Several key players make up the agricultural sector and its value chain, which is crucial to the national economy. Among them, the role of grain exporters stands out, with their main functions being:
- Acting as intermediaries between producers and global buyers.
- Setting prices and mitigating risks in the local market.
- Optimizing logistics and storage.
- Adding value to primary production.
- Facilitating regulatory compliance.
- Promoting new technologies and production practices, among others.
Additionally, grain exporters contribute significantly to the economy through exports (grain complexes accounted for nearly 37% of the total export value in 2023), foreign exchange generation (exceeding USD 25 billion), employment, infrastructure and transportation investment, and capital market development.
The success of the grain and oilseed origination business depends on predicting factors such as global agricultural production, global demand, consumption trends, competitors’ actions, international prices, and weather conditions. Moreover, it is crucial to predict the commercial behavior of local producers to make decisions about timing and volumes for purchases, sales, and risk management.
Viterra, one of the leading grain exporters in Argentina and the world, approached our consultancy with these concerns. For Viterra, understanding and quantifying the factors that determine the selling decisions of agricultural producers in different regions is vital to the success of their business and profitability.
Thus, being able to predict the volume of sales positions at different times of the year—before harvest (forward sales), during harvest (direct settlement), or post-harvest (spot sales)—is crucial for developing marketing strategies and optimizing decisions. It should be noted that, in the grain and oilseed market, the moment when the producer takes a sales position (Forward, Contract) does not necessarily coincide, in most cases, with the settlement moment.
In response to this demand, at Simpleza, and in collaboration with Viterra, we developed the “Farmer Selling” model. This tool applies state-of-the-art data science to integrate and model data from various sources and answer the questions: “When will the producer decide to sell their production during the year? What will be the volume to sell?”
“Farmer Selling” allows forecasting, based on historical data, how the selling intention will be during the next campaign and the relative importance of each determining variable. Specifically, the tool combines two models: one at an annual scale and another at a monthly scale. The annual model aims to forecast the total sales volume of each campaign based on the total produced and some covariables. Then, the monthly model forecasts the evolution of sales within the commercial campaign.
In addition to specific predictions, the model allows quantifying the uncertainty associated with the predictions and proposing different scenarios. This is achieved by characterizing campaigns according to the relative sales evolution using clustering analysis and simulating future events. This way, the user can evaluate “what if…” scenarios. All of this is presented through a user-friendly graphical interface or dashboard, where data series and key industry indicators are summarized.
We built the tool over 12 months through four stages: generating a historical database with public and private data (evolution of sales positions by product and region, and the variables that define their behavior); econometric analysis of the time series; construction and calibration of predictive models; and finally, result visualization. The tool’s updating and maintenance are continuous.
