What is forecasting?
Forecasting determines what will happen in the future — stock prices, weather, or booking patterns for rental properties. Revenue management has used time-series forecasting for over 40 years in airlines and hotels, examining historical trends to project future events. Its strength is that historical data is reliable; its weakness is that it’s slow to learn new trends, react to special events, and update seasonality.
Why do we forecast?
Managers use forecasts to make decisions today that shape future outcomes. Knowing that this October’s demand will be down 25% versus last year gives months of notice to adjust pricing, marketing, and promotions to recover the lost revenue.
The “what if” problem
Demand for perishable goods like rental nights requires statistical methods to estimate hypothetical scenarios. When a property sits vacant at $500 a night with a 3-night minimum, you can’t know what would have happened at $450 with a 2-night minimum. Machine-learning algorithms estimate outcomes at different price points, enabling better decisions for future bookings.
The unique challenge in short-term rentals
STR forecasting is distinct because each unit operates independently. A 300-room hotel can absorb a 10% forecast error across its rooms. An individual rental faces a binary outcome — completely booked or completely vacant — leaving no margin for error.
The solution
Most STR investment focuses on dynamic pricing, which captures 80–85% of the possible gains revenue management can achieve. Reaching the final 15–20% requires robust demand forecasting — converting those binary outcomes into price probabilities and overcoming the limits of pure time-series models.