On June 5th, some members of TradeHeader enjoyed a hackathon organized by Algoritmica in Barcelona. We tackled several challenges related to structured finance with AI and spent the entire day participating in various activities. Here's the chronicle of the day.
When we got to Aticco, we greeted other participants and the event organizers. We grabbed a cup of coffee to get the day going. It happened to be Gina's birthday, so she surprised us all with some delicious croissants, following the tradition of bringing in treats for birthdays.
Luca Borella, the founder of Algorítmica, kicked off the day by introducing us to the hackathon. He delved into the essence of the event, stressing the role of finance in our everyday lives and focusing on asset backed securities, the central theme of the event.
Following Luca's introduction, Maurizio Pillitu, CTO of FINOS, took the stage to discuss their project and provide updates on the training courses they're developing. These courses are designed to be invaluable resources for improving our skills in the open-source environment.
Then, Marc Gratacós stepped up to introduce us to the CDM model. He emphasized its significance in standardizing financial products and highlighted its ability to cover all associated processes, offering a comprehensive approach to the financial industry.
Following these informative presentations, we had the opportunity to collaborate with our teams and complete the solutions we had been thoroughly implementing over the previous weeks. This time allowed us to finalize the code and distribute it according to the requested channels. Working closely with our teammates, we coordinated our efforts to reach our goals and ensure that the delivery was done on time.
Now, it was Martina Guzmán's turn to guide us through the Best 3 Minutes approach, designed by David Becket. This method underscores the significance of effectively communicating our ideas within a concise timeframe. Martina stressed how pitches serve as vital tools for forging new connections, engaging in thrilling projects, and showcasing our ideas. The ultimate aim? To ensure that our pitches are not only heard but also comprehended and retained. Through engaging activities, Martina helped us apply these fundamental principles in real-time.
When lunchtime arrived, we were pleasantly surprised with boxes filled with a variety of meals from the nearby restaurant, Cacho. There were even some extra boxes left over, leading to a stroke of luck for a few of us who enjoyed free dinners for the remainder of the week. As a sweet gesture to celebrate Gina's birthday, a special dessert was served, and she generously shared it with everyone. It was absolutely delicious!
After lunch, we got back to work on our challenge. We focused on improving our presentations based on Martina's advice about the 3-minute pitch. We wanted to make sure our ideas were clear and impactful. As the deadline approached, we hurried to gather all our materials – code, presentations, and demos. We wanted to be ready to showcase our work to the judges before time ran out.
After the breakout session, we had a conference by Sam Griek, Founder of EaglEyes AI, where he featured an interactive session where a lawyer posed real-time questions to a developer demonstrating a cutting-edge legal tech tool. It was very interesting and original.
The moment arrived to present the results of our challenge. Time was short, and although we wished we could have explored our findings more thoroughly, we concisely presented our challenges to fellow participants and a jury.
Find here a summary of the challenges tackled by both teams and the proposed solutions:
Tradeheader-1: The financial industry currently faces significant challenges in managing legal documentation due to the sheer volume and complexity of documents. Manual processing is time-consuming and error-prone, leading to inefficiencies and inconsistencies.
To address these issues, our project involved using Retrieval-Augmented Generation (RAG) to extract specific legal agreements from extensive legal documents and convert them into a Common Domain Model (CDM) format. The system identifies and extracts various agreements such as master agreements, master confirmations, and contractual supplements from lengthy legal prospects. It then transforms these extracted codes into the CDM format, appending them to the contract details of the trade. This solution offers several key benefits. It significantly streamlines the document handling process, making it much more efficient by reducing the time and effort required. Additionally, it improves accuracy by leveraging advanced natural language processing techniques to minimize errors. Finally, converting these prospectuses into the CDM format ensures consistency and better integration with existing systems.
Overall, this project demonstrates the potential for advanced technologies to revolutionize the management of legal documents, offering substantial improvements in efficiency, accuracy, and standardization.
Tradeheader-unicorns: Existing tools are not able to efficiently analyze the high volume of reported data and detect anomalies. The solution targets ABS product reporting compliance through having a standardized representation of the impacted transactions which is essential for integrating the data with the appropriate tools to devise a solution.
The Common Domain Model (CDM) was used as a standard representation of the financial instruments. Firstly, the object model was extended supporting the representation of the targeted products at the required taxonomy qualification granularity level. Secondly, the functional model was extended creating a qualification interface specific for ABS products.
Finally, the chosen implementation for the code generator was Java which fully covers all DSL functionalities by default.
Using the structured transactions and the qualification results as input for the first IA model, the eligibility of the transactions was determined, producing a list of jurisdictions for report submission. Once the relevant reports were externally generated, a second IA model used one of them, along with the original CDM transaction representation, and returned the report updated and compliant with the latest regulations. This process was implemented in a business-friendly environment where live transactions were monitored, and their updated reports were displayed, highlighting differences and enabling the user to make corrections or modifications.
The jury asked each group multiple questions, exploring the intricacies of our projects. Despite the time constraints, the pitches were captivating and insightful. With our presentations concluded, the jury withdrew to deliberate and select the winner.
While the judges were making their decision, Nuria Manilas, who manages the Barcelona Finance Hub, came over to talk to us about it. She gave a really interesting talk, asking why so many people in finance are drawn to Barcelona, even though it's not usually seen as a big financial city like London, Frankfurt, or New York. Nuria said Barcelona's strong points are its education, programs for startups, and focus on new ideas. She promised that the hub would keep supporting these things to make sure Barcelona stays a place full of creativity and chances in finance.
The jury announced the winners of the hackathon, and it was the team "Tradeheader-unicorns". Congratulations to all of them! They presented a fantastic end-to-end solution for Challenge 3, and we're excited to see if it can be further developed in the future!
As the day's schedule came to an end, we gathered for a casual networking session, enjoying refreshing drinks and lively conversations. Against the backdrop of the setting sun, we formed friendships and strengthened connections, marking the end of a fulfilling day of collaboration and innovation.
Personally, I thoroughly enjoyed the experience. On one hand, I gained new knowledge in both economic and technical areas, and I attended some highly engaging conferences. On the other hand, collaborating with colleagues whom I don't typically work with on a daily basis was very enjoyable, and we had a great day together!
Laura, graduated with a degree in Economics and a master's degree in Finance, joined TradeHeader as a consultant. Being part of the payments team, she specializes in the ISO 20022 Standard, analyzing payment flows and messages. Laura also works in risk reporting.
To be informed of more content like this, subscribe to our blog!