Name: Alessandro Taglietti
Title: Data Scientist & Portfolio Manager
Organization: Denarius Conseils & Gestion SA
Denarius Conseils & Gestion is an independent Investment Management firm based in Geneva, focusing on Portfolio Management and Multi-Family-Office services. We focus on active Investment strategies, with a strong expertise in Global Macro trading across multiple markets.
- Best MLOps Use Case
- Best AI Democratization Program
- Best ROI Story
We are an independent Investment Management firm. Our investment style involves highly active trading across multiple markets (stocks, commodities, interest rates, forex, derivatives such as options and futures...). A key challenge for our portfolio managers was to simultaneously analyze all the markets we trade, detecting opportunities, patterns, and trends to generate investment ideas and define our macro view of the markets.
Without AI, we were forced to manually analyze each individual security, thus limiting the range of opportunities we could act on as well as our ability to explore interconnections between different markets. Furthermore, we also felt the need to have a centralized and collaborative environment for R&D, where we could test new quantitative models, validate them with data, and easily put them into production in order to make our investment process more systematic and disciplined.
We chose Dataiku because of my previous experience with the tool as a Consultant. I saw Dataiku as a perfect fit for Denarius because of the way it successfully combines ETL functionalities with an R&D environment where we can prototype and engineer models in Python. I am the main user and administrator of the tool; my primary role is that of Portfolio Manager and Trader, but I also act as the team's lead Data Scientist.
As an individual user, I have been able to accomplish all of my goals with the Community Edition, which I think exemplifies well the power of the tool. With Dataiku, I was able to develop an end-to-end platform performing the following tasks:
- Ingest market data from the Interactive Brokers API, data APIs of other vendors, and custom flat files.
- Structure a market data warehouse (our entire DW, based on PostGreSQL, is orchestrated and managed using Dataiku as the key interface and ETL tool).
- Perform ETL to calculate a large number of trading indicators on each security.
- Identify trading signals.
- Analyze new datasets.
- Run in production a quantitative model for systematic trend-following strategies (a key hedge fund strategy where long-term and short-term trends are estimated on multiple markets, with traders positioning themselves in the direction of the trend to ride with the flow).
- Produce PDF and Excel reports for Portfolio Managers displaying the most promising trading opportunities (e.g., key market trend direction and strength, oversold-overbought securities, price patterns with predictive power).
- Backtest investment/trading strategies through simulations on historical data (this module is still a work in progress).
- Integrate with a number of internal APIs, acting as the key nexus in a microservice architecture (e.g., microservice to turn HTML strings into PDF reports, microservice to send PDF reports to Portfolio Managers by email).
Thanks to Dataiku, I was able to develop a complex end-to-end platform, without the need to hire a dedicated software development team. Dataiku's flexibility to function in a fluid and rapidly evolving environment has also proved essential, as I had to experiment with multiple technology architectures:
- I started the project on the Cloud but was able to quickly move it to an on-premise server with very limited effort.
- I started the DW on MySQL, but subsequently moved it to PostGRESql in only three days of work.
Dataiku has had a major impact on our ability to scan multiple markets and make our investment process more systematic. Every night, Dataiku analyzes more than 3000 securities and markets for signals, patterns, and trends, producing a report for our traders indicating trends in key markets as well as highlighting securities that are attractive to trade. This has allowed us to have:
- A fully automated trend-following strategy that we can integrate into our portfolios.
- A rapid overview of market opportunities before the Opening.
- The possibility to spot trading signals in stocks/sectors that, before we had our algorithmic tools, we would not consider a core area for investments.
We have noticed a great improvement in our ability to invest thanks to the AI tools we developed using Dataiku:
- We have more signals. As a result, we can trade more and have greater diversification, with a sharp improvement in terms of risk management (we trade more often and across more markets; therefore, we have less risk on each trade).
- We are more disciplined, systematic, and consistent (we now have some strategies in the trading book that are purely model-based, and, in our discretionary strategies, we have improved risk management by never taking positions that go in the opposite direction compared to what the models are telling us).
- We have the optionality to develop and test new strategies quickly.
Business Area: Analytics
Use Case Stage: In Production
- We have increased by a factor of three the number of trades we perform.
- We have reduced our Data & Analytics costs by 50%, as we cut software tools that were not needed once we could fully exploit Dataiku.
- We have a very flexible software platform for Trading Intelligence; we are able to put a new strategy into production in one day.
Value Brought by Dataiku:
- Enhanced tech stack efficiency Integration of ETL, R&D / AI, and reporting in a single tool.
- Flexibility to connect across data sources in a hybrid environment.
- Marketing: screenshots of Dataiku flows were useful in presenting clients our capabilities.
- Increase revenue
- Reduce risk
- Save time
- Increase trust
- Other - nice visuals for marketing
Value Range: Thousands of $