Dynamic pricing of short-term rental properties

We have developed a dynamic pricing framework for short-term rental properties, aimed at maximising revenue and enhancing revenue management for property owners.

Assisting non-professionals with price settings and revenue management

Platforms such as Airbnb, Vrbo and Homeaway have enabled property owners worldwide to earn extra income through short-term rental accommodation. However, for non-professionals, setting the right prices and keeping rates up-to-date depending on shifts in demand can present challenges. Typically, individual owners are unlikely to have the information necessary to inform pricing decisions, nor the revenue management experience required to use the information effectively.

Monitoring the short-term vacation market

In collaboration with Wheelhouse, we have developed a dynamic pricing framework which enables property owners to save time and boost revenue from their short-term holiday rentals.

Wheelhouse is a San Francisco-based company that monitors the entire short-term vacation market globally. This means processing millions of individual bookings daily, across several platforms. In order to do this, Wheelhouse uses models developed by NR. Our models serve as the foundation for their pricing strategy, which can change based on different variables.

We have worked with Wheelhouse since 2014, when they were a startup with four team members. Over the years, we have essentially served as the research and development arm of the firm. Our team of researchers has grown steadily as the company has expanded.

Statistical modelling and machine learning

Our models are a combination of old and new methods. Since rental properties are distinct from hotel rooms and each property is unique, we grounded our framework in econometric models from the 1970s, which are designed for markets with diverse products.  Additionally, as rental properties can be booked at any point before the day in question, we made a link to the classical modelling of survival time, a standard technique in medical statistics.

These traditional models are effective tools, but were never intended to be trained on hundreds of billions of datapoints. Consequently, the models are insufficient on their own. We have therefore coupled these approaches with modern machine learning techniques in order to meet the requirements of Wheelhouse’s colossal data collection. Our semi-parametric approach is fast, robust, and enables us to adjust price recommendations for all units across the entire calendar daily.

Investment models for property management

On top of the core statistical routines for pricing, we have also developed extensive market intelligence reports, which have proven useful for driving investment models used by property management companies to determine which markets offer the best opportunities.