A Predictive model for hedging futures contracts to stabilize Kenyan coffee farmers' income

Abstract

The volatility of global coffee prices presents significant financial risks for Kenyan coffee farmers, leading to unpredictable incomes and economic instability. Predicting price fluctuations is crucial for stabilizing smallholder farmers’ earnings and reducing financial uncertainty. While machine learning models have been extensively used in forecasting financial and commodity prices, their application within developing economies—particularly in Kenya’s coffee sector—remains underexplored. This study aims to address this gap by developing a predictive model tailored for hedging futures contracts. The research utilizes historical auction data alongside key economic indicators to forecast coffee prices using a Long Short-Term Memory (LSTM) neural network. The model is trained and evaluated on a dataset comprising monthly coffee auction prices from 2019 to 2022, incorporating macroeconomic variables such as inflation and exchange rates. The results demonstrate that the LSTM model effectively captures price trends, achieving a Root Mean Squared Error (RMSE) of 34.94 and an R² value of 0.96. These findings indicate that the model’s predictions align closely with historical price patterns, making it viable for real-world applications. Additionally, a user-friendly interface was designed to allow farmers to select a future date and receive price forecasts. This study highlights the potential of machine learning in enhancing risk management within the agricultural sector by enabling data-driven decision-making. Implementing such a predictive system could contribute to greater income stability for Kenyan coffee farmers and serve as a scalable approach for other commodities affected by market fluctuations.