Axyon AI specialises in delivering AI-powered investment strategies through our proprietary ML platform. Daily, our systems compute predictions to rank assets based on their expected performance.
While robust quantitative methods drive these predictions, asset managers (our clients) face significant challenges in navigating and interpreting the vast amounts of data generated by our platform, and industry-standard techniques such as SHAP are not sufficient to bridge the gap between quantitative data and qualitative insights.
Integrating generative models such as Large Language Models (LLMs) with our existing platform will constitute a pivotal step forward in addressing these challenges, providing users with a more holistic view of investment opportunities.
LLMs, with their advanced natural language processing capabilities, can interpret and generate valuable insights from complex numerical and tabular data sets.
In this project, with the help of our partner AImageLab, AI research laboratory at the University of Modena and Reggio Emilia (UNIMORE), Axyon AI proposes to develop a specialized LLM on financial data and forecasts, using parameter-efficient fine-tuning or preference optimization techniques.
The resulting LLM will be able to generate reports, written in natural language, of the market trends forecasted by the Axyon AI platform, with insightful explanations of such predictions.
Such reports will be of great value in assisting asset managers in understanding the underlying factors driving predictions and thereby making data, forecasts and insights more accessible and actionable.
The successful completion of this project will have a substantial business impact, leading to
- (i) increased revenues driven by access to a broader target market and higher pricing potential, and
- (ii) reduced reliance on human resources for analysis and report drafting, as these tasks can be efficiently handled by the LLM.
Giacomo Barigazzi, AXYON AI SRL - Italy