Financial forecasting of sales is vital for companies to make appropriate and informative decisions. Unlike point forecasting, probabilistic forecasting provides realistic uncertainty estimates, which result in more informed decisions. Additionally, interpretable solutions allow for understanding the rationale behind a model’s decision. Stochastic differential equations (SDE)-based models offer probabilistic and more interpretable solutions. This thesis presents an SDE model capable of providing probabilistic forecasting for the weekly sales of 45 Walmart stores. A neural network implementation is used to learn the parameters of the SDE. Extracted features, such as the month and the day of the weekly sale, are also investigated to further improve the SDE parameters estimation. More context of the sales in the previous and recent weeks to a specific date is additionally provided to the model in the form of time-delay embedding to further boost the estimation. Using the estimated SDE parameters, multiple trajectories of the time series of each Walmart store are simulated from which the confidence intervals are computed. Different feature scaling and transformations are experimented with and applied to the data. Two baseline models, a prophet model and a SDE-based one were additionally built to which all the experimented models were compared to. Two models were found to be superior to the rest, surpassing the SDE-based baseline and matching the prophet-based one. The first model, which used a 4-week time-delay embedding of the weekly sales, produced the most accurate results with realistic uncertainty estimates. However, the mean of the confidence intervals was roughly constant or featureless. The second model, which utilized a standard scaled 52-week time-delay embedding of the weekly sales, grasped the structures of the original time series extremely well and provided accurate predictions with realistic uncertainty estimates. Yet, the second model was less accurate than the first and generates wider confidence intervals.