Secondary Market Valuation Model

The model continuously consumes live market data to anchor valuations in current reality, not historical averages. This prevents underpricing high-demand items or overpricing stale inventory.

• Data Sources: B2B auction platforms, B2C recommerce marketplaces (eBay, Poshmark), search trend APIs, and competitor pricing scrapers.
• Velocity Tracking: Measures the rate of sale for specific SKUs in secondary channels to calculate a liquidity premium.
• Seasonality Adjustment: Automatically weights demand signals based on time-series patterns, boosting prices for winter coats in autumn regardless of current slow sales.

The valuation engine ingests a structured defect ontology from the computer vision grading system to quantify depreciation. It does not treat all 'used' items equally, but prices against a granular condition vector.

• Cosmetic Grade: Maps scratches, dents, and discoloration to a statistical discount factor derived from historical sales of similarly graded units.
• Functional Status: Applies a binary multiplier—items with verified functional defects are routed to a 'parts/repair' pricing curve, not a 'used' curve.
• Packaging Integrity: Integrates the Packaging Integrity Score to determine if an item qualifies for 'Like New' pricing tiers, which command a significant premium in recommerce.

The model calculates distinct optimal prices for each recovery channel, recognizing that a pallet of mixed-grade electronics sold B2B has a different value curve than individually listed B2C items.

• B2B Liquidation Logic: Prices are optimized for lot velocity, factoring in buyer acquisition costs and expected holding time. The goal is to maximize Net Recovery Rate per pallet.
• B2C Recommerce Logic: Prices are set dynamically against individual listing competition, incorporating shipping costs and platform fees to calculate a net margin.
• Channel Arbitration: The core decision engine compares the predicted net recovery across all channels to inform the Automated Disposition Engine, routing goods to the highest-yield path.

The algorithm builds and continuously refines product-specific depreciation curves that predict value decay over time, distinct from standard linear accounting depreciation.

• Time-Since-Launch Factor: Applies a non-linear decay function that recognizes steep initial depreciation immediately after a new model release, which then plateaus.
• Substitution Effect: Detects when a new product generation cannibalizes the secondary market value of the previous generation and triggers an immediate markdown.
• Repair Cost Deduction: For items routed to refurbishment, the model subtracts a predicted parts and labor cost, sourced from a Re-kitting Workflow analysis, from the final resale value to calculate a true net recovery.

The valuation model is not static. It ingests actual sale prices from completed transactions to measure prediction error and trigger automated retraining cycles, preventing model drift.

• Grade-to-Net Recovery Analysis: Continuously correlates the assigned grade and initial valuation with the final sale price to refine discount factors for specific defect types.
• Anomaly Detection: An unsupervised monitor flags SKUs where the predicted price consistently deviates from the clearing price, triggering a model review.
• A/B Pricing Experiments: The system can programmatically test pricing strategies on a small percentage of inventory in recommerce channels, using the results to update the global model.