Prediction Method
Bid smart
You probably know the feeling: you want to bid on a house and immediately start a search for as much information as possible. You look around the street to see if many other houses are for sale, ask the real estate agent countless questions about the condition of the house and how many others are bidding. You might even grab a coffee in the neighborhood to see if it's a place you would want to live. In short, you gather data to decide if and what you want to bid.
At SlimBieden, we do pretty much the same thing. The only difference? We don't ask the real estate agent, but our data. It's much more fair and transparent because only one of those two acts without self-interest. On this page, we’d like to explain how our model works. We provide examples of how we combine macro and microdata to come up with a bid. Macrodata includes things like consumer confidence, stock markets, and housing market trends. Microdata concerns the house and its surroundings, such as the number of rooms, proximity to schools, and the condition of the property. By bundling this information, we arrive at a bid that is, on average, closest to the winning offer.
Market
The Dutch housing
market
Housing prices have been on the rise for a while, with an average increase of 1.05% per month in 2024. There are clear differences between various types of homes (apartments 2% more than detached houses) and cities (Amsterdam twice as much as Tilburg). Although the trend in the average overbid is strongly correlated with the asking price, they do not follow exactly the same pattern. Together with other economic condition variables, these time series form a reference point for the development of the overbid for a specific property, and each individual property moves to its own rhythm within the trend.
Bron: DNB
Bijgewerkt: 16 april 2025
Macroeconomic factors
When the demand for houses increases but the supply lags behind, housing prices rise, often along with the relative overbid. When interest rates increase, borrowing capacity decreases, which in turn affects what people can afford to pay for their new home. And if a financial crisis looms, making large purchases becomes more daunting: do you sell your house quickly for a good price, or do you wait it out? Even the best economists don't always agree on the effects of these kinds of relationships. SlimBieden’s model lets the data speak. We use reliable data from institutions such as De Nederlandsche Bank and the Centraal Bureau voor de Statistiek to estimate the correlations that make the overbid predictions as accurate as possible. Finally, we’ve tested these data-driven outcomes against economic intuition.
Property
Property specifications
Every house is unique, yet not that unique. At least, that's what our data shows. Our characterization of a property is so detailed that we can find enough comparable properties, even for the most eccentric homes. It’s crucial to take all these features into account, because only then can we understand why bids on properties in the same street and the same month can vary so much. The information we use is much richer than what you see in the semi-automatically filled form on the website. The underlying connections provide far more data, and we also retrieve this data for a collection of the most comparable properties, for which the winning overbid is known. Part of this information is summarized in the accompanying value report.
Demographic
Distribution
Bron: CBS
Bijgewerkt: 12 december 2024
The neighbourhood
Single-family homes located closer to a primary school are often more in demand than other properties. For apartments, the vibrant Pijp neighborhood in Amsterdam, with its trendy shops and restaurants, tends to see higher overbids compared to quieter flats in some outer districts. It might seem obvious, but for an accurate prediction, it's crucial to account for these kinds of factors. The proximity to museums, hospitals, police stations, and the amount of green space nearby often also plays a role, to varying degrees. It's precisely this large number of potentially relevant variables that makes our AI methods so powerful; they thrive on big data. By carefully fine-tuning our models, we prevent overfitting and retain only the variables that genuinely contribute to an accurate prediction of the overbid.
