top of page

Market Stadium's machine learning model for housing market forecasting

Market Stadium provides customers with property investment guides by providing forecasting models using machine learning models. The change of median housing value in the future of the census tract area which is a proxy variable of housing market growth is forecasted based on changes in the characteristics of the real estate market in the past. Determinants of median housing value such as real estate factors, sociodemographic factors, employments factors, and accessibility factors are extracted and used from the sufficient and in-depth dataset of the Market Stadium. We're delivering a forecast for the future median housing value in three years from the input data, and the CAGR(compound annual growth rate) of housing value growth in the area is calculated based on the predicted median housing value.

The Market Stadium research team tested various statistical and machine learning models in terms of data selection, model structure, and various algorithms. The latest Market Stadium forecasting model is delivering the 2024 forecasts of approximately 45,500 tract areas within 122 metropolitan areas that are selected as potential promising real estate investment areas by Market Stadium. In terms of percentage error in median housing value, our model shows approximately 90% of accuracy.

The table below shows the validation results of our forecasting model. The model for validation estimates forecasts in median housing value of the census tract area in 2021 using the data until 2018. The actual median house price for these census tracts increased by 9.08 % per year from 2018-to 2021. Meanwhile, the top 10% of census tracts expected to grow the most by our forecasting model grew by an average of 13.13% per year, and the top 20% grew by an average of 11.47%. This is 44.60% and 26.32% higher than the average growth rate of the benchmark.

Average CARG of actual median housing price change (2018-2021)

Actual median housing price change (2018-2021)

Top 10% from MS forecasting model

Top 20% from MS forecasting model




The map below shows the CAGR of actual median housing value in downtown Houston from 2018-to 2021, and the top 10% of census tracts in terms of the value growth derived from the MS forecasting model. Census tracts with darker red have a higher CAGR of actual median housing values. Try our demo and check the future housing market prediction of the market you are into.

You can see areas with black boundaries that are the top 10% areas predicted by Market Stadium's forecasting model are aligned with areas with higher growth.

156 views0 comments


bottom of page