Home Value Explorer (HVE) FAQ

Home Value Explorer^® (HVE^®) is consistently rated among the top automated valuation models (AVMs) in the industry for accuracy and coverage. Freddie Mac participates in testing with third-party AVM testing companies and with large lenders. HVE’s average national hit rate, makes it among the leading AVMs for overall geographic coverage in the United States.

HVE is tested by third-party companies that perform AVM testing and large national lenders. HVE undergoes regular analysis within Freddie Mac as part of our internal model validation process. We perform extensive in- and out-of-sample testing. Each month, the model is recalculated based on the most recent available data and the performance is evaluated two months after the model build to assess the ongoing accuracy of the estimated values.

HVE is licensed and serviced through a distributor network; each distributor provides unique value-adds with distinct pricing, packaged products, reseller networks, and/or service commitments.

Freddie Mac does not endorse or make any representations or warranties regarding any distributor or reseller. Each end user is solely responsible for the decision of whether to engage a given distributor or reseller, and on what terms, and for ensuring that the distributor or reseller it engages have in place and maintain controls, processes, and procedures that satisfy the end user's business, operational, risk, legal and regulatory requirements.
 

HVE is an ensemble model, with machine learning subcomponents combined with an industry- leading house price index. It blends model estimates returned by sub-models – the repeat sales model, which uses property sales data, and the hedonic model, which uses property characteristics data.

The HVE point value estimates can be returned on single-family, condo, co-op and 2-unit properties.

Yes, HVE covers all 50 states and the District of Columbia, with data from more than 3,100 counties.

When evaluating any property, if data is available, HVE uses both hedonic modeling and repeat sales modeling to generate a point value estimate. In many cases, however, rural areas are harder to evaluate due to the lack of available sales activity. Generally, confidence level tends to be lower in rural areas – those outside metropolitan and micropolitan statistical areas, where homes are commonly separated by a great distance - with insufficient data on which to model.

The HVE model is recalculated each month based on the most recent available data.

Less than 25% of all properties evaluated by HVE receive a low confidence rating. And although the confidence level is low, the HVE valuation can provide valuable information, especially in the case of rural properties. Low confidence is commonly the result of a lack of good data or recent sales data in the area. An appraiser will have the same difficulty obtaining good information in rural areas. An HVE value with a Low Confidence Score can be used for example, as an indicator to review the appraisal more closely. The goal is to enhance the collateral review of as many loans as possible.

HVE is consistently rated as one of the industry’s top AVMs in accuracy and coverage. All AVMs have geographic strengths and weaknesses that are due to the availability of data to the AVM in that locale. HVE has an advantage over other AVMs because it incorporates the property data from Freddie Mac’s portfolio of loans. We have more than 103 million property addresses in our database. In addition, because our large database is based on our portfolio of loans, HVE has the unique advantage of being able to value many properties within the 12 nondisclosure states.

Freddie Mac obtains new data each month and completely rebuilds the model using this data.

The hedonic model is a statistical model that estimates a property value by treating it as a bundle of characteristics (number of bedrooms, number of bathrooms, etc.). There are no advantages to using only a hedonic model. Best estimates are provided when both hedonic information and repeat sales information are used and combined statistically.

Each HVE point value estimate is accompanied by a Forecast Standard Deviation (FSD). The FSD represents the probability that the AVM value falls within a statistical range of the actual market value, measured against a sales price. The lower the FSD, the smaller the error in predicting actual market value and the higher the level of confidence HVE has in the value (the closer the AVM value will be to the actual sales price).

HVE derives the confidence levels from the FSD and summarizes within high, medium, and low value ranges. The confidence levels can be defined in the following manner:

• High – at least as precise (typically more precise) than an average appraisal
• Medium – comparable to or somewhat less precise than an average appraisal
• Low – less precise than an average appraisal (more common for rural areas)
   

Confidence Levels Related to Standard Deviation

• High: <.130 FSD | ~50% of HVE estimates
• Medium: > .130 and <.200 FSD | ~25% of HVE estimates
• Low: > .200 FSD | ~25% of HVE estimates
   

After some natural disasters, HVE may temporarily stop returning point value estimates for certain properties located within zip codes identified as eligible disaster areas. After a determination that sufficient time has passed since the natural disaster, HVE will then begin returning values for those impacted properties.

A newly constructed home is incorporated into the HVE model when either Freddie Mac purchases the loan and obtains the property information, or the deed of the property is recorded in the local tax assessor’s office and that information is captured and sold to Freddie Mac by independent data aggregators.

No, there are no known REO, short sales, or restructured loans included in the model.

No, there are no known manufactured homes included in the model.

In many cases HVE can provide an estimate on properties in non-disclosure states.

HVE contains limited data on cooperative properties (co-ops). Therefore, hit rates may be lower for co-op property types.  Additionally, Freddie Mac frequently receives co-op data that is coded incorrectly as a condominium and could be included in the model as a result.

Some key reasons why a property may not receive an HVE point value estimate are:

• The submitted property address could not be verified by United States Postal Service standards.
• The property is a newly constructed home and information on the property is not yet available to HVE.
• The property may be located in an area where HVE does not have enough raw data to generate a value estimate with a sufficient degree of confidence (e.g., rural properties).
• The property is a manufactured home or a 3- to 4-unit property. HVE does not provide data on these property types.
   

In some areas, HVE does not have assessment records and must complete the valuation using the repeat sales method ONLY. In that instance, HVE must have prior transaction information in order to valuate. If no prior sales data is available in the area, then there will be no valuation.

HVE performs statistical checks on its data. If the variance on that data is too high, HVE will not produce a value estimate.
 

Home Value Explorer Fact Sheet