2025 Product Enhancements
ALG Model 8.0
Beginning on May 15th, ALG residual values will be calculated using our new and improved forecasting model. Learn more about this new model below.
What is Model 8.0
Model 8.0 represents a revolutionary analytical approach to forecasting designed to greatly improve accuracy and deliver the most comprehensive residual values in the market. By uniting proprietary PIN data from J.D. Power reflecting factors such as incentive spend, brand value, and market trends, with our unique Voice-of-the-Customer redesign impact intelligence and more, Model 8.0 delivers an unparalleled level of precision.
By accurately covering all manufacturers and models in the U.S., including EV-only OEMs such as Tesla and Rivian, and by correctly weighting attributes that have the greatest impact on residuals, such as range retention vs. consumer expectation insights derived from Recurrent data and J.D. Power research for electric vehicles, Model 8.0 has been designed with the future in mind to remain accurate even as technology continues to evolve.
The ALG 8.0 model forecasts future used vehicle values (as % of MSRP and dollar values) over key ownership intervals (12, 24, 36, 48, and 60 months). It delivers:
- Monthly, model-level residual value forecasts
- Outputs normalized to 15,000 miles/year
- Trim-level translation for practical use cases
- Bi-monthly updates aligned with industry leasing cycles
Key Enhancements
Version 8.0 introduces significant improvements over previous versions, particularly compared to 7.0:
- Improved Redesign Impact Detection
- Incorporates APEAL survey data to capture early signals of consumer sentiment on new or redesigned models. Allows forecasts to reflect vehicle lifecycle changes more quickly—often within months.
- Advanced Statistical Techniques
- Replaces Stage 2’s WLS with a Bayesian regression model, which better handles collinearity and allows for informative priors based on historical data.
- Adds an Extreme Gradient Boosting (XGBoost) component for ensemble learning, improving accuracy without sacrificing interpretability.
- Minimized Forecasted Input Error
- Shifts to using observed sales-period data (e.g., incentives, lease/fleet penetration) rather than forecasting these features, reducing the risk of compounding forecast errors.
- Shifts to using observed sales-period data (e.g., incentives, lease/fleet penetration) rather than forecasting these features, reducing the risk of compounding forecast errors.
Every Detail Powered by The Voice of The Customer
Conceptual Framework
Version 8.0 uses a two-stage modeling approach – similar to 7.0 with big changes to the Stage 2 process:
Stage 1 – Segment Forecasting (Top-Down):
Weighted Least Squares (WLS) regression forecasts the overall segment trend using macroeconomic indicators (GDP, gas prices, supply chain stress) and market-level automotive data (incentive spend, lease penetration, sales volume).
Stage 2 – Comparative Model Positioning (Bottom-Up):
A Bayesian regression predicts how each vehicle’s retention deviates from its segment average. It incorporates vehicle-level features such as performance specs, incentive history, APEAL scores, redesign cycles, and sales strategy.
These two stages are combined to produce a robust, granular forecast that balances macro trends with vehicle-specific nuances.
Get In Touch Now About Your Account or to Learn More
Contact an ALG representative today to discuss how we can help your business. Get in touch with someone today.
Sign up to get all our automotive insights - Sign Up Now
If you are an existing ALG customer Login To Your Account Now.
