Appraisal Institute Controversy Part 2

[glenn walker]

Now now...... the dude announced he was retiring.... cut him a little slack.

At least he's consistent he's called almost everyone on the AF over the year's either Idiot's, morons etc. Honestly I don't know if he's ever given Kudos to anyone.

 

[Surf Cat]

At least he's consistent he's called almost everyone on the AF over the year's either Idiot's, morons etc. Honestly I don't know if he's ever given Kudos to anyone.

I wrote a post not too long ago where I gave him kudos. He contributed to an appraisal problem....no MARS, super computers, or derogatory statements. It was pretty cool. But he admitted he fell into a moment of weakness and went right back to everyone is a moron mode. LOL

I don't know, it kind of makes the forum fun to see all the different personalities...

 

[Non Sequitur]

Back to the OP. The "insta-toot" is still in lockdown holding secret meetings. Are we closer to appointing a new CEO? Members will be the last to know. Nothing has changed, nothing will change.

 

[glenn walker]

I wrote a post not too long ago where I gave him kudos. He contributed to an appraisal problem....no MARS, super computers, or derogatory statements. It was pretty cool. But he admitted he fell into a moment of weakness and went right back to everyone is a moron mode. LOL

I don't know, it kind of makes the forum fun to see all the different personalities...

Back to the OP. The "insta-toot" is still in lockdown holding secret meetings. Are we closer to appointing a new CEO? Members will be the last to know. Nothing has changed, nothing will change.

None of these organizations are democracy's so members are represented by their board members.

Member's are always last to know. That's not changing and if one doesn't like it they should quit supporting it.

 

[RCA]

I've read many of your posts carefully. You seem to believe you have found the best way of extracting adjustments using MARS statistical regression methodologies. That's fantastic. I commend and applaud the novel use of any methodology which assists the appraiser in developing a credible opinion of value. But I'm confused. Are you also saying that the use of other methods that are out side of statistically significant analysis are nothing more than unsupported nonsense?

*** Well, this is one of those rare and tantalizing posts. ***

Valuation is a larger subject than you might expect. There are many valuation tasks outside traditional appraisal and even real estate. But they all share some important features. --- Anyway:

An example of another methodology is using Prolog Constraint Logic Programming (or "CLP").

Think about it. RCA also uses constraints: RCA uses MARS to develop the best possible estimate of property value based on measurable attributes. Then we subtract that estimate from the actual net sale price of comparables to get a fairly accurate estimate (aka the "residual") of the sum of all the value contributions of their unmeasured attributes. That is a constraint. We magically find and prove a theorem, that states that you can explain that the total residual by any linear combination of residual unmeasured or residual or, if you prefer, "subjective" features, without altering the final value conclusion. That is pure magic, - well relatively speaking. With this, the appraiser is given wide latitude in explaining his subjective valuation of condition, quality, functional utility, view (insofar as view lacks objective measures), aesthetics including architectural design, - and so on, AS LONG AS THE TOTAL OF HIS ESTIMATES OF VALUE CONTRIBUTIONS IS EXACTLY EQUAL TO THE RESIDUAL FOR THE GIVEN PROPERTY. Of course, the subject has to have its residual estimated by comparing it to the sorted list of comparable residuals, a list that is objectively created by the R programming that does all the calculation and associated reporting work. The one single spot in the RCA process subject any "subjective" judgment is comparing the subject's subjective properties with the set of photos for the comparables that have been sorted from least to most appeal by the R program.

Getting back to Prolog CLP. Consider that most commercial appraisals, use all three approaches to value: MV, Income, Cost (and there are some others of course). Appraisers often use each of these methods to derive an estimate of value for the subject property. These values are invariably different - unless they have been tweaked otherwise. Question: How are these three values resolved into a final value conclusion? Probably they are weighted at the discretion of the appraiser. But, really these are three completely different values (possibly with some overlapping dependencies), and these three values are invariably constraints on the final value conclusion - that is, we expect a final conclusion somewhere between them, - unless one or more are considered based on inadequate or weak supporting evidence.

In real time valuation, the buyers and others often consider different approaches and create a lengthy and complex set of constraints like:


- From Zoning Regulation Z(1), this or that is not allowed.
- From Zoning Regulation Z(2), this or that is not allowed.
...
- From point of view A, the value cannot be less than or more than ( for example, from the point of view of a certain segment of potential buyers, for example the 15-35% of residential buyers who would, according to conducted surveys, pay extra for a stained glass window in a master bathroom)
- From point of view B, the value cannot be less than or more than.

We can even work with min-max-most probable and even probability curves, whereat we might venture into calculus for our valuation.

All of these constraints can form a hierarchy, especially when working through the HBU.

YES, you can easily question at what point we are overreaching in the amount of work we are setting ourselves up for. But, it can be like manufacturing- you can spend a lot of effort setting up a production line to replace individual workers. Although, quite frankly, to create much of a production line for high quality appraisals, is probably a contradiction in terms. The only thing that will save the individual appraiser who wants to produce appraisals with 1-2% or less error, is some pretty powerful tools. Well, now, with AI and really fast computers and network systems, - and better data services, this is becoming possible.

[ Many will quip: Who in the hell wants a minimum 1-2% error in appraisal estimates? Errors add up. Certainly they are perhaps really only needed at the outer fringes of value: At the lower end which calls for possible destruction and reconstruction. At the higher end, where valuation often becomes a nightmare. The overall logic being, perhaps, if you can't get within 1-2% of your of the ideal value per your definition; then whatever rough value your conclusion is, it can be called into question in.a court of law. In a court of law, where juries and even judges can be swayed based on their lack of knowledge. Oh yes, then we have politics. I could go on.]

If so, what is your proposed methodology for developing opinions of market value where the subject property is of a type where the universe of comparable sales is either not sufficient in number, or does not exhibit characteristics that derive statistically significant confident predictions?

Correct me if I am wrong: The appraiser states the above, ie. he CANNOT give a value that has any degree of predictable accuracy, because the data provided is not statistically sufficient to do so. He can give the best possible value from the data provided, because that is all that can be done. And that value can not be asserted to hold any probability of being the true value of the subject property, whose market value, at least hypothetically, is unknown. It can only be argued, that no other value could be any better - and that THAT market value is all that the client has to work with in negotiation. Of course, in reality, the client will likely give far more weight to the Cost or Income approaches.

(continued)

 

[RCA]

(continuted)

It's perfectly reasonable to utilize regression or MARS modeling to derive adjustments when there is sufficient relevant data in the market to generate results at a significant confidence level. Since you have dismissed other methods as inaccurate, and used only by over the hill, pompous appraisers who believe they are god, like paired sales for example, what is your solution if data collection presents observations with high degrees of variability?

I try to always preface my opinions as being for the San Francisco Bay Area - or at least Northern California. In the San Francisco Bay Area, we have a good MLS systems, as far as MLS systems go. It is fairly accurate. And understand, by using MARS you can account for many factors that most appraisers don't account for. I often go back 5 years for data and my have 300–600 comps. By going back in time so far, even if I don't have a retrospective appraisal to do, I can pick up many more different types of house with different features and pull in more "intelligence" in deciding the value of a current subject property that has some unusual combination of factors ===>>. and that happens a lot in the SF Bay Area, because we have:

1. Homes over a hundred years old.
2. High incomes and high wealth
3. Rich and poor
4. Old wealth and new wealth.
5. Many different styles
6. Many different types of geology and terrain.
7. Many different kinds of updating and remodeling,
8. And soon.

How does MARS work with conservation easement donation value opinions for IRS documentation?

You need to study MARS and come to your own conclusions, because how well it works for any application is largely in the hands of the person using it. MARS juggles around thousands of possible values to find a model that does the best possible job of predicting a target value such as "net sale price," give a set of inputs such as actual sales. This is essentially very similar to how LLMs are trained. It has ways to partition the data into training and. testing data sets. where it might for example take 4/5'ths of the data through random selection to create the model, and use the other 1/5 of the data for testing - and do this over and over to find a model that is good not at just predicting value for the training data - but also for the "unknown" or "test" data. The MARS user is also, please understand, the one who select the multitude of parameters for setting up a MARS run - as well as the data that goes into it. The MARS user is responsible for removing systemic errors in the data which are specific to particular neighborhoods, cities and counties. He is also responsible for making corrections to the data as necessary. The quality of the final pricing model, is usually quite dependent on having a competent "modeler," - somebody who knows what they are doing. The last thing in the world we want is to dump MARS onto a world of incompetent practitioners. The result would most likely be a list of horror stories a mile long. You want to see something like a math degree and at least several years of experience using MARS and doing appraisal.

How about market values per ton for in ground green slate versus in ground red slate?

It really doesn't matter. The general principles of RCA apply to nearly all kinds of valuation. But, for it to make sense, you should be able to predict at least 70-80% of the value from measurable attributes.

I'm curious as to what variables would be in your model in the estimation of contributory values per acre of hydroelectric project associated flowage easements over tillable acreages versus pasturage.

This sounds like more of a DCF type of problem.

Though maybe something simpler, say the size adjustments for various categories of finished areas found within in-patient juvenile psychiatric hospitals?

DCF+.

Would MARS help with impacts on value for mitigated landslide stigma, fuel oil contamination, post clean up, from the basement of a $2.5 mil. ski slope side vacation home?

This would involve market value. Insofar, yes.

Your own hubris reveals your inabilities. Sure mathematical consistency and solid statistical analytics is preferable. And the data is available with straightforward, commonly transferred property types with few and relatively invariant characteristics in markets with similar general price levels. But those are the assignments for beginners.

(continued)

 

[RCA]

(continued)

From this point on, I would have to say you follow the common pattern of attacking someone's views by starting off trying to say whatever good things you can conjure up for the beginning to argue that you are objective and unbiased, before eventually throwing your statements to the wind and diving in to your attack. Such an approach is intended to play up to someone to get their sympathy, only to attack them and supposedly create emotional harm as a kind of revenge for some reason or another, such as perhaps you are just simply a psychopath.

You, I would say, are most likely in that group of sheep appraisers who follow the herd, are protected by the heard, and yet feel encroached up and threatened by some other point of view.

Well, I have got very good news for you - your ARE protected by the heard of idiots you belong with, including the Appraisal Institute.

But, lets be clear, there is a reason that appraiser like you are deadly afraid of the word "accuracy!" "Accuracy" IS indeed a direct attack on you and the philosophy of the Appraisal Institute.

You've made the mistake of thinking that appraisers avoid using the tools you're an advocate of because they don't know their own tools are inaccurate and don't have proper education or experience with modern statistics.

Actually, I don't know any appraisers capable of using MARS. I don't know of any who could make themselves competent, but I would assume they do exist. I have used MARS since 2004, without RCA and without many later refinements with R. That's what beginners should do. - Just using MARS to substitute for linear regression is the first but probably the hardest step. - But to so so without screwing up things with overfitting takes knowledge, and several years of experience and the ability to do critical thinking.

My thinking with respect to RCA is mostly related to what I see in SF Bay Area, with respect to residential properties. Of course, I have appraised office, multi-residential, industrial, land and environmentally impacted sites in this area. They all have their own problems. Office - OMG - Smash and Crash.
Think of San Francisco. Those poor guys over at (well I won't name it).

But that's not it at all. You just haven't received a tough assignment which is devoid of any applicability as to statistically significant analytics.

Of course you don't know what the **** you are talking about. Vermont is nothing like California, especially the SF Bay Area. Who do you think you are kidding?

You've made remarks which indicate a questionable grasp of the fundamental economic concepts that underlie the function of the appraiser.

You are just just showing your bias here. You are a fake, a pretender. Vermont?

You talk about attaining an accurate value.

Yes. Accurate is relative to significance.

That's not even our job.

You do not consider accuracy part of your job? Interesting. Actually, you have the relationship of concept screwed up. You need a course in philosophy and logic.

Our assignments is provide a supported opinion of value.

The support, in case you don't know, IS SUPPORT FOR THE ACCURACY OF YOUR OPINION.

Again, you don't understand the fundamental concepts you are using here.

Value is not a fact to be found in the real estate market.

That depends on your exact definition of value - and there are all kinds of definitions - fact which every appraiser should understand. So, let's be clear stupid. "MARKET VALUE" does indeed exist in the real estate market.

Why?

Many people are just born that way.

Because there is no way to measure it precisely; it is an imperfect market.

It's apparent you don't know what precision is, really. I'll let you figure it out.

Buyers and sellers have imperfect information about the property they are buying and selling; they have imperfect knowledge of the competition in regard to all of the potential alternative properties, the market inventory fluctuates, people have variable motivations and time frames, and on and on. The point is, the market participants themselves don't create an accurate value point.

Again, you are in "don't know what you are talking about" land.

Let me give you a small hint: No body knows the exact value of irrational numbers - because the digits are infinitely long. So, does it make sense to ask someone what the exact value of the mathematical "pi" or "e" is? Of course - yet they are going to counter, "to how many digits of precision?"

So, in this light, what you are talking about is utter nonsense, with respect to accuracy and precision. You simply do not understand the basic concepts. You are in respect to the subject matter an idiot, and in my opinion you should NOT be a certified appraiser. This is basics. Just basics. I certainly don't want an appraiser who doesn't understand the basics of precision and accuracy.

That's why we have appraisers develop supportable opinions of value.

Good grief! "Opinions of Value" by themselves are absolutely NOTHING.

Because that's the best you'll ever get.

Of course, we can do better, that's supposedly why we want appraisers - we want something MORE than "opinion of value" - which we can get anywhere.

Use statistics if you can for sure.

Statistics in not some simple single thing. It is a very big subject. You would be far better off to be more specific about which part of statistics you are talking about - as if you could possibly know (you clearly don't have a clue).

But if I'm reviewing an appraisal that simply uses statistic that aren't significant, and it's going to trial, I know country bumpkin attorney's who will make you look as smart as George's friend Lenny.

Well, you're in Vermont (I assume unless you are working out state). And you admit you have "bumpkin attorney's." Good, you admit that much. I don't know who George or Lenny are for sure; although I assume you mean George Dell. I can assure you that they are extremely limited in their knowledge base of statistics. In fact, you could say that of every statistician. George's views are, as I take it and from talking to him, limited to what he considers "kosher" for the students he has to teach. So, to be sure, I have absolutely no idea of how competent he is at using MARS - or whether he has ever used. I asked him a long time ago, and he said he had never used MARS. Now, hard to say. I have talked to some others who attend Appraisal Institute conferences and are MAIs and use R mostly for Graphics. - and they said when I asked them, that they hadn't used MARS. I have no evidence he (George D.) has ever used MARS, and I expect that if I asked him, he would say something to the effect that it is beyond the realm of usability by most of your smartest appraisers.

In fact, I keep hearing comments from instructors and the like as to how difficult it is to get appraisers to do even simple things (such as multi-linear regression) in statistics. And if I had to bet, I bet they would say I am absolutely insane to argue that appraisers should use MARS and use R programming to the extent I do.

So, in this respect, I completely split from "George" and the other instructors - and just go off on my own track at whatever the cost. Yea. That's exactly what I am doing. I am really aiming at the future and could not care to much about, for example, GSE guidelines - except to perhaps analyze now and then how much damage they are REALLY capable of doing to the future economy.

Of course those other tools have severe weaknesses. How effing stupid do you think we are? That's the job for crying out loud. To know how weak the data is, and come to conclusion that nevertheless are rational, reasonable, logical, and support the conclusions you reach. From what I've read in your comments, I don't believe you actually understand the process. I really question your competency frankly. I just hope to god one day you're on the other side of the a case I'm consulting for. Your over estimation of your own abilities talk volumes about your abilities to estimate at all.

That data is not generally as weak as you think it is, and also I should add, weak data can still be used. In fact, you are using weak data by your own admission. You just don't have the clarity of thought to understand you yourself are really doing. You don't really know what you are doing, in my own opinion.

 

[chad hampton]

Damn, I ain’t readin all dat.

 

[Joe Flacco]

It's too much text to read. Did he call him an idiot or not?

 

[glenn walker]

Thank god he's leaving he is just rambling a lot of nonsense.