House A - Agent listed at $1.175M 2670 sqft, 4737 sqft lot Redfin estimate - ~$1.176M https://www.redfin.com/WA/Redmond/11922-173rd-Pl-NE-98052/home/2115659 House B - Redfin listed at $1.4M 2530 sqft, 4560 sqft lot Redfin estimate - $1.40M conveniently https://www.redfin.com/WA/Redmond/17371-NE-119th-Way-98052/home/2072936 Shocking fact is that House B is smaller and literally right behind House A and both are in the same community and share same schools. Redfin estimate seems manipulated and not trustworthy. So buyers beware of this estimate shit on redfin. #redfin #microsoft #google #Facebook #amazon #tech
I sleep well at night knowing there are people actually using Redfin estimate or Zestimate
I'm not an expert, but here's the differences I see. House B is North facing, House A is West facing (superstitious) House B has more space in the backyard. House B is listed with Redfin But I've seen similar/questionable estimates as well - 3 houses in a row, identical layout, furnishing, and lot size, one was surprisingly 30% high than the other 2.
That’s prob because it’s an model that adds or subtracts to the listing price
I dont see anything wrong here. Here’s the algorithm for redfin estimate = Price listed +/- 500 So that makes sense /s
House S is listed at 2.04m. Redfin estimate at 2.10m. House S sold for 2.72m a month later, Redfin estimate caught it up in a week after closing. https://www.redfin.com/CA/Sunnyvale/1655-Kamsack-Dr-94087/home/838534?utm_source=ios_share&utm_medium=share&utm_campaign=copy_link&utm_nooverride=1&utm_content=link
1500 sq ft in Sunnyvale sold for 2.7 mil? What the fuck man
This isn't surprising. Both Zillow and Redfin's algorithm heavily weights current asking price for new listings. If you ever see their estimates, they always jump up (or down) to a new listing price whenever a house comes on the market and the price is substantially different than the pre-listing estimate. Once a house stays on the market unsold for a while, the estimate starts to trend down to reflect that. On the one hand, this is definitely dumb and easy to manipulate. On the other hand, on average, you would think that listing price is the best predictor of sale price, so it kind of makes sense.
Its kinda silly though, the estimate is supposed to offer a secondary data point that can be factored in evaluating the home. Without that you still need agents to do comp analysis, call up other listings that went pending recently to come up with a reasonable estimate. Not saying that estimate needs to be 100% accurate but close enough to give a rough idea. If estimates are literally based on the listed prices, then it might as well not show that and just show the list price
We get complaints about this all the time from customers. That team is incredibly understaffed.
Just that team? Seems understaffing is more and more prevalent these days.
Good point. We are heavily understaffed across the engineering org. Consumer Apps has it pretty bad too though seems like they're getting more people now. I don't know about your team, but we are constantly building tech debt that we never have time to address because we are so understaffed.
ML in action