Hybrid Prediction Framework of Company Growth Based Scaling Law and Neural Network

Predicting company growth carries significant implications for strategic adjustment, operational decision-making, risk assessment, and loan eligibility reviews. Traditional attempts to model company growth either overly focus on theory, neglecting practical forecasting, or rely on time-series forecasting techniques without incorporating the inherent mechanisms of corporate growth. A recent study established a first-principles growth model for enterprises based on scaling laws, providing predictions superior to baseline methods regarding average business growth. In this paper, we propose a hybrid prediction framework for company growth, blending the growth equation of enterprises with time-series forecasting technology. Specifically, we used a neural network model as a prediction method to make our framework more extensible. This model captures both the intrinsic growth mechanisms of enterprises and the fluctuations in the external environment, demonstrating superior predictive performance compared to methods using time-series forecasting technology alone. Moreover, we found that the advantages of our framework are more prominent for long-range prediction tasks.