Request RELAR Demo
Loading...
Request sample report.Request sample report.


Testing and Evaluation of the Product

The primary measure by which the RELAR product is evaluated internally is its ability to accurately predict sales price and market time for an individual property. One major issue is to determine the appropriate standard against which both RELAR and other products should be evaluated. In particular, CDR is reluctant to measure the results of a RELAR analysis by comparing the results to an appraisal or appraised value of a property. Appraisals represent opinions, albeit ones based on standards and experience, about the value of a property. While such a comparison may provide a gross indication of the accuracy and utility of the RELAR analysis, it is insufficient to provide the statistical accuracy evaluation many of our customers require as input to their risk analysis and valuation models.

Instead, our approach has been to evaluate the RELAR analysis against actual sales experience with lists and pools of properties.  We have compiled pools of properties representing over 3500 individual residential properties in the Southern California area. We are in the process of expanding those lists to include 20 major metropolitan areas nationwide.

The testing and evaluation criteria for the RELAR product, in order of priority, are as follows:

  • Average price accuracy for a list of properties – Measure the average price for a group of properties predicted by the RELAR analysis and compare those predictions to actual sales prices. To date, CDR has tested over 700 aggregate properties, and aggregate valuations are within +/-0.4 percent for a list of 100 or more properties.
  • Aggregate sales time accuracy for a list of properties = Measure the average time for a group of properties to sell and compare that to the aggregate RELAR analysis prediction. Using over 700 properties, our current aggregate sales time error is +/-5%, representing a typical error for a list of properties of +/- 6 days on market.
  • Price accuracy for individual properties – Determine the number of individual properties in a list of properties where the analysis accurately predicts the sales price of the property to within +/-5%. To date, CDR has tested over 500 properties and shown that between 85% - 90% of individual valuations meet the +/-5% criterion. As MLS data coverage becomes more extensive, CDR expects to meet its goal of a minimum of 95% of properties for individual valuations.
  • Sales time accuracy for individual properties – Determine the number of individual properties in a list of properties where the analysis accurately predicts the time to sell to within +/- 10%, representing a typical error of +/-11 days. Currently, our results show that this criterion is met by approximately 70% of properties in a list. We expect to meet the goal with 90% of properties as MLS coverage in major metropolitan areas is expanded and rationalized (made more consistent in format from area to area).

Summary Results for Aggregate Price Prediction for 100 Random Properties

San Diego Orange County
Aggregate Sales Price $74,342,415 $72,820,250
Aggregate Predicted Price $73,984,354 $73,074,638
Difference $358,061 $254,388
% Difference 0.482% 0.349%


Expected Confidence Level 50% 90% 95%
Observed Confidence Level - San Diego 53 87 96
Observed Confidence Level - Orange County 37 84 94

San Diego 100 Property Value Differences

San Diego 100 Properties Percent Difference

San Diego Property Details

Orange County 100 Properties Value Difference

Orange County 100 Properties Percent Difference

Orange County Property Details