• Welcome to AppraisersForum.com, the premier online  community for the discussion of real estate appraisal. Register a free account to be able to post and unlock additional forums and features.

UAD 3.6: No need to drive comp photos

yeah, my general advice the the production folks in my realm is that they're not saving anything with a desktop option - as they'll still have to get a PDR. The do save quite a bit of time and money both (at least the borrower saves the money) with the PDR only option.
That's what im afraid of. The hybrid option just bombs and the GSEs goes straight to the waiver with a PCR. Cutting us out completely.

I think it will go that way in 5-10 years, with appraisers getting 20% of the complex or value pie.

But if the hybrid fails, it may happen sooner.
 
That's what im afraid of. The hybrid option just bombs and the GSEs goes straight to the waiver with a PCR. Cutting us out completely.

I think it will go that way in 5-10 years, with appraisers getting 20% of the complex or value pie.

But if the hybrid fails, it may happen sooner.
Certainly, the more advanced they believe CU becomes, the more loan types they'll offer VA+PDR. I'm not sure what the current ratios of VA/VA+PDR/Hybrid/Desktop/Traditional are.
 
That's what im afraid of. The hybrid option just bombs and the GSEs goes straight to the waiver with a PCR. Cutting us out completely.

I think it will go that way in 5-10 years, with appraisers getting 20% of the complex or value pie.

But if the hybrid fails, it may happen sooner.
This has more of chance than the GSEs admitting defeat and returning to, for a lack of better words, traditional reports.
 
Does anyone really understand all of the outputs TrueTracts or similar software produces? I’ll wager there may be a unicorn or two out there, but the vast majority do not understand Bayesian Adjustment Framework, Theil-Sen Regression, and several degrees of Polynomials with widely different trend lines.

Be careful what concepts you banty about. You might be accused of charlatanism:

Me to Grok: "It [i.e. "Bayesian Adjustment Framework"] appears to be non-academic, lacking review."

Grod: "The term "Bayesian Adjustment Framework" does not correspond to any established methodology in the field of appraisal, whether in real estate valuation, performance evaluation, or related disciplines. Comprehensive searches across academic databases, professional resources, and industry discussions reveal no references to this exact phrase as a recognized framework.

Bayesian statistical techniques are indeed applied in various adjustment contexts, such as:

  • Bayesian Adjustment for Confounding (BAC): A model averaging approach used in causal inference and epidemiological studies to account for uncertainty in confounder selection.
  • Bayesian methods in survey weighting, process control, or clinical trial covariate adjustment.
However, these are distinct, peer-reviewed concepts primarily in statistics and causal inference, not appraisal-specific practices.

In real estate appraisal, adjustments (e.g., in the sales comparison approach) are typically derived from market data analysis, paired sales, regression models, or qualitative techniques, as outlined in standards such as the Uniform Standards of Professional Appraisal Practice (USPAP) and The Appraisal of Real Estate. No evidence supports a "Bayesian Adjustment Framework" as a standard or proposed method in this domain.

Your observation that it appears non-academic and lacking peer review aligns with the available evidence: the term yields no credible publications, guidelines, or professional discussions. It may represent a misunderstanding, informal proposal, or conflation of Bayesian principles with traditional adjustment processes. If you can provide the original source or additional context where the term was encountered, further clarification could be pursued."

Note: There are a few charlatans in the field. Oh yeah, "conflation" is that magic word.
 
Be careful what concepts you banty about. Yity and/or Inference doesn't work for housing valuation, it [i.e. "Bayesian Adjustment Framework"] appears to be non-academic, lacking review."

Grod: "The term "Bayesian Adjustment Framework" does not correspond to any established methodology in the field of appraisal, whether in real estate valuation, performance evaluation, or related disciplines. Comprehensive searches across academic databases, professional resources, and industry discussions reveal no references to this exact phrase as a recognized framework.

Bayesian statistical techniques are indeed applied in various adjustment contexts, such as:

  • Bayesian Adjustment for Confounding (BAC): A model averaging approach used in causal inference and epidemiological studies to account for uncertainty in confounder selection.
  • Bayesian methods in survey weighting, process control, or clinical trial covariate adjustment.
However, these are distinct, peer-reviewed concepts primarily in statistics and causal inference, not appraisal-specific practices.

In real estate appraisal, adjustments (e.g., in the sales comparison approach) are typically derived from market data analysis, paired sales, regression models, or qualitative techniques, as outlined in standards such as the Uniform Standards of Professional Appraisal Practice (USPAP) and The Appraisal of Real Estate. No evidence supports a "Bayesian Adjustment Framework" as a standard or proposed method in this domain.

Your observation that it appears non-academic and lacking peer review aligns with the available evidence: the term yields no credible publications, guidelines, or professional discussions. It may represent a misunderstanding, informal proposal, or conflation of Bayesian principles with traditional adjustment processes. If you can provide the original source or additional context where the term was encountered, further clarification could be pursued."

Note: There are a few charlatans in the field. Oh yeah, "conflation" is that magic word.

If you want to know the reason Bayesian Probabilty and/or Inference doesn't work for housing valuation it is that the use of prior probabilities introduces bias. It introduces unreliability:


Grok: "Yes, the arguments presented regarding the limitations of Bayesian probability in house valuation—particularly the subjectivity of prior distributions and computational complexity—apply directly to Bayesian inference more broadly, as Bayesian probability forms the foundational framework for Bayesian inferential methods.

Bayesian inference updates prior beliefs about model parameters (e.g., coefficients representing the impact of property features like size, location, or condition on value) with observed data to produce posterior distributions. In real estate valuation contexts, such as hedonic pricing models or mass appraisal regressions, this process inherits the same challenges:
  • Subjectivity in priors: Specifying informative priors requires expert judgment on parameter values, which can vary among appraisers and introduce bias if priors are misspecified. In property markets with heterogeneous data or limited transactions, weakly informative or non-informative priors may be used, but even these choices influence results, potentially deviating from purely data-driven estimates preferred in regulatory frameworks like USPAP, which emphasize empirical market evidence and reproducibility.
  • Computational demands: Posterior inference typically relies on techniques such as Markov Chain Monte Carlo (MCMC) or variational inference, especially in hierarchical models accounting for spatial dependencies or submarket variations common in real estate. These methods are resource-intensive, requiring specialized software and expertise, which contrasts with the efficiency of frequentist approaches (e.g., ordinary least squares regression) that provide point estimates quickly and are more prevalent in practical appraisal tools.
While Bayesian inference offers benefits, such as quantifying uncertainty through credible intervals and incorporating expert knowledge in data-sparse scenarios, these advantages are often outweighed in routine house valuation by the need for objective, transparent, and timely results. Frequentist methods align better with standard practices in individual appraisals and mass assessments, contributing to the limited adoption of Bayesian approaches outside academic or specialized research settings."

When you hear someone talking about using Bayesian this or that for appraisal - you are talking to a charlatan. Sad to say.
 
Calm down, I don't understand that fancy stuff either. It's TrueTracts talk, not Non Sequitur talk.
ARE YOU SERIOUS?


This is CLEARLY NOT USPAP compliant. Holy Christ! HOLY CHRIST!

Honestly, it was at least well over 15 years ago, after some conference presentation, that GD asked me if I had ever used Bayesian Inference. From the get-go, I knew that it had very serious problems. It is one of those things based on probabilities. Yes, if you were very careful, very detailed, and in specific cases, it could work. But it is a VERY risky to implement except in rather simple cases. I wouldn't trust any appraisal based on it, except perhaps for some clear-cut cases, even then they are doing all kinds of extra work to filter out bias, e.g., additional sensitivity analysis, hierarchical modeling, and so-called regularization of priors.

You could say its weakness is its dependency on assumptions about prior distributions.

I would say, it just can't compete with MARS in so many ways. MARS is non-parametric: It doesn't depend on prior assumptions about distributions. MARS is fast. MARS does your categorization and finds the most important variables that impact the target variable (SALE PRICE). And it is highly accurate. And a few other things.

Bayesian Inference has to struggle to try to get close.
 
Last edited:
Speaking of Bayesian Probability, crazy thing happened this morning.
Y'all know I'm not a morning person but I woke up early today and started thinking (I do that trying to plan my day).
Then I worried about my Tesla and how I charged it last night.
I figured it's too late to fix whatever problem I thought and just stay in my warm bed with my wife.
But I had to go downstairs and check my Tesla app on my phone.

When I got to my office at 6:51 AM, the phone rang and I recognized it was from my tenant.
I answered the phone and resolved the issue which could have been a major issue.
What are the odds of that? Of all days and morning, I must be the luckiest person.
 
Find a Real Estate Appraiser - Enter Zip Code

Copyright © 2000-, AppraisersForum.com, All Rights Reserved
AppraisersForum.com is proudly hosted by the folks at
AppraiserSites.com
Back
Top