coding a "Best Deals in real estate" algorithm

[coastal]

I am not a programmer but I am curious as to what language you would write a algorithm in to judge best deals from IDX data and what would the code look like?

For instance if I wanted to pick the best deal in a community with 10 homes for sale. I guess some of the factors you could look at would be...
time on market - longer may = more likely to accept a low offer.
Price - obviously the lower the price the better.
Square footage - determine the price per square foot
Taxes - in the same community they should all be about the same but if comparing one community vs. another it would be a factor.

I guess a lot of this would be similar to an assending-decending sort, just with a few more variables thrown in. Different variables could be weighted more or less depending on their importance.

Anyway I apologize for my ramblings but I am very interested in knowing how someone would go about this.

[DMike]

Here's how i would do it
- i would make a config area where the user could enter his details what he wants the "decision" to be made on - price, square footage, taxes ...
After that i would a search in the database with those filters and give the best option for the user

[jessiesc]

I am not a programmer but I am curious as to what language you would write a algorithm in to judge best deals from IDX data and what would the code look like?

I would use a simple SQL query.

For instance if I wanted to pick the best deal in a community with 10 homes for sale. I guess some of the factors you could look at would be...
time on market - longer may = more likely to accept a low offer.
Price - obviously the lower the price the better.
Square footage - determine the price per square foot
Taxes - in the same community they should all be about the same but if comparing one community vs. another it would be a factor.

This is the rub though isn't it...

While these variables will produce results, will they produce correct or expected results?

What I mean is, how do you define 'best deal'?

1. Is 'best deal' defined by price? If so, the query is simple.
2. Is 'best deal' defined by price per square foot? If so, the query is simple.
3. Is 'best deal' defined by lowest taxable value? If so, the query needs to calculate the new tax, a bit more complicated but doable.
4. Is 'best deal' defined by the most upscale neighborhood? If so, again an easy query.

Perhaps as you point out, 'best deal' may be defined as:
Longest time on market (low offer ability) + lowest price + lowest price per square foot + least amount of taxes + within area 'X'.

What about the following factors:

• Upgrades to property - and which upgrades?
• Amenities of property - Pool, spa, sauna etc?
• Amount of land

Are we talking primary residence or income investment?

IMO, this is where many of these 'best deal' alternatives fail. The target is not fully quanitifiable unless it is narrowed down.

What is a 'best deal' to one, may not be a deal at all to another.

IMO, this algorythm is subjective to too many variables and is best made into sub categories:

• Lowest price for a primary residence in community 'X'
• Lowest price per square foot in community 'X'
• Lowest price per square foot with the most land in community 'X'
• Lowest taxes for a primary residence in community 'X'
• Property most likely to accept 'low offer' in community 'X'

etc.

[coastal]

It would be interesting to build a algo that learned. It could ask their primary concerns and then go from there, like a virtual agent.

Because if we can think it we should be able to write a program that simulates our thinking, but I'd like to go way beyond just sorting database info.

Other factors might include which neighborhoods are hot sellers. Of my clients X amount who make a purchase end up buying in y neighborhood.

I'm guessing it would take months to document the entire process and then a year to write a program that could decipher it all, and then in the end you still have to see the property in person and compare it to others to make an educated, informed decision. And as you mentioned what one person loves another may hate.

Off on another tangent you could also add some social networking... Home owners could log on and gauge their willingness to accept lower offers, neighbors could log on with neighborhood ratings. Neighborhood pool = 1 - 5 stars, neighborhood kids = obnoxious - Sweet and helpful etc... The list is endless. Maybe when I'm mega rich and have a few programmers with nothing better to do I'll set them on it or go back to school and learn how to do it myself.

[nemaroller]

I'm a programmer, and I'm laughing because coastal is an agent promoting the use of a 'virtual' agent?

What? You want us programmers to replace your job too?

[jessiesc]

I'm a programmer, and I'm laughing because coastal is an agent promoting the use of a 'virtual' agent?

What? You want us programmers to replace your job too?

You'd be surprised (or perhaps not) how many times I am asked to create this type of program for offices.

[DMike]

jessiesc, could you get more accurate info on it ? I mean it would be nice to see what exactly the offices are looking for, what exactly are they interested in, that way we can build exactly what they want

[jessiesc]

DMike,

I am under an NDA (I also design software) and can not disclose any proprietary details as I have a client that wants something similar but not as virtual as described above.

However, if you search around the forums here you'll find lots of similar requirements by many of this communities members.

Most of the brokers wish lists here should give you a pretty good idea of what they want.

[EFS Webmaster]

Another suggestion (that zillow haters aren't going to like) is to look for the largest delta between the Zillow "Zestimate" and the actual listing price.

Even if you despise Zillow, consumers seem to like the concept of the zestimate and the "best deals" could essentially be those that showed the largest variance. Zillow makes an API available for developers so I would think something like this would be feasible unless they have limitations on how their zestimates can be utilized.

You would need to package this as a 'virtual' deal finder / toy and reinforce the fact that an agent should be consulted for accuracy.

[Joe Cline Austin]

probably would be cheaper to run some sql queries with your general criteria and then just flip through the results to hand pick the ones you know are good from experience..