Adjustment Support Request

[Sadie]

Paired sales are very common. In fact, paried sales are the comps right there on the grid, with applied adjustment.

MATCHED paired sales are uncommon. Paired sales simply means that sales have adjustments applied to minimize the differences, resulting in a narrower adjusted value range. It then becomes easy to reconcile the MVO from that range.

Regression and larger datasets are useful for time adjustments or for certain trends. But the actual adjustments make more sense when extracted from the local market, similar to subject comps, and it is right there on the grid for a user, reader, or client to see and understand. The heart of a credible appraisal is choosing the right comps, and that takes time and local market knowledge, both of which the AMC's and GSE's want to discount.

The algorithms and regression models for large datasets face a problem: as more data is introduced with dissimilar properties, the results become distorted. But it looks slick and shiny, with charts and graphs. Thus, I is perfect for churn-and-burn appraising, to data-dump, with "proof" of (misleading) support. A non-appraiser can use the same software tools. And THAT is the ultimate goal - if the profiteers can swap out non-licensed PDC collectors to "inspect", then the rationale will go- appraisers are using data software tools on properties they never saw, so then a trained "data analyst" can use these same software tools.

.

Deleted

 

[Sadie]

That's true however, USPAP is the minimum requirement. If your Client specifed at the time of your engagement, that you would provide that sort of data to them and you accept the assignment then, you do it.

Most engagement letters are boiler plate for all loan types and simply say conform the USPAP, then a subsection for when you have something other than a conventional site built home. And a laundry list of what exhibits they want. If they want access to your workfile contents, that would be up front and unusual.

 

[J Grant]

TerryRoherer is arguing for the sake of arguing. If he, or any appraiser, uses regression or large data set statistics with supportable results and can explain what they did, great. My comments about being used in the wrong hands do not apply to the competent appraisers.

It applies to churn and burn appraisers who do not understand these tools and press click and dump the results into a report. Whether the local market or simialr limited set of comps align with the results or not.

 

[TerryRohrer]

TerryRoherer is arguing for the sake of arguing. If he, or any appraiser, uses regression or large data set statistics with supportable results and can explain what they did, great. My comments about being used in the wrong hands do not apply to the competent appraisers.

It applies to churn and burn appraisers who do not understand these tools and press click and dump the results into a report. Whether the local market or simialr limited set of comps align with the results or not.

In a discussion about the use of relevant tools used in appraisal, it should go without saying that competent and ethical use of every is a given. I am objecting to your continued debasement of perfectly suitable tools that you have never used and have no understanding of. The pivot to talking about what you mean comes after pages of objections to what you said, which was pointedly about the shortcomings of tools you have no competence in using, much less a functional understanding. Clearly, the distinction is beyond you.

 

[J Grant]

In a discussion about the use of relevant tools used in appraisal, it should go without saying that competent and ethical use of every is a given. I am objecting to your continued debasement of perfectly suitable tools that you have never used and have no understanding of. The pivot to talking about what you mean comes after pages of objections to what you said, which was pointedly about the shortcomings of tools you have no competence in using, much less a functional understanding. Clearly, the distinction is beyond you.

If you carefully read my posts, I am not debasing perfectly suitable tools.

I am arguing against the bad use of those tools. Clearly, the distinction is beyond you.

 

[TerryRohrer]

The algorithms and regression models for large datasets face a problem: as more data is introduced with dissimilar properties, the results become distorted. But it looks slick and shiny, with charts and graphs. Thus, I is perfect for churn-and-burn appraising, to data-dump, with "proof" of (misleading) support. A non-appraiser can use the same software tools. And THAT is the ultimate goal - if the profiteers can swap out non-licensed PDC collectors to "inspect", then the rationale will go- appraisers are using data software tools on properties they never saw, so then a trained "data analyst" can use these same software tools.

Here is what I fear in res mortgage loan appraising. USPAP has a standard of peer practice. As more and more appraisals are done by AMC's, who are the ones who need churn and burn, and AMC;s use hybrids and PDC collections - thus they want big data/slick "support" from computer programs - charts, graphs, and statistics that "support" the adjustments will become the expectation, even if the results are misleading or awful. Even if the appraiser simply dumped the data and clicked a few prompts to get it.

When appraisers give over their agency to that, the argument becomes, who needs the appraiser to click on software to do the work when anybody with a short training course can use the software?

The problem with using large data sets from computer programs that do not vet the sales is this: in my market area, the subject and best similar sales are in a non-gated, no amenity subdivision. But within a half mile are a number of similar-sized pool or lake view home sales in a gated subdivision with a golf course and mandatory buy-in country club fee of 300k and yearly dues of 30k. and HOA fees of 16k annually. Property sales from these superior amenity communities with high fees are NOT comps for the subject, yet a non-vetted software search will bring up these sales as comps to get adjustments from.

80,000+ posts and you don't have any idea what you were saying, much less an ability to comprehend the plain English of others.

 

[ZZGAMAZZ]

What “sensitivity analysis”​

In appraisal practice, sensitivity analysis is not a standalone valuation method. It’s a diagnostic or reasonableness-check tool that answers questions like:

• If this adjustment were a little bigger or smaller, does my value conclusion materially change?
• Which adjustments matter most to the final indication?
• Is my reconciliation fragile or stable?

Used properly, it tests conclusions that were first derived from:

• paired sales analysis,
• regression or other statistical modeling,
• market extraction,
• cost/income support (when applicable),
• or qualitative market behavior evidence.

It does not replace those things.

---

The core problem with using only the sales grid
If the appraiser’s “sensitivity analysis” uses only the adjusted sales in the grid, several fatal issues arise:
1. Circular reasoning
The adjustments are derived from the same sales being tested.
That means:

• The model is validating itself
• There is no independent evidence
• No external market signal is introduced

This is equivalent to saying:


That’s not analysis—it’s recursion.

2. The grid is not a dataset; it’s a

A URAR grid typically contains:

• 3–6 sales (occasionally more)
• pre-selected and already-filtered comparables
• often intentionally constrained to “most similar”

This is not a representative market sample.
It’s a curated subset designed for illustration and reconciliation.

Sensitivity analysis assumes:

• a sufficiently large and variable dataset
• observable response of price to changing inputs

A 4–6 sale grid simply does not meet that condition.

---

3. Adjustments require

USPAP and long-standing appraisal theory are clear (even if not always followed in practice):



Sensitivity analysis can tell you:

• which adjustment matters more than another
  It cannot tell you:
• what the adjustment should be
• whether it reflects buyer behavior

If no paired sales, regression, or other extraction exists outside the grid, the adjustments are unsupported—no matter how elegant the sensitivity narrative sounds.

---

Why this fails the URAR definition of market value

The URAR definition centers on “the most probable price” under specific conditions.

Key word: probable, not “mathematically convenient” or “internally consistent.”

To estimate probability, you need:

• exposure to the broader competitive market
• evidence of how buyers actually pay for differences
• confirmation that results are not driven by one or two influential sales

A grid-only sensitivity analysis:

• cannot demonstrate probability
• cannot show robustness
• cannot show market breadth
• cannot distinguish signal from noise

At best, it shows internal consistency.
Market value requires external market validation.

Can the concept of "most probable" properties within the subject's sub-market--if defined appropropiately--be reflected by properties that an appraiser selects to report in the SCA grid?

 

[Elliott]

This from a report that cost $870. Not my report, not saying the explanations are sufficient. One day loan approval. State of the Art?

"-View adjustments are made based on market data and reflect sites which offer a view / location considered superior than typical neighborhood view. A $175,000 city / strip view adjustment was applied based on market reaction and using paired sales analysis with comparables 1 and 2."

"The adjustments utilized are reasonable and within an acceptable market range based on market analysis. Based on MLS data there is an acceptable range variance for properties differing in a variety of attributes. Adjustments are extracted from and are supported by the actions of the market. Positive and negative adjustments were applied to comparables in areas of dissimilarity to produce an indicated value of the subject property based on paired sales analysis. Not all adjustments in the Sales Comparison Approach can be directly extracted or supported by the available market data with a high degree of accuracy. Some adjustments have an element of subjectivity and professional judgment which the appraiser has applied based on prior observations of the reactions of the typical / knowledgeable buyers' and sellers' in the marketplace. These adjustments are then refined using sensitivity analysis within the grid and tested for reasonableness with the selected comparables. This method is standard and well accepted practice within the appraisal industry. All adjustments are rounded to the nearest $100."

Other adjustments are 'supported' with one or two lines of narrative. 30% LTV and the appraiser hit value. 42-page report.

 

[J Grant]

When the defense of large data sets or regression rests on insulting people, what does that say about the methodology? (or the person making the insults ) Describe what works about the methodology, and how the sales or listings in it can be vetted, for example.

The best support for adjustments is multiple methods, which I do, even though I post here about how relevant the comps on the grid are and how the adjustments made to them are clear to the reader and users. I spend hours looking at additional sales and listings, not just the ones on the grid. It takes hours to eliminate the less relevant or similar sales, but they are looked at and considered. This how buyers purchase - they spend days or weeks or months eliminating properties until they find the "one", from a smaller set of contenders.

If an appraiser vets the sales and listings of a larger data set and compares the results to the most relevant smaller local market data, the combination provides the best support. I run the MC analysis of a larger data set as well as the MLS stats, since I can control the input and exclude outliers. Using a cost approach for depreciated costs is another method that can support making an adjustment. Not every minor difference needs an adjustment. No matter what method we use ,it is a model of MV, not an actual sale price prediction.

Buyers adjust with their wallets, paying more for positive features and paying less for adverse features. Buyer identification is crucial, as buyer profiles expect different things. A luxury property buyer looking for a move-in ultra-renovated or new is not the same buyer as a middle-class person on a budget, and they are not the same buyer as an investor looking for a bargain and willing to make repairs. Yet I see appraisers using the different properties and making huge adjustments, instead of narrowing down the search to what the indeifitied typcial buyer for a property is looking for.

 

[Tim Hicks (Texas)]

Just have TC email you the list of adjustments.