Principle of Substitution, ....

[TerryRohrer]

That is the point of narrowing it down to 3-6 properties. Why would you want to add more variables and then just pretend they do not exist?

Well, you will find the glaring holes in you myopic approach if you ever get outside the cookie cutters!

 

[RCA]

I'm no expert on it but know enough about it. But I understand real estate and the data that is databased, organized, and available for regression or MARS. That type of analysis ignores anything that is not organized in a database to be analyzed. You pretend the countless number of variables do not exist. Plus the variables are dynamic depending on combination of other variables. Real estate market is very complex.

The above is very mixed up and I really have no idea what you are saying. I enter all possible MEASUSRED variables into the regression - and it decides which are significant through intensive analysis. These varaiables are typically:

Lot Size
GLA
HOA Fees
Bathroom Count
Bedroom Count
GIS Coordinates (Longitude rounded to 3 decimals, Lat round to 3 decimals) Or Location.
MLS Area Name
# of Fireplaces
Age
Garage Spaces (Bays)
Carport Bays
Parking (cars)
Stories
Pool (Y/N)
SaleDate (as days before effective date)
Frontage
Style

From the above MARS (R/earth) will decide which variables are significant anc create a model from them)

From the model, my R program will create price estimates and residuals for the 100+ comps. (And understand I have two kinds of so-called "comparables" - those that go into the regression and then those that go into the sales grid.

The comps will be ranked according to the size of the residual and then receive CQA (or Residual) Scores from 0.00 to 10.00. From the scores vs residuals and scores vs residual/sf, hash functions will be created along with graphs.

The residual ranking you will see, if you have a good regression model, OBJECTIVELY rank the comparable properties from those with the most negative residual or lowest quality/condition to those with the highest residual or highest quaity/condition. You should be able to place the subject in that ranking fairly easily by comparing its photos to those of the comparables and then it give it a CQA score between the lower and the upper. From that, you can also assign a residual value (I would use Residual/SF * GLA).

The next step, for each comparable in the sales grid, is to break the residual down into UNMEASURED VARIABLES such as:

Condition
Quality
Design
Functional Utility
View (this is the portion of View value that remains after taking account of location which is handled by MARS)
Yard/Landscaping?
Any other features that didn't go into the regression that you deem important

Remember the total value of all value contributions for these unmeasured attributes must add up exactly to the residual value.

Once you have all value contributions for the comparables and the subject - then you can take the differences between the subject and comparable values to get the adjustments.

then add all adjustments to the comparable sale prices to get the adjusted sale prices - and those should be exactly the same for all comparables.

 

[Joe Flacco]

The above is very mixed up and I really have no idea what you are saying. I enter all possible MEASUSRED variables into the regression - and it decides which are significant through intensive analysis. These varaiables are typically:

Lot Size
GLA
HOA Fees
Bathroom Count
Bedroom Count
GIS Coordinates (Longitude rounded to 3 decimals, Lat round to 3 decimals) Or Location.
MLS Area Name
# of Fireplaces
Age
Garage Spaces (Bays)
Carport Bays
Parking (cars)
Stories
Pool (Y/N)
SaleDate (as days before effective date)
Frontage
Style

From the above MARS (R/earth) will decide which variables are significant anc create a model from them)

From the model, my R program will create price estimates and residuals for the 100+ comps. (And understand I have two kinds of so-called "comparables" - those that go into the regression and then those that go into the sales grid.

The comps will be ranked according to the size of the residual and then receive CQA (or Residual) Scores from 0.00 to 10.00. From the scores vs residuals and scores vs residual/sf, hash functions will be created along with graphs.

The residual ranking you will see, if you have a good regression model, OBJECTIVELY rank the comparable properties from those with the most negative residual or lowest quality/condition to those with the highest residual or highest quaity/condition. You should be able to place the subject in that ranking fairly easily by comparing its photos to those of the comparables and then it give it a CQA score between the lower and the upper. From that, you can also assign a residual value (I would use Residual/SF * GLA).

The next step i for each comparable in the sales grid, to break the residual down into UNMEASURED VARIABLES such as:

Condition
Quality
Design
Functional Utility
View (this is the portion of View value that remains after taking account of location which is handled by MARS)
Yard/Landscaping?
Any other features that didn't go into the regression that you deem important

Remember the total value of all value contributions for these unmeasured attributes must add up exactly to the residual value.

Once you have all value contributions for the comparables and the subject - then you can take the differences between the subject and comparable values to get the adjustments.

then add all adjustments to the comparable sale prices to get the adjusted sale prices - and those should be exactly the same for all comparables.

Yes, exactly. You analyze only 20 or so number of variables and then pretend there are no other variables. According to Terry, the number of variables is countless. He is correct.

 

[TerryRohrer]

Yes, exactly. You analyze only 20 or so number of variables and then pretend there are no other variables. According to Terry, the number of variables is countless. He is correct.

And you select a few sales that are similar in 6 variables and assume all and the subject are identical in the rest. Vastly inferior methodology that is highly misleading.

 

[Joe Flacco]

And you select a few sales that are similar in 6 variables and assume all and the subject are identical in the rest. Vastly inferior methodology that is highly misleading.

I can't really blame you for anything since you live in Montana. How far do you live from the nearest human being?

 

[TerryRohrer]

I can't really blame you for anything since you live in Montana. How far do you live from the nearest human being?

I guess this is where you go when you don't understand the discussion and can't answer any of the questions?

 

[Joe Flacco]

I guess this is where you go when you don't understand the discussion and can't answer any of the questions?

I'm just saying it doesn't seem like you have a grip on comparable selection and the impact on valuations. I am pretty sure it is because you live in Montana.

 

[TerryRohrer]

I'm just saying it doesn't seem like you have a grip on comparable selection and the impact on valuations. I am pretty sure it is because you live in Montana.

And yet you have been unable to answer a single question in this discussion. Sewer gas getting to you?

 

[Joe Flacco]

SSO | Bright MLS

SSO | Bright MLS

matrix.brightmls.com

This is really straight forward. Let's use this house as an example. It's a 100 year old rowhouse. 18' wide and one parking space at the alley. Two bathrooms upstairs. There are more than a hundred rowhouse sales in the neighborhood. In selecting the comparables, I am looking for these features.

Original floor plan
Original hardwood floors
Original transom windows
Original trim
Original interior doors and hardware
updated in the last 10 years
refinished lower level

You can't get more reliable results than narrowing it down to the 3-6 sales out of 100 in the neighborhood that has these features.

 

[TerryRohrer]

SSO | Bright MLS

SSO | Bright MLS

matrix.brightmls.com

This is really straight forward. Let's use this house as an example. It's a 100 year old rowhouse. 18' wide and one parking space at the alley. Two bathrooms upstairs. There are more than a hundred rowhouse sales in the neighborhood. In selecting the comparables, I am looking for these features.

Original floor plan
Original hardwood floors
Original transom windows
Original trim
Original interior doors and hardware
updated in the last 10 years
refinished lower level

You can't get more reliable results than narrowing it down to the 3-6 sales out of 100 in the neighborhood that has these features.

My original question...how do you measure reliability?