qualifying all of the pertinent property details to pump into the math machine still requires a good deal of professional judgement.
Of course. MARS will create a sometimes complex regression model - although I have to always work to simplify it as much as possible to avoid overfitting. Then I go through the 200+ comparables, create the price estimates from the model based on measurable attributes, and the residual differences between the estimate and the actual sale price. Then I rank the comps by the residual and compare the properties to their ranking. The most appealing properties will tend to have the highest residuals and the least appealing the lowest. But sometimes you have to wonder just why, for example, some property has an especially high residual
Now backup, an interesting thing I have discovered is that there is a pretty good correlation between the Sale Price/List Price ratio and the residual. In fact, with the current appraisal in Ocean Colony (Half Moon Bay) I was getting a correlation of 70%, which surprised me. That is not bad.
Anyway, I was looking at a house with a high residual. It sold around the beginning of 2022 way above the MARS estimate. There were very few homes sold at that time. But one thing I noticed was that the sale price was much higher than the list price. The Sales Agent had put the house on the market with a list price that was pretty close to the MARS estimate. That kind of makes sense, they probably based their estimate on GLA and to a lesser extent on their subjective judgment of its appeal (which apparently they weren't sure of). The house had been upgraded and from the photos it was impressive. But still, I was not at all sure it deserved that much higher of a price. But with so few houses available to purchase, it appears there was a bidding war that pushed the final sale price way above list. That's market value. That is very real. You can't discount it. Who am I to say the buyers were wrong? It is what it is. So, this house had a CQA of 9.9 out of 10.0 - max more or less.
That residual in my RCA method - it does NOT miss a thing. It captures 100% of everything. See the Residual - that is the "other" stuff. You can split it any way you want between condition, quality, functional utility, view or whatever - and the value conclusion will not change.
It is 100% - the only entry for bias is in your judgment of the CQA for the subject property -- Just go to your ranking and find which two properties it fits best between.
You have to see it, I suppose, to believe it. But it is very real.
Now going back to the point I think you were trying to make --> If you don't know how to use MARS, --> YES, you will find that your ranking of the residuals does not match appeal. It will be out of whack. So, you need skill. To interpret deviations, you need a good understanding of real estate and appraisal.
You need a rare combination of skills.
