Structured finance is a data-intensive field, with success depending on the ability to analyse swaths of information. But for all its importance, the data itself is often messy and difficult to work with. Plus, stakeholders expect insights even when the data isn’t really sufficient.

Once, a static file and a few formulas might have been enough. But now, every servicer or originator has their own way of reporting, with inconsistent formats, missing fields, and weird quirks that take hours (or days) to clean up. Most of us end up with three options: clean everything manually, build ETLs in-house, or pay for some expensive data solution that doesn’t always give you exactly what you need.

The path forward begins with one fundamental change: open access to clean and better-structured data.

What is Deeploans?

Tools like Lotus 1-2-3 were revolutionary in their time, and Excel has since become an indispensable upgrade. But as asset-level datasets grow in scale and complexity, even our beloved Excel struggles to keep up.

Deeploans is an open-source framework designed to simplify the processing and analysis of loan-level data. 

At its core, it is a collection of ETL (Extract, Transform, Load) pipelines tailored to the needs of structured finance professionals and innovators. These pipelines improve the process of cleaning, structuring, and preparing loan-level datasets for analysis.

Deeploans currently supports ETL pipelines for several key asset classes including:

• Auto Loans: Recent trends, such as the rise of electric vehicles and shifting consumer preferences, have added new layers of variability. Changing repayment and refinancing behaviour tied to EVs, for example, means that traditional models may not apply well to EVs.
• Consumer Loans: Consumer loans make up a diverse and dynamic segment of structured finance. Predictions in this segment are heavily reliant on behaviour patterns, and the diversity of loan types adds to the challenge, but with the rise of Fintech lending, the volume and granularity of this data are becoming almost unmanageable. 
• Corporate Loans: Small and medium-sized enterprises drive much of the economy, but their loans can be tricky to analyse due to their varying risk profiles. 
• Residential Mortgages: This pipeline transforms raw residential mortgage data into a high-quality, organized format ready for building advanced solutions that enrich ESG and other key variables. 

Each of these pipelines is designed to address the unique challenges of its respective asset class, improving the structure of data and preparing it for machine learning models or dashboards in tools like Looker and PowerBI.

Why Open Source?

For this sector to innovate at any meaningful speed, open-source collaboration is essential. While innovation is possible within closed systems, the pace is too slow to meet the demands of an increasingly complex landscape.

Research has shown that open-source tools can greatly speed up development and improve the overall quality of the final solutions. 

The sector needs a shared foundation to innovate independently, with less costs and less strings attached. Open sourcing Deeploans is our response to this challenge. 

There are two key things we aim for:

1. Democratising data access

A hallmark of innovation is expanding access to all. PCs didn’t just evolve—they democratised access to computing power. High-level languages, e.g. Python, democratised access to software development, with no-code platforms and Gen AI models continuing this trend today.

When access improves, solutions flourish. 

Structured finance professionals should be able to access tools to process and prepare data themselves, avoiding reliance on intermediaries and establishing more control over data pipelines.

2. Making AI-Powered finance a reality 

A current pet peeve across industries is that AI tools are impressive but they cannot do anything useful without the right data foundation.

For innovators and professionals to move beyond expensive data providers and generic AI models, they need to be able to:

• Access and process granular data efficiently.
• Train or fine-tune AI models tailored to their unique markets, asset classes, and risk profiles.

With deeploans now open, professionals can experiment, prototype, and deploy models that genuinely add value.

How to get involved

Structured Finance Professionals

Leverage Deeploans to streamline your workflows, reduce your reliance on third parties, and start building models tailored to your needs. 

For Developers and Contributors

Contribute to the Deeploans framework by:

• Onboarding new asset classes.
• Improving data quality rules and standards.
• Refining tools for advanced analysis and reporting.

Getting Started

Explore Deeploans on GitHub today, access documentation, and join the community!

https://github.com/Algoritmica-ai/deeploans

For any comments or questions, drop me a message at luca.borella@algoritmica.ai