This episode is a deep dive into the paper by Jorge Faleiro defining Sigma, an open and domain-specific language designed to allow large-scale collaboration in economics, especially within financial trading and modeling. The paper is publicly available in the author’s arXiv.org page.

This audio podcast is produced by the author using generative artificial intelligence technology, and the publicly available version of the Sigma paper as the main source.

This episode explores the application of incremental learning, specifically using a Stochastic Gradient Descent (SGD) model, for building a robot trading system. It contrasts this approach with batch-supervised learning and reinforcement learning, arguing that incremental learning is more suitable for the non-stationary and rapidly changing nature of financial data.

The article presents a simplified demonstration of how such a system can be built and trained, while also cautioning against over-optimistic expectations of guaranteed profit.

This audio podcast is produced by the author using generative artificial intelligence technology, and the article describing the end-to-end process of developing the nowcasting model is used as the main source.