Market areas and Market Trends

[djd09]

tds off your own comments...

 

[RCA]

View attachment 93406

That graph looks just like my MARS regression graph for date of sale!!!

 

[RCA]

Their example brings up more questions than answers. How were these rolling trend data points determined? What does "Comparable sale 1=5% increase as of contract date?" mean? Is this the yearly appreciation rate the comparable sale experienced as measured from its prior sale?

Other questions:

• Why does the best fit line have pivots on phantom data points?
• Why is there a trend line starting at 1% from the origin to comparable 1?
• How was the market from comp 4 to the effective date determined to be stable?
• Why does the 'Decreased value' line not follow the slope of the trend line from comp 2 to comp 4?

They have to be using MARS regression!

With MARS regression you will get an equation for the graph to give you the value contribution for the subject and comparable sale dates. Just take the difference and that is your adjustment. You can use R or Python to move all these equations into an Excel grid for easy computation.

But the real catch is that in any case, you need to understand how to use MARS, regardless of what ever software someone dumps on you

Remember YOU are always going to be the one liable for deficiencies - even if it is really the fault of the software. Software companies NEVER accept liability for failures of their own software. Think hard about that - you frigging sucker!

 

[RCA]

They have to be using MARS regression!

With MARS regression you will get an equation for the graph to give you the value contribution for the subject and comparable sale dates. Just take the difference and that is your adjustment. You can use R or Python to move all these equations into an Excel grid for easy computation.

But the real catch is that in any case, you need to understand how to use MARS, regardless of what ever software someone dumps on you

Remember YOU are always going to be the one liable for deficiencies - even if it is really the fault of the software. Software companies NEVER accept liability for failures of their own software. Think hard about that - you frigging sucker!

Actually, they are using % instead of expected value contribution. 0% is for the sale prices at the beginning of the comparable sale period being used. The percentages to the right are the percentage increase or decrease with respect to the beginning of the period. I would just use the average sale price. The result comes out some think better with % because it would compute based on comp sale price. A blip … what are you going to use for the subject sale price when it doesn’t exist? The better approach is to go to second degree regression and regress on DateOfSale+GLA, BathCount, and maybe a few other variables and then aggregate the individual adjustments into one date adjustment, showing your calculations - somewhere.

 

[RCA]

The smartest appraisers at Fannie Mae and Freddie Mac - are, at best, only half-smart.

 

[Fernando]

View attachment 93708


Maybe. maybe not.....

Before pandemic, I was looking at bank properties as safe investments. Too large to fail.
The cap rates were low but strong tenants and good if they continue to stay there.
I did notice more Citibanks were for sale and thought that bank was in trouble.

 

[Tom D]

Does regression take this into account,

Russian money laundering in the UK has pushed up house prices, Yvette Cooper and David Lammy have said.

In an article for The Telegraph, the Home Secretary and Foreign Secretary said that the Government aimed to “call time” on London’s financial system being used as a clearing house by criminals and London property being used as “Bitcoin by kleptocrats”.

 

[RCA]

Does regression take this into account,

Russian money laundering in the UK has pushed up house prices, Yvette Cooper and David Lammy have said.

In an article for The Telegraph, the Home Secretary and Foreign Secretary said that the Government aimed to “call time” on London’s financial system being used as a clearing house by criminals and London property being used as “Bitcoin by kleptocrats”.

Regression takes EVERYTHING in the input into account, as long as it impacts the value of the target variable, - i.e. sale price. Yep.

So, if you want it to take into account news on money laundering, then create a column for it and some kind of value system for "money laundering". Think. You will need the date the news broke, several delay factors (e.g. weeks or months). It can be done. It might provide results. You will have to be creative, no doubt. But even if you don't do this, as long as you have date of sale, anything that occurs over different dates that disrupts sale price trends will have an impact; only, you will have to do some research to discover whether it was the hurricane, stretch of heavy rain, Trump getting elected again, an earthquake, forest fires, war or whatever.

 

[Fernando]

Regression takes EVERYTHING in the input into account, as long as it impacts the value of the target variable, - i.e. sale price. Yep.

So, if you want it to take into account news on money laundering, then create a column for it and some kind of value system for "money laundering". Think. You will need the date the news broke, several delay factors (e.g. weeks or months). It can be done. It might provide results. You will have to be creative, no doubt. But even if you don't do this, as long as you have date of sale, anything that occurs over different dates that disrupts sale price trends will have an impact; only, you will have to do some research to discover whether it was the hurricane, stretch of heavy rain, Trump getting elected again, an earthquake, forest fires, war or whatever.

More significant would be mortgage rates. Do you have rates as your variable? If so, does it have much affect on sales price?
Instead of many detail variable, there are limited number of variables in each market which have most affect on price.

 

[Joe Flacco]

Regression and avms can't take everything into account because there are differences that are not tracked or organized in a database.

I just did one where the most recent comp sold for like 10% less compared to other comps. The reason is it's on a hill and you have to park on the street and walk up 25 steps to get to the door.