Are Appraisals Really Relevant?

[glenn walker]

Appraisers mostly live in a world in which they are focused on one property and one borrower. Few ever thought about why the Avocados are not sold by the LBS weight or size. The sign says $1.50 per avocado take a small one or big one a fresh one or a old one still $1.50 each. Why Not sold like most other fruit's and vegetables by the LBS weight. That's mortgage backed securities the risk not individual values.

 

[ZZGAMAZZ]

In certain cases appraisals are valuable. In most they aren't anymore. In the case of difficult or high dollar properties left over will migrate to MAIs and SRAs.

Mine is prolly a naive perspective, but isn't the absence of "condition" considered to be a failure of AVMs? I ask bc that almost ALWAYS is the primary adjustment factor in every residential appraisal I conduct.

 

[glenn walker]

Now AI can be used to assess the Avocado Delema. Lmao )

 

[J Grant]

Curious: How is "error rate" defined? Who publlished that statistic? And, what is the corresponding rate for appraisal-based values? [Really interesting post]

What is the benchmark of the "error" compared against? A price median or average ? That is the problem with these programs - they are not thinking conceptaully about market value, they are using numbers with averages, mean, etc - an AVM develops a range , with a degree of confidence about the point value.

 

[J Grant]

I personally think AI can be used as an integration and anlayss tool with an appraisal rather than replace the appraisal (or appraiser )

 

[George Hatch]

Instead of telling the users what they should think about the utility of appraisals it might be instructive to ask them what they actually do think of the utility of appraisals.

It literally doesn't matter what the investors or the RE brokers or other outsiders think about the lenders' usage of appraisals. Nobody asked them their opinion on the matter and nobody cares about their opinion.

 

[J Grant]

Appraisers mostly live in a world in which they are focused on one property and one borrower. Few ever thought about why the Avocados are not sold by the LBS weight or size. The sign says $1.50 per avocado take a small one or big one a fresh one or a old one still $1.50 each. Why Not sold like most other fruit's and vegetables by the LBS weight. That's mortgage backed securities the risk not individual values.

It's an inane comparison because one avocado is like the other, easily interchangeable, whereas properties are vastly different from each other.
Mortgage-backed securities may be sold in bundles; however, compare the individual property of any loan, a property worth hundreds of thousands of $ each, vs a fifty-cent avocado.


When each property backing a loan is worth hundreds of thousands of loaned money, more than a few valuations being "off" can really impact things.

 

[glenn walker]

It's an inane comparison because one avocado is like the other, easily interchangeable, whereas properties are vastly different from each other.
Mortgage-backed securities may be sold in bundles; however, compare the individual property of any loan, a property worth hundreds of thousands of $ each, vs a fifty-cent avocado.


When each property backing a loan is worth hundreds of thousands of loaned money, more than a few valuations being "off" can really impact things.

It completely went over your head and actually just the opposite as no two Avocados are the same size or weight but not sold by the pound. Don't overthink it you'll get a headaches or maybe if you do think through it a Bell will ring.

 

[Elliott]

Curious: How is "error rate" defined? Who publlished that statistic? And, what is the corresponding rate for appraisal-based values? [Really interesting post]

I asked Perplexity what Redfin and Zestimates error rate was.


"How Zestimates ‘Error Rate’ Is Determined

Zestimate error rate refers to how closely Zillow’s property valuation algorithm matches actual sale prices of homes. This is a statistical measurement, reflecting Zillow’s ongoing effort to track the accuracy of its automated estimates.
Calculation Method
• Zillow calculates the median error rate by comparing its Zestimates (the prices it predicts) with the actual sale prices of homes that have recently sold.
• For instance, if the Zestimate was $500,000 and the home sold for $510,000, the error would be (|500,000 - 510,000| / 510,000), or 1.96%. Zillow runs this calculation for a large sample of recent transactions, then finds the median value from all those error percentages.
• Median error rate means that half of the Zestimates are within this percentage of the sale price, and half are not.
Different Rates for On- and Off-Market Homes
• For on-market homes (actively listed for sale), the median error rate is 1.83–1.94%.
• For off-market homes (not currently listed), the error rate is higher: around 7.01–7.06%.
What “Error Rate” Indicates
• If the error rate is 1.9%, this means half of Zestimates for on-market homes are within 1.9% of the actual sale price, and half are more than 1.9% off. The same logic applies to the 7% error rate for off-market homes.
• Zillow also reports the percentage of homes where the Zestimate was within 5%, 10%, or 20% of the sales price. For most major markets, more than 95% of Zestimates fall within 10% of the final sale price.
Factors Affecting Error Rate
• **Accuracy depends on available ** Areas with more MLS data, recent comparable sales, and user-fed real estate information have lower error rates.
• Uniqueness of properties: Unique or rarely sold properties usually have higher error rates, especially if recent comparable sales are unavailable.
• Market activity and data quality: Frequent sales and good, up-to-date data improve algorithm accuracy.
In essence, the Zestimate error rate is a statistical comparison of predicted property values versus real sales, reported as a median percentage across thousands of transactions. It’s regularly updated and varies based on market conditions, data coverage, and property type."

 

[J Grant]

I asked Perplexity what Redfin and Zestimates error rate was.


"How Zestimates ‘Error Rate’ Is Determined

Zestimate error rate refers to how closely Zillow’s property valuation algorithm matches actual sale prices of homes. This is a statistical measurement, reflecting Zillow’s ongoing effort to track the accuracy of its automated estimates.
Calculation Method
• Zillow calculates the median error rate by comparing its Zestimates (the prices it predicts) with the actual sale prices of homes that have recently sold.
• For instance, if the Zestimate was $500,000 and the home sold for $510,000, the error would be (|500,000 - 510,000| / 510,000), or 1.96%. Zillow runs this calculation for a large sample of recent transactions, then finds the median value from all those error percentages.
• Median error rate means that half of the Zestimates are within this percentage of the sale price, and half are not.
Different Rates for On- and Off-Market Homes
• For on-market homes (actively listed for sale), the median error rate is 1.83–1.94%.
• For off-market homes (not currently listed), the error rate is higher: around 7.01–7.06%.
What “Error Rate” Indicates
• If the error rate is 1.9%, this means half of Zestimates for on-market homes are within 1.9% of the actual sale price, and half are more than 1.9% off. The same logic applies to the 7% error rate for off-market homes.
• Zillow also reports the percentage of homes where the Zestimate was within 5%, 10%, or 20% of the sales price. For most major markets, more than 95% of Zestimates fall within 10% of the final sale price.
Factors Affecting Error Rate
• **Accuracy depends on available ** Areas with more MLS data, recent comparable sales, and user-fed real estate information have lower error rates.
• Uniqueness of properties: Unique or rarely sold properties usually have higher error rates, especially if recent comparable sales are unavailable.
• Market activity and data quality: Frequent sales and good, up-to-date data improve algorithm accuracy.
In essence, the Zestimate error rate is a statistical comparison of predicted property values versus real sales, reported as a median percentage across thousands of transactions. It’s regularly updated and varies based on market conditions, data coverage, and property type."

An error off a sale price is not the same thing as an "error " off a market value opinion.