- Softened predictions point toward a more balanced market: 65.3% of current on-market listings are predicted to sell at a discount to their original list prices in Q3 2019, flat from the Q2 2019 prediction of 65.3%.
- The average predicted discount buyers can expect to see is 3.1%, but 30.1% of homes are predicted to sell at a discount of 5% or more.
- Top 10 markets where homes are expected to sell at a discount include major cities like Miami (#1), Chicago (#2) and New York (#8), with more high-priced inventory and frequent domestic migration to secondary cities.
- These markets were also among the top 10 cities for deals on homes sold in Q2 2019, when 61.5% of homes sold below their original list prices – a 5.2% increase year-over-year.
Knock, on a mission to make trading-in a home as easy as trading-in a car, today released the results of the Q3 2019 National Knock Deals Forecast, predicting which of 45 of the largest U.S. Metropolitan Statistical Areas (MSAs) will have the highest percentage of homes that sell at a discount to their original list prices, or at a “Deal,” among current listings. Knock also analyzed actual deals among Q2 2019 and Q2 2018 home sales to determine market trends over time.
“In line with the Q2 2019 Knock Deals Forecast, housing commentary during the Spring Homebuying Season pointed toward a shift to more of a buyer’s market,” said Sean Black, Co-Founder and CEO of Knock. “But as the market slows heading into the second half of the summer when families want to be settling in for the new school year, less inventory will mean increased competition for buyers. With our Forecast, Knock aims to provide these consumers with transparency into where homes are most frequently overpriced so that they can negotiate the best deal.”
The National Knock Deals Forecast is developed using Knock’s proprietary machine learning-driven algorithms that predict real time outcomes for listings currently on the market. The algorithms take into account over 200 characteristic, historical, seasonal and behavioral data points to determine outcomes like final sale prices for homes it predicts are likely to sell. In June 2019, 50% of U.S. homes sold within 1.7% of the Knock predicted final sale price, and 70% sold within 2.8% of the predicted final sale price.
Graph 1: Rate of homes selling below, at, above original list prices
Moving Toward a More Balanced Market: Home Prices
Knock forecasts that 65.4% of current on-market listings in 45 of the largest U.S. MSAs will sell below their original list prices. This number is relatively unchanged from the Q2 2019 prediction, reflecting the transition out of the busy Spring Homebuying season and that inventory is at near historic lows with a 4.3 month supply, according to the National Association of Realtors.
Knock did see a 5.5% year-over-year increase in the rate of deals among actual Q2 2019 home sales. Knock also saw a significant year-over-year decrease in the rate of homes selling above their original list prices, down to 23.5% in Q2 2019 from 28.3% at the same time last year – this number is forecasted at 19.5% for current on-market listings.
Graph 2: Home sale prices by range of variation from original list prices
“While we’re not necessarily seeing the same jump in homes selling below original list prices that we forecasted going into Q2, the good news is that there has been a steady drop in homes selling above their original list prices,” said Jamie Glenn, Co-Founder and COO at Knock. “Buyers are being more cautious, which is forcing sellers to price their homes more realistically. Of course, lower prices combined with seasonality may lead to less inventory, but I think we’ll see buyers being less hasty to pay over market value regardless of availability.”
Among homes predicted to sell at a deal, 30.1% are predicted to sell for 5% or more below original list prices, compared to just 5.2% that are predicted to sell for 5% or more above original list prices. The average predicted discount to home prices across the markets analyzed is 3.13%. This number wasn’t much lower in Q2 2019, at 2.7%, when it was also only up 0.7% year-over-year. However looking at more competitive markets illustrates the shift toward greater balance. For example San Francisco, forecasted to be the worst market for deals heading into Q3 2019, is predicted to have an average premium to original list prices of just 0.7%. On average homes sold for 3.1% more than original list prices in Q2 2019, down from 8.9% in Q2 2018.
San Francisco’s steady softening is a reflection of larger market trends. Comparatively, the greatest instances of increased competition seem to be in markets that have experienced significant economic events. For instance, the Washington, DC area, which is still experiencing the aftermath of Amazon’s decision to make it the company’s only official second headquarters, is the second worst market for deals on our Q3 Forecast. Furthermore, it is the only market in the bottom 10 for Q2 2019 sales to see a year-over-year decrease in the rate of deals, down 4.2% from Q2 2018. On the other hand New York, which Amazon ultimately decided against as a headquarter location, saw a 3.4% year-over-year increase in the rate of deals, and is ranked number eight on the Q3 2019 Forecast. New York is also a prime example of a market that has been seeing extensive domestic migration to secondary cities in recent years.
Moving Toward a More Balanced Market: Locations
Seven of the top 10 markets on the Q2 2019 Knock Deals Forecast were in the south, including four markets in Florida. While Miami continues to be the number one forecasted market for deals headed into Q3 2019, the top 10 are evenly split between MSAs in the North and in the South of the country, with New York and Pittsburgh moving up the list.
Table 1: Top 10 markets with the most current listings predicted to sell below original list prices (listings added within the past 16 weeks)
|Forecast Ranking||Metropolitan Statistical Area (MSA)||Percent of homes predicted to sell below original list price||Percent of homes predicted to sell at original list price||Percent of homes predicted to sell above original list price||Average predicted discount/ (premium) to original list price|
|1||Miami-Fort Lauderdale-West Palm Beach, FL||84.42%||7.31%||8.27%||5.57%|
|3||Hartford-West Hartford-East Hartford, CT||76.45%||7.57%||15.98%||4.10%|
|4||Houston-The Woodlands-Sugar Land, TX||75.68%||11.55%||12.77%||4.01%|
|5||New Orleans-Metairie, LA||75.51%||14.16%||10.33%||4.68%|
|6||Tampa-St. Petersburg-Clearwater, FL||73.20%||15.28%||11.51%||3.97%|
|8||New York-Newark-Jersey City, NY-NJ-PA||72.78%||8.65%||18.56%||3.91%|
|10||St. Louis, MO-IL||71.34%||13.94%||14.71%||4.85%|
Previous editions of the Knock Deals Forecast have found a correlation between rising prices and a higher rate of deals. Many home price reports that create buzz around rising prices are dated by one to two months, and can encourage sellers to list at prices higher than market value, particularly taking factors like seasonality into account.
For instance, the number nine predicted market for deals is currently Tampa, which is also a top three market on the latest S&P CoreLogic Case-Shiller Home Price Indices due to its year-over-year price increase of 5.6% – as of April. However Tampa saw the second highest rate of deals in Q2 2019, with homes selling at an average discount of 4.7%. This number rose to an average 8.7% for the 29.1% of homes that took at least two months to sell. On a home listed at $300,000 that could mean a savings of over $25,000.
Another example of a market in a different U.S. region that is also frequently in the news is Las Vegas, which just missed the Q3 top 10 at number 11, but was the number eight best market for deals among Q2 2019 home sales. Las Vegas has consistently been a top ranked market by Case-Shiller, most recently at number one with a 7.1% year-over-year price increase as of April. However 71.8% of homes that sold in Q2 2019 sold for below original list prices, up 17.0% year-over-year. 24% of Q2 2019 sales in Las Vegas were on the market for over 60 days, and 91.7% of those homes sold at a discount.
“Between decelerating home price increases, mixed reports on inventory and sales, and lower mortgage rates, conversations about the housing market have been somewhat uncertain over the past several months,” said Paul Habibi, Economic Advisor to Knock and Lecturer of Real Estate at UCLA Anderson School of Management. “However this ping-ponging could just be reflective of the market finding its balance, as further evidenced by the reduced instances of homes selling above list prices and geographic variation in the rate of deals reported in this latest National Knock Deals Forecast.”
Table 2: Top 10 markets with the most homes sold below original list prices in Q2 2019
|Forecast Ranking||Metropolitan Statistical Area (MSA)||Percent of homes that sold below original list price||Percent of homes that sold at original list price||Percent of homes that sold above original list price||Average discount/ (premium) to original list price|
|1||Miami-Fort Lauderdale-West Palm Beach, FL||86.48%||7.55%||5.96%||6.98%|
|2||Tampa-St. Petersburg-Clearwater, FL||74.09%||14.33%||11.58%||4.66%|
|6||New Orleans-Metairie, LA||72.23%||15.02%||12.74%||4.81%|
|7||Houston-The Woodlands-Sugar Land, TX||71.97%||13.76%||14.27%||3.98%|
|8||Las Vegas-Henderson-Paradise, NV||71.76%||15.13%||13.11%||3.10%|
|9||New York-Newark-Jersey City, NY-NJ-PA||70.07%||9.05%||20.88%||3.62%|
|10||Hartford-West Hartford-East Hartford, CT||67.52%||11.69%||20.79%||3.73%|
Knock is building an end-to-end home buying and selling marketplace that will provide consumers with more transparency into and control over the home buying and selling process. At the core of this platform are Knock’s proprietary algorithms, which help determine the market value and time to sell of our Home Trade-In customers’ homes. These machine learning-driven models apply insights previously held exclusively by real estate agents to data that can be accessed by the average home buyer and seller. With the quarterly National Knock Deals Forecasts, Knock processes data on the entire U.S. market through these algorithms with the goal of empowering buyers with information about the realities of the market and encouraging sellers to price their homes to true market value, to ultimately increase overall market fluidity.
Q3 Predictions: Interactive Map of the Findings
Q2 Sales: Interactive Map of the Findings
Knock (www.knock.com) is the first online Home Trade-in platform, a revolutionary new approach to home buying and selling that makes it as easy to trade-in your home as it is to trade-in your car. Launched by founding team members of Trulia.com, the company uses data science to price homes accurately, technology to sell them efficiently and a dedicated team of Licensed Experts to guide consumers through every step of the process. Knock has raised over $600 million in debt and equity, closing its Series B in January 2019, from top tier investors including RRE Ventures, Foundry Group, Redpoint, Greycroft, Corazon Capital, Correlation Ventures, Great Oaks Venture Capital and FJ Labs. The company has offices in New York, San Francisco, Atlanta, Charlotte, Raleigh-Durham, Dallas, Fort Worth, Denver and Phoenix, with several more on the way.
Knock trains a suite of machine learning models on historical real estate data going back three years. The models, trained on 200 features across the top U.S. Metropolitan Statistical Areas by population size, are able to predict listing outcomes such as the likelihood in selling, the selling price, when various price drops will occur and by how much, and how long it will take to sell. These models take into account seasonal trends, long-term market trends, and hyperlocal information, inclusive of the real life pricing activities of individual real estate agents, to make their predictions.
To understand what is happening at the market level, the probable outcomes for each listing are aggregated over all the listings in a market to produce what is likely to happen for the entire market. In order to predict the proportion of upcoming deals to expect in each market, Knock predicts the probability of a listing selling at various levels below the original list price. We also adjust these aggregated probabilities by the likelihood that the listings will sell. These probabilities are then aggregated within each market resulting in a market-wide prediction. Additional factors:
- Time Frame: Predictions are based on 583,949 active listings added to the market in the past 16 weeks as of June 30, 2019.
- Market size: 45 of the top 50 largest U.S. MSAs were analyzed; not included due to data sourcing and therefore accuracy challenges are Milwaukee, WI, Richmond, VA Seattle, WA, Sacramento, CA, and Virginia Beach, VA.
- Likelihood of Selling: Knock’s algorithms predict the likelihood of a listing selling during its current listing period as opposed to being withdrawn/removed/closed. If a listing is 50%+ likely to not to sell, its impact on aggregated price estimates is discounted by half.
- Outlier filtering: In order to remove price changes that are likely due to typos or unusually priced listings, we remove listings from our calculations where the price has increased or decreased by two-fold. This removes about 0.1% of the listings.
- Home type: Study includes single family homes (attached, detached, condo); excludes land sales, mobile homes, income properties, multi-family properties, foreclosures, short-sales, new construction.
- Price band: Study includes homes in the $50,000-$5,000,000 range.
- Days on market: Knock measures these as the difference between list date and pending date.
Additionally, Knock analyzed data on 525,826 home sales from April 1, 2019 through June 30, 2019, as well as 652,819 home sales during the same period last year, in the same 45 MSAs across various time frames, comparing data from their original listing to data from their sale. These data include the original list price, final sale price and days on market. e.g. year to date, by month, etc., to determine trends in homes selling below original list prices, and how they compare to the Knock Deals Forecast predictions.
Data for both the predictive and historical analyses sourced from ATTOM Data Solutions.
Knock Deals Forecasting Model 2.0: Updates Methodology
Starting with this Q3 2019 National Knock Deals Forecast, the Knock model treats selling at exact list price as its own type of outcome. The model previously treated selling at list price as being both 0%-1% above list and 0%-1% below list. If before the model thought it was 30% likely to sell for exactly list, that likelihood would get split between above and below list 0%-1%. Therefore this number was split approximately evenly between homes that sold below original list prices and homes that sold above original list prices.
Additionally, the model has been updated to more accurately make actual predictions on the variation ranges from original list prices, e.g. 0%-5% below original list prices. Knock previously applied an interpolation to the predicted to sold ratio categories, but now directly predicts the original list price outcomes.
 2.0 version of machine learning-driven Knock Deals Forecast algorithm more accurately predicts homes expected to sell at exactly list price, bringing predicted deals more in line with forecasted market patterns. Q2 prediction adjusted using new model provided, showing greater accuracy to the actual Q2 sales numbers.