Are Price Estimates EVEN Accurate Anymore?

Today I want to talk about price estimates, and have a conversation around algorithms estimating prices, as well as the pros and cons to something like Zillow’s Zestimate, a Redfin estimate, or these other syndicate websites that have public facing price estimations for the value of a home.

First, we need to lay the ground rules for what the true value of a home is. The only number for the value of a home is the highest amount of money that a willing and able buyer chooses to spend. That is because real estate is real property, which differs from personal property insofar as there can never be more than one of a single thing. That single thing in this case is a house, but more importantly, the location of that house, or the land that it is on. You may be able to build the same home over and over and over again, but it cannot take up the same space as another home, and therefore cannot be exactly the same as another piece of real estate as real property.

It’s important that we define this because, and if you’re starting to catch on you can see where I’m going with this, there is no method, algorithm, or entity that can tell you the value of a home. Period.

Estimating the value of a home exists in all sorts of capacities and people will always try to tell you one is more accurate than another, etc etc. However one change happened recently in this market that is putting these estimates at the forefront...or rather, the fact that they are no longer at the forefront is very telling.

See, as a licensed real estate agent, I have access to all sorts of tools that consumers do not. I have internal programs that help me search and give me information. But the fact is, companies like Zillow and Redfin have made entire business models on user interfaces dedicated to bringing consumers information faster and more clearly, and quite frankly, they do a pretty damn good job, I can’t lie. I will very often use Redfin simply because it can deliver information that I know is already public, pretty quickly and easily. And of course, Redfin is one of the popular sites that very clearly displayed their “Redfin Estimate” boom, right there at the top of every single house. Until recently.

First off, they have relocated the estimate from the top to pretty far down towards the bottom. Zillow is often doing this as well, they’re no longer prominently displaying their Zestimates right at the top, and often you’ll get the little robot scratching his head and having no displayed price at all. In short, what Zillow is saying to the consumer is “we don’t have enough information to give an estimate”. That really isn’t the truth. What is the truth is their algorithms are calculating a price that is significantly, sometimes hundreds of thousands of dollars less than what the home is listed for.

Now, previously, you might have seen a Redfin estimate at say 10, 20 thousand dollars less, maybe 30 less than the list price and a few years ago they might have been right sometimes. The house might need a price adjustment and it might drop. Or it was just wrong but hey its an estimate, estimates don’t have to be completely correct right?

But here’s the problem. Consumers are starting to take these estimates as gospel. They’re starting to think any time an estimate is lower or even higher, that that is the true value no matter what, no questions asked. I mean, after all, these companies are pumping millions of dollars into these websites and algorithms, they have to count for something. Plus, for most people their home is the most valuable thing that they own. No one should feel comfortable making decisions based on estimates that are “pretty close”

I don’t know for sure exactly how each company’s estimation tools work but I would imagine they are some combination of two things: traditional comparable sales analysis and some type of machine learning algorithm. With a comparable sales analysis, past homes that have sold are compared to this particular home you want to know the value of, you try to put a dollar amount on differences between those homes and come to some kind of comparable value. You try to select the homes closest in area, size, type, condition, etc and build out your analysis from there. But despite the fact that this is the method almost all appraisers use, it still has its faults. How much is a bathroom remodel worth? Well we might be able to put a pretty good dollar amount on that. But how much is it worth to be at the end of a cul-de-sac on a quiet street versus a main road? How much are mature trees worth? How much is it worth to have a master suite instead of a detached master bathroom? How much is north versus south facing? And of course so many aspects can appeal to different people in different ways. Some people love a pool, major bonus. Some people devalue it because it is costly to maintain, could go either way.

And as for the machine learning part? I would imagine that it is pretty good in areas of the country where the market moves very slowly, with lots of very similar track homes being built in predicable ways and patterns. So for Middle America? Sure, I’d trust an algorithm. Los Angeles? Forget about it. Even your best models couldn’t keep up with the pace of this market and the wild diversity of homes and neighborhoods in this city.

See, a few years ago, I started to get the idea that these estimates were off so significantly that they were starting to do damage to consumers instead of helping them. In previous markets, they might have overestimated the value of a home, which often happened to me in areas of Los Angeles where a home was on the border of a neighborhood or even more importantly, a city. A great example: I sold a home in South Pasadena where the Redfin estimate was lower than my list price, and I sold it over list price. But Redfin was pulling all of these comparables from Pasadena since the home was close to the border.

You might thing, eh, Pasadena, South Pasadena close enough right? Nope. See, they are actually two completely different cities, the city of Pasadena and the city of South Pasadena. One of them has an award winning school district where every school elementary through high school is 10/10 on Great Schools. The other? Not so much. Not bad, but nothing special. So of course, access to that amazing school district through property ownership in South Pasadena wouldn’t just be a little off, no big deal. It’s a HUGE deal, a massive difference in real estate values. Not to Redfin’s algorithm apparently.

I’ll give you another fantastic example. I recently sold a custom built 1959 mid century home that was a truly special one of a kind architectural gem. I listed at 949k and sold it for 1.215 Million. Why? Because I understand the value of Mid Century Modern architecture when it is a truly one of a kind home. No estimate understood that. Zillow had the home estimated at 886k the entire time it was listed. If you were an uneducated consumer not working with a smart and connected real estate agent, you might have thought the home was way over priced and might have tried to lowball. That wouldn’t have gotten you very far.

Now of course, I’m doing a lot of harping on these estimates being low, but that is simply the market we are in right now. It’s moving fast, prices are appreciating very quickly, and not even appraisers can keep up. But there will be a time in which the market stabilizes and we enter a time where the estimates are overestimating value. It’s happened in the past and certainly can happen again in the future. The easiest scenario to imagine is that none of these algorithms can know the state of a home. They can’t tell what shape the roof is in, if the paint is a dated color, if the kitchen has cracked tiles, if the water heater works. But here’s the thing, a human being can tell that instantaneously by doing some easy reconnaissance. Sometimes you can even tell from photos.

Home is personal and houses are unique. For the time being and the very significant future, this is a human driven business because only human beings can understand all of the little intangible things that make real property different from personal property, and more importantly how to truly value the space you live, work, play, cook, sleep and dream in.

Thank you for watching, I hope you learned something about home value estimations, and if you did be sure to like and subscribe, and if you found this valuable, I’d incredibly appreciate you sharing this video with someone that could benefit by connecting with me. It’s super simple, and on my website there’s a button to Book A Consultation that connects directly to my calendar.

At Stephens Real Estate, we offer a few value adding programs for both buyers and sellers that no one else is offering.

For buyers, we put together a full marketing plan that paints you as the right buyer for your dream home. We can help recommend local lenders that will fit your financing situation, even if you’re freelance. And don’t worry about the inspections, we’ve got you covered. They’re free, all of them, no questions asked.

And If you’re looking to sell your home, we offer the ability to REMODEL NOW, and PAY WHEN YOU SELL with Zero fees, no interest, no upfront costs. Or, if you just want to freshen up a bit, we offer the HANDYMAN SPECIAL where I pay $1,500 towards all the little fixes can yield up to a 10x return. Plus, we pay for all of the aspects of digital marketing from photography, drone footage, videography, 3D walkthrough and virtual open houses, social media campaign, custom website, targeted emails and more.