premium net of age and mileage effects. Each coefficient will carry a p-value, enabling significance-based filtering.
Price Elasticity Curves
The schema is in place; the script is not. Per-model elasticity curves will quantify exactly how many additional days on market each 10% price premium costs — turning pricing strategy from intuition into arithmetic.
Computer Vision for Salvage Assessment
Georgia's salvage import pipeline involves visual assessment of damage from US auction photos. An image classification model trained on auction imagery can automate severity scoring and estimate repair costs before a car is purchased — mitigating the primary source of margin compression.
Insights-as-a-Service API
Dealers, banks, and insurance companies operating in Georgia need market intelligence embedded in their own systems — DMS, CRM, underwriting platforms. The planned API layer will expose price forecasts, time-to-sell curves, and feature impacts as structured endpoints, making AutoBridge a data infrastructure layer for the broader automotive financial ecosystem.
What This Means for Market Participants
Price forecasts are directional signals, not certainties. If the model shows a segment falling steadily over 30 days with an R² above 0.60, the rational response is to price at or below today's market — not to hold for a price that will not arrive.
Feature premiums are durable but not permanent. The right-hand drive discount exists because of a closed export route. If new eastern markets open, that discount narrows. The data will reflect this before dealer intuition does.
Average days on market is the most underused metric in the Georgian market. Knowing that hatchbacks clear in 19 days while wagons take 21 changes which cars are worth buying at auction — and at what price discount.
Methodology
AutoBridge proprietary database.
131,420 active and historical listings. Price forecasts computed for 312 make/model/year segments with ≥60 days of price history. Feature impact analysis based on 127,000+ listings with complete specifications.
Historical window: January 2025 — April 2026. Forecast horizon: 30 days from April 24, 2026. Survival analysis based on resolved listings from the past 12 months.