Which is more accurate: Zillow’s Zestimate or Redfin’s Estimate?
When you type in a home’s street address to obtain an online valuation from the two biggest players in the field — Zillow’s Zestimate or Redfin’s Estimate tools — how good is what you get?
Both are used by millions of home shoppers, owners, realty agents, anyone curious about what a house in their neighborhood might be worth. Both also have been criticized for estimates that are off the mark; some homeowners have actually sued Zillow over their Zestimates, although unsuccessfully. Zillow’s own chief executive, Spencer Rascoff, famously sold his Seattle home for 40 percent below its Zestimate.
Accuracy matters a lot in this arena because many buyers and sellers use the online estimates to price their homes or make purchase offers, literally handing sellers or buyers the estimates as part of their bargaining strategy. This is despite both companies’ warnings that these are not appraisals, only algorithm-based computer estimates. They are starting points, not holy writ.
So which company’s estimate is the more accurate?
For two years, Redfin has claimed that it produces estimates that are superior, based on the results of an independent study. When it values homes that are on the market, Redfin says its median national error rate is just 1.77 percent. That is, the selling price, compared with the estimate, is within that margin of error half the time. On houses that are not for sale, Redfin’s median error rate is 6.66 percent. Redfin has a total of 74.4 million properties in its valuation database — 1.3 million on the market and listed for sale, 73.1 million off the market.
[How accurate is Zillow’s Zestimate? Not very, says one Washington-area agent.]
But it looks like bragging rights for accuracy could be shifting to Zillow.
Following an international contest involving teams of data scientists, Zillow announced that its median error rate on valuations of the 110 million U.S. homes in its database will soon drop to 4.0 percent or even below, from the current 4.5 percent. Zillow does not provide a breakout that distinguishes between its error rates for homes already listed on the market and off-market homes, so there is no direct comparison to Redfin’s claimed 1.77 percent figure for listed houses. But the overwhelming majority of homes in Zillow’s 110 million-property database are off-market, which are more challenging to value because there’s usually less detailed information available on them.
Note the difference in Redfin’s 1.77 percent error rate for listed homes vs. its 6.66 percent rate for off-market homes. Given this, Zillow’s claim that it will have a 4 percent composite error rate on 110 million homes — the vast majority of them off-market — looks better.
An error rate of 4 percent or less would put Zillow close to a standard that many appraisers consider passable for their own work. Ryan Lundquist, an appraiser in the Sacramento, area, told me that for many colleagues, a 4 percent median error rate “would be a fairly acceptable range.”
Pat Turner, an appraiser in the Richmond market and a longtime skeptic about automated valuations, says the only way Zillow could ever get to a median error rate of 4 percent would be in “cookie cutter” subdivisions, where houses are similar and comparable properties are plentiful. In neighborhoods with greater diversity of home types, ages, interior improvements and land sizes — or in nonurban areas where comparable homes and data are hard to find — he seriously doubts the claim.
[Redfin launches a service to compete with Zillow’s valuation estimator]
Does it really matter what these companies say about improvements in their error rates? Absolutely — if you make use of Zillow Zestimates or Redfin Estimates. If they don’t produce value estimates you can rely on within their published error rates, why would you waste your time looking at them?
But remember: “median national error rate” can be a tricky concept. “National” does not mean your local market. Your neighborhood may have a much better — or far worse — error rate than the national medians.
Before using either tool, it’s a good idea to go their Web pages and check how far off their estimates tend to be where you live. You can find them at Zillow.com/#acc and Redfin.com/redfin-estimate.
And focus on the key term “median.” In Chicago, the median Zestimate error rate is an impressive looking 3.8 percent; but 41.4 percent of Zestimates are not within 5 percent of the actual sale price. That’s sobering. In Washington, D.C., the median error rate is 3.1 percent. But fully a third of Zestimates aren’t within 5 percent of being accurate.
Ken Harney’s email address is harneycolumn@gmail.com.
Read more Harney:
Study claims older homeowners are preventing millennials from buying houses. But is that true?
Large breach of mortgage borrowers’ data raises new concerns, questions
Your FICO score doesn’t always preordain your mortgage chances
Napolitano says if NYT report is accurate, Trump may be implicated in attempted obstruction
Fox News judicial analyst Judge Andrew Napolitano on Tuesday considered a New York Times report that claimed President Trump tried to interfere in an investigation into his former personal attorney Michael Cohen, and the analyst said if the report is accurate Trump could be implicated in attempted obstruction.
The New York Times report claims that Trump called Acting Attorney General Matthew Whitaker last year to ask if Geoffrey S. Berman, the United States attorney for the Southern District of New York and a Trump ally, could be put in charge of the investigation into Cohen.
Berman, who was appointed by Trump, has recused himself from the Cohen investigation. In December, Cohen was sentenced to three years in prison after pleading guilty to numerous crimes while cooperating with federal prosecutors.
"If the Times reporting is accurate, the phone call would be evidence of what?" Fox News anchor Shepard Smith asked.
"Corrupt intent," Napolitano replied. "That is an effort to use the levers of power of the government for a corrupt purpose: to deflect an investigation into himself or his allies."
Rudy Giuliani, Trump's lawyer, told Fox News in a statement, "The President said today he had no such conversation with the Acting AG and I believe Mr. Whitaker issued a statement to the same effect. The rest of the piece is just a regurgitation of previously refuted obstruction theories. They all fail as obstruction because as Professor Dershowitz’s recent book and many other authorities make clear all of the alleged actions were within the President’s sole discretion under Article II of the US Constitution."
Napolitano called the report “dynamite, lengthy and well-documented.”
CLICK HERE TO GET THE FOX NEWS APP
“If you try to interfere with a criminal prosecution that may knock at your own door by putting your ally in there, that is clearly an attempt to obstruct justice," Napolitano said. “Where it goes from here, I don’t know. But I tell you who is reading it as we speak: Bob Mueller and team.”
RG3: Lamar Jackson will develop into a more accurate passer
OWINGS MILLS, Md. -- When it comes to Lamar Jackson, few have a better perspective than Robert Griffin III.
Part backup quarterback and part mentor, Griffin has watched nearly every pass that Jackson has thrown since Jackson joined the Baltimore Ravens as a first-round pick last season. He believes completion rate is one of the biggest areas where Jackson will improve.
"With Lamar, I equate inaccuracy with just knowing, experiencing and feeling the way the game is played at the NFL level," Griffin said. "Some guys are just inaccurate. They have a strong arm and they can’t hit a target. But Lamar can. He is an accurate quarterback. We’ve seen it in practice. We’ve seen it all throughout training camp. It’s just when the game comes and things are happening a little faster, it’s just about knowing and feeling it out. From Year 1 to Year 2, I would not expect his completion percentage to be 58 percent. But that just comes with playing."
Accuracy has long been a question mark with Jackson, who connected on less than 60 percent of his passes in each of his three seasons at Louisville and finished with a 58.2 percent completion rate in his first season in the NFL. As a rookie, Jackson delivered strikes -- like when he hit tight end Mark Andrews in stride for a 68-yard touchdown in Los Angeles against the Chargers -- but he also thew routine passes in the flat into the dirt and had other throws sail over the heads of open receivers downfield.
Some quarterbacks have gone from sporadic throwers in college to more precise passers in the NFL. Brett Favre, Carson Palmer, Matthew Stafford and Matt Ryan all completed less than 60 percent of their college passes and went on to produce completion rates over 62 percent in the NFL.
Other quarterbacks, specifically the ones that Jackson has often drawn comparisons with, have failed to fine-tune their accuracy. Michael Vick completed 56 percent of his passes at Virginia Tech and ended with a 56.2 percent completion rate in the NFL. Colin Kaepernick hit 58.2 percent of his targets at Nevada and connected on 59.8 percent of his passes with the San Francisco 49ers.
Griffin expects Jackson's efficiency to grow along with his comfort level. When Jackson sees a coverage or a blitz, he won't overthink the situation like a rookie and will instinctively know where to throw the ball because of experience.
The biggest difference this offseason will be Jackson's snaps. Last year, Jackson watched as Joe Flacco directed the first-team offense in the spring and summer. With Flacco reportedly traded to the Denver Broncos, there will be more opportunities for Jackson to work on his throwing mechanics and footwork.
"His development in the passing game will naturally come," Griffin said. "This offseason, he’s going to be QB1. He’s going to get most of the reps. He’s going to lead the offense. All of the things that he might have struggled with as a rookie, he’s naturally going to get better at them. He doesn’t have to press to say, ‘Oh my god, I have to get better at this right now.’ It’s honestly not his personality to do that anyway. Just by repetition and by work, he’s going to get better in those areas. He’ll continue to build the trust amongst players, coaches and then ultimately the fans as they see him continue to develop."
• Trade grade: Why Broncos get low marks• Broncos still need long-term QB solution• Flacco's legacy defined by playoff success• Will trade cause ripple effects in NFL?'• Flacco's smooth handoff to Jackson
Even though Jackson failed to complete more than 60 percent of his throws in college, he did trend upward in that area. His completion rates were 54.7 percent, 56.2 percent and 59.1 percent.
Last season, Jackson was one of the more accurate first-year passers. His completion percentage ranked second among rookie quarterbacks in 2018, trailing only Baker Mayfield (63.8 percent).
But in his first playoff game, Jackson misfired on seven of his first 10 passes against the Chargers. He finished 14-of-29 (48.2 percent), which was the most inaccurate performance in last season's playoffs.
"[The critics] can keep talking," Jackson said on First Take earlier this month. “When I started playing [in Week 11], I really didn’t have chemistry with the starting receivers. I was always with the rookies and stuff like that. Flacco was able to produce with the team. But I felt like each and every week, I was able to progress. But we’ll have to see next year."
The Ravens can help Jackson become a more consistent passer with their personnel moves and tweaks to the offense. Jackson can have more time to throw if Baltimore upgrades the interior of the offensive line. He can improve in the short passing game if the Ravens add a pass-catching running back or if the coaching staff calls more wide receiver screens.
It will be a team effort in the evolution of Jackson and the Ravens' passing attack.
"That’s the beauty of this place," Griffin said. "They’re invested and committed to Lamar and doing it the way that Lamar can do it. That means in Year 2, you don’t switch it and try to make him a dropback passer. You do what you do and he’ll get better at the dropback stuff and it’ll be harder to stop. I look forward to seeing what happens.
No comments:
Post a Comment