Introducing the Fregnan Engine
An advanced, machine-learning driven forecasting and modelling tool for unbiased, high-quality equity research.
Integration into your existing workflow
The Fregnan Engine connects to you existing data provider, providing seamless integration to your existing workflows.
Please contact us to discuss your specific setup and data requirements.
The Fregnan Engine creates forecasts using best practice data science.
- Prior to any forecasting or modelling, the Fregnan Engine transforms, cleans and scales company fundamentals using custom data pipelines.
Machine-driven regression analysis
- Connections are formed between individual company fundamentals and to other relevant data using machine learning. Feature selection is constrained by our proprietary ontology, which is designed by financial experts and identifies the possible paths of cash flow. For most companies, this whole process is accomplished in less than a minute.
Advanced time series forecasting
- Forecasts are generated using other machine techniques, which handle both trend and seasonality. The hyper-parameter space of each machine learning model is automatically determined to find an optimal configuration.
The Fregnan Engine is built around a modelling platform that allows users to inspect and rebuild the machine-generated financial models, tailing them to specific scenarios.
- Individual company reports and their financial models are organised into portfolios, which in turn are differentiated into "user portfolios", "sector portfolios" and "global portfolios". For ease of comparison and benchmarking, Fregnan's standard portfolios cover all major global indices.
Portfolio-wide modelling assumptions
- Which companies are most exposed from the possibility of GDP dropping by 10%, or of oil prices increasing to $100/bbl? Which companies are the most exposed to significant changes in USD/GBP? If the revenue of a sector falls on average by 20%, then what are the corresponding changes in profit to each constituent company? These questions and questions like these can be answered by making portfolio-wide changes to financial model assumptions. Hundreds of individual, bespoke company models can be built concurrently, and then clear comparisons can be made between them.
- Reports and their financial models can be rebuilt against a given point in time, allowing users to validate their modelling assumptions. Only fundamental, market and alternative data that would have been available at that point in time is used in model creation.
Machine-generated models of the Fregnan Engine are transformed into Excel formulae.
The result is a fully connected financial model comparable to what a human analyst would normally create from scratch.