System, method and computer program for improved forecasting residual values of a durable good over time
A residual value forecasting system may utilize heterogeneous data, such as used market data, industry-specific data, and non-industry-specific data, from disparate data sources to produce residual value forecasts of an item based on a sophisticated residual value forecasting model particularly configured for agility. The system can dynamically and quickly adapt to change in data inputs and produce custom outputs. The system may determine a baseline value for an item using the used market data, a microeconomic factor using the industry-specific data, and a macroeconomic factor using the non-industry-specific data, as well as adjustments such as locality adjustments and modifications. Given the macroeconomic factor and the microeconomic factor relative to the locality-adjusted value of the item and in view of the competitive sets of similar and/or substitute items in the same industry, the system can generate an accurate forecast residual value of the item at a future time point.
1. A method for generating forecasts of residual values of an item of interest in an industry, the method comprising:
programmatically receiving or obtaining, by a system, used market data, non-industry-specific data, and industry-specific data from multiple data sources, the system having a processor and a non-transitory computer-readable medium;
transforming, by the system, the used market data, the non-industry-specific data, and the industry-specific data into data representations internal to the system;
applying, by the system, the data representations of the used market data, the non-industry-specific data, and the industry-specific data as input to a residual value forecasting model, the residual value forecasting model having a first computational component driven by a baseline value for the item of interest, a second computational component driven by macroeconomic factors not specific to the industry, and a third computational component driven by microeconomic factors specific to the industry, the first computational component having a baseline value variable for representing the baseline value for the item of interest with a base configuration at an initial time point, the second computational component having a macroeconomic factor represented by a linear combination of macroeconomic variables that represent macroeconomic features under consideration for the item of interest, the third computational component having a microeconomic factor represented by a linear combination of microeconomic variables that represent microeconomic features specific to the industry, the applying producing a forecasted residual value for the item of interest at a future time point; and
providing the forecasted residual value for the item of interest for presentation on a client device.
2. The method according to claim 1 , further comprising:
determining the baseline value for the item of interest, the determining comprising deriving, from auction data, an observed current market value of the item of interest with the base configuration at the initial time point;
determining, at the initial time point, a reference period at which adjustments are to be made to the item of interest for value alignment with items that compete with the item of interest in the industry;
determining, at the initial time point based at least on the initial time point and the reference period, a constant width of time intervals at which forecasts are to be generated for the item of interest;
determining a locality adjustment to the item of interest, the locality adjustment determined at the initial time point and at a time a value modification being made to account for a modification to the item of interest;
collecting or determining incremental values of modifications to the base configuration of the item of interest;
determining, based at least on the locality adjustment and the incremental values of modifications to the base configuration of the item of interest, a locality-adjusted value of the item of interest having the modifications to the base configuration;
constructing competitive sets of similar and substitute items in the industry, the constructing comprising partitioning items that compete with the item of interest in the industry based on a measure of similarity between pairs of the items;
determining the macroeconomic factor by computing the linear combination of macroeconomic variables that represent the macroeconomic features under consideration for the item of interest,
determining the microeconomic factor by computing a linear combination of observed or forecasted values of the microeconomic variables that represent the microeconomic features specific to the industry; and
computing the first computational component, the second computational component, and the third computational component of the residual value forecasting model utilizing the baseline value for the item of interest, the locality-adjusted value of the item of interest having the modifications to the base configuration, the macroeconomic factor, and the microeconomic factor thus determined.
3. The method according to claim 2 , further comprising:
computing, utilizing a baseline value of a closest competing item in the competitive sets, an adjustment value; and
generating a final forecasted residual value for the item of interest by adjusting the forecasted residual value for the item of interest with the adjustment value.
4. The method according to claim 2 , wherein the modifications to the base configuration of the item of interest comprise value-affecting changes to the item of interest at any time point.
5. The method according to claim 2 , further comprising:
combining values received or obtained from the multiple data sources into a single value which gives more weight to a data source or data type in which a higher confidence is held, the combining including computing a combining equation at a time point in the reference period for a similar or substitute item that competes with the item of interest in the industry.
6. The method according to claim 2 , wherein the reference period is 36 months and wherein the constant width of time intervals is two months.
7. The method according to claim 1 , wherein the second computational component further includes a coefficient for the macroeconomic factor and wherein the third computational component further includes a coefficient for the microeconomic factor.
8. The method according to claim 1 , wherein the used market data comprises open auction data, closed auction data, and certified pre-owned data.
9. The method according to claim 1 , wherein the non-industry-specific data comprises at least one of inflation, unemployment rate, gas prices, an economic index, interest rates, or industry-wide used market supply.
10. The method according to claim 1 , wherein the industry-specific data comprises vehicle-specific data and wherein the vehicle-specific data comprises modifications to the base configuration of the item of interest.
11. A system for generating forecasts of residual values of an item of interest in an industry, the system comprising:
a processor;
a non-transitory computer-readable medium; and
stored instructions translatable by the processor to perform:
programmatically receiving or obtaining used market data, non-industry-specific data, and industry-specific data from multiple data sources;
transforming the used market data, the non-industry-specific data, and the industry-specific data into data representations internal to the system;
applying the data representations of the used market data, the non-industry-specific data, and the industry-specific data as input to a residual value forecasting model, the residual value forecasting model having a first computational component driven by a baseline value for the item of interest, a second computational component driven by macroeconomic factors not specific to the industry, and a third computational component driven by microeconomic factors specific to the industry, the first computational component having a baseline value variable for representing the baseline value for the item of interest with a base configuration at an initial time point, the second computational component having a macroeconomic factor represented by a linear combination of macroeconomic variables that represent macroeconomic features under consideration for the item of interest, the third computational component having a microeconomic factor represented by a linear combination of microeconomic variables that represent microeconomic features specific to the industry, the applying producing a forecasted residual value for the item of interest at a future time point; and
providing the forecasted residual value for the item of interest for presentation on a client device.
12. The system of claim 11 , wherein the stored instructions are further translatable by the processor to perform:
determining the baseline value for the item of interest, the determining comprising deriving, from auction data, an observed current market value of the item of interest with the base configuration at the initial time point;
determining, at the initial time point, a reference period at which adjustments are to be made to the item of interest for value alignment with items that compete with the item of interest in the industry;
determining, at the initial time point based at least on the initial time point and the reference period, a constant width of time intervals at which forecasts are to be generated for the item of interest;
determining a locality adjustment to the item of interest, the locality adjustment determined at the initial time point and at a time a value modification being made to account for a modification to the item of interest;
collecting or determining incremental values of modifications to the base configuration of the item of interest;
determining, based at least on the locality adjustment and the incremental values of modifications to the base configuration of the item of interest, a locality-adjusted value of the item of interest having the modifications to the base configuration;
constructing competitive sets of similar and substitute items in the industry, the constructing comprising partitioning items that compete with the item of interest in the industry based on a measure of similarity between pairs of the items;
determining the macroeconomic factor by computing the linear combination of macroeconomic variables that represent the macroeconomic features under consideration for the item of interest,
determining the microeconomic factor by computing a linear combination of observed or forecasted values of the microeconomic variables that represent the microeconomic features specific to the industry; and
computing the first computational component, the second computational component, and the third computational component of the residual value forecasting model utilizing the baseline value for the item of interest, the locality-adjusted value of the item of interest having the modifications to the base configuration, the macroeconomic factor, and the microeconomic factor thus determined.
13. The system of claim 12 , wherein the stored instructions are further translatable by the processor to perform:
computing, utilizing a baseline value of a closest competing item in the competitive sets, an adjustment value; and
generating a final forecasted residual value for the item of interest by adjusting the forecasted residual value for the item of interest with the adjustment value.
14. The system of claim 12 , wherein the modifications to the base configuration of the item of interest comprise value-affecting changes to the item of interest at any time point.
15. The system of claim 12 , wherein the stored instructions are further translatable by the processor to perform:
combining values received or obtained from the multiple data sources into a single value which gives more weight to a data source or data type in which a higher confidence is held, the combining including computing a combining equation at a time point in the reference period for a similar or substitute item that competes with the item of interest in the industry.
16. The system of claim 12 , wherein the reference period is 36 months and wherein the constant width of time intervals is two months.
17. The system of claim 11 , wherein the second computational component further includes a coefficient for the macroeconomic factor and wherein the third computational component further includes a coefficient for the microeconomic factor.
18. The system of claim 11 , wherein the used market data comprises open auction data, closed auction data, and certified pre-owned data.
19. The system of claim 11 , wherein the non-industry-specific data comprises at least one of inflation, unemployment rate, gas prices, an economic index, interest rates, or industry-wide used market supply.
20. The system of claim 11 , wherein the industry-specific data comprises vehicle-specific data and wherein the vehicle-specific data comprises modifications to the base configuration of the item of interest.