UAD 3.6 discussion

[J Grant]

AI Overview
(wrt trusting a tool )

While not required to replicate Automated Valuation Models (AVM) or software, appraisers must have a solid understanding of how they work to use them compliantly under the Uniform Standards of Professional Appraisal Practice (USPAP)
. This is mandated by the Competency Rule and is essential for ensuring the appraisal's credibility.
Key USPAP requirements when using AVMs or valuation software:

• Basic understanding: An appraiser must possess a basic understanding of how the AVM or software functions. This includes knowing the model's methods, its underlying assumptions, and the level of user intervention it allows. Without this knowledge, an appraiser cannot properly judge the tool's output.
• Appropriateness for the assignment: An appraiser must determine if the AVM is appropriate for the specific assignment, given its intended use. If the output of the AVM would be misleading, the appraiser must use other tools or decline the assignment.
• • Independent, credible conclusion: The appraiser cannot rely solely on the AVM to determine value. The final value conclusion must be independently supportable, with the software serving as a tool to aid the appraiser's analysis, not replace it.
  • Validation of data and results: The appraiser is responsible for confirming that the AVM's input parameters are correct and that the model produces reliable results for the specific market. This demonstrates the appraiser's understanding of the market and verifies the software's reliability.
  • Communication of use: When an AVM is used, the report must satisfy all applicable reporting requirements. This includes citing the name and version of the software, describing its methods, and identifying the data used. The appraiser must also make it clear that the AVM's output alone is not an appraisal.
  • Avoiding bias: Appraisers must not use or rely on information from software that incorporates bias or reflects discriminatory patterns.

The USPAP Competency Rule emphasizes that appraisers must continuously improve their skills to stay proficient with new technologies and changes in the market. This includes understanding how to properly use and vet advanced tools like AVMs to produce credible and unbiased appraisal results.

 

[shrubberyvaluation]

Of course you're using extreme hyperbole, so I shall as well. If the unknowing appraiser makes a $200k time adjustment when the tool is telling them to make a $2000 time adjustment do you just do that? No questions asked?

No, not unless I know the methodology and the data set and I know it is sound. This reliance on technology for the sake of it being newer, isn't any better than rote adjustments IMO.

 

[alebrewer]

No, not unless I know the methodology and the data set and I know it is sound. This reliance on technology for the sake of it being newer, isn't any better than rote adjustments IMO.

This should be the litmus test for using a tool as well, no? I don't think anyone's promoting the reliance on technology simply for the sake of it being newer. I think the argument goes something like this: "Since appraisers either don't understand, or are too lazy, to use recognized methodologies and techniques, the marketplace has developed tools to assist them." At no time should a tool be used within a vacuum of common sense - that's just a silly accusation.

 

[shrubberyvaluation]

This should be the litmus test for using a tool as well, no? I don't think anyone's promoting the reliance on technology simply for the sake of it being newer. I think the argument goes something like this: "Since appraisers either don't understand, or are too lazy, to use recognized methodologies and techniques, the marketplace has developed tools to assist them." At no time should a tool be used within a vacuum of common sense - that's just a silly accusation.

Have you heard of appraisal modernization? If someone is using rote adjustments, giving them technology isn't going to suddenly make them do research.

 

[alebrewer]

Have you heard of appraisal modernization?

Yes.

If someone is using rote adjustments, giving them technology isn't going to suddenly make them do research.

If they don't know how to perform recognized methodologies and techniques, then I fully agree. Giving them a tool isn't going to somehow impart the knowledge to do so. Using that tool, though, might result in more meaningful (and credible) results than not using it. Again: in the absence of understanding, I'll trust a tool more than a human.

 

[J Grant]

Yes.


If they don't know how to perform recognized methodologies and techniques, then I fully agree. Giving them a tool isn't going to somehow impart the knowledge to do so. Using that tool, though, might result in more meaningful (and credible) results than not using it. Again: in the absence of understanding, I'll trust a tool more than a human.

But it is not separable - a human is using the tool - and if the human is incompetent, they will use the tool poorly, and conversely, if the person is competent, the tool can be used well.

The problem remains is that the profiteer element and the and conflict of interest present when the fee is a determining factor for appraiser selection lets the incompetent slide along and they will indeed use a tool poorly.

A software tool is more dangerous than a rote cheat sheet of adjustments because the software tool is trusted by people like yourself, especially as it can produce impressive-looking charts and graphs. And has math to back it up. But the software tool is widely available and can be misused for bad results on a mass scale ( whether in appraisals or alternative valuation products )

 

[DTB]

Since I have NOT completed any UAD related appraisal, it was the straw that broke this camel's back, I only can ask a question. What is the end goal of this search for mountains of data in residential appraisal work? Appraisal accuracy? Were that many appraisals inaccurate enough to cause default related issues?

I can still peg a value quickly and accurately w/o any mumbo jumbo, and back it up by using that as a basis to put MY money on the line. Ain't been too far off, ever.

I just don't understand where this is all leading.

 

[alebrewer]

A software tool is more dangerous than a rote cheat sheet of adjustments because the software tool is trusted by people like yourself, especially as it can produce impressive-looking charts and graphs.

(a) if you honestly believe this, there is no point in debating this topic any further. (b) a tool cannot be dangerous just because it is trusted by someone. It's either dangerous or it's not - doesn't matter who trusts it. Let's take a simple bi-variate regression for example. There are still a LOT of appraisers who don't know how to use this 'tool' (and it's been around for several centuries). AND - it produces impressive looking (and credible) charts and graphs. I trust regression - when used in the correct context - and modeling market behavior over a particular period of time is a correct context for the use of regression (I personally like to restrict it to the subject's CMS). So is that 'dangerous', or are you just intentionally poopooing something you don't understand? 2nd question: is the use of bi-variate regression - even by someone who doesn't completely understand it - less credible than a 'rote cheat sheet'? And if you say yes - just go ahead and admit that you're gaslighting.

 

[DWiley]

A software tool is more dangerous than a rote cheat sheet of adjustments because the software tool is trusted by people like yourself, especially as it can produce impressive-looking charts and graphs. And has math to back it up.

I see lots of fancy charts and graphs that are very well done.

I also see some where it is very evident that the person who put them in the report had no idea what they were doing.

There is no assumption that charts and graphs are inherently better. On the other hand, there is an assumption that for reports with no data and analysis at all (even though it has now been required for more than six months), there probably was no data and no analysis.

 

[Non Sequitur]

This is all anyone needs for fancy charts and graphs:

ChatExcel AI - Chat With Any Spreadsheet Data | AI for Excel

Discover ChatExcel – the revolutionary AI platform that transforms spreadsheet analysis. Simply talk to your data and instantly generate insights and answers from Excel files and CSVs. Perfect for analysts, business professionals, and data enthusiasts seeking effortless data exploration.

www.chatexcel.co