Use Case Financial
Drawdown management in portfolios (Factor investing & ML)
The use case focuses on managing drawdown risk in factor-based investment portfolios. Since the development of the CAPM model and the Fama-French factor model, tools have been created to test various fundamental and technical variables in order to generate market returns without adding additional risk. This use case explores how artificial intelligence techniques can help manage this risk.
Challenges
The project faces challenges in obtaining real-time data, minimizing over-optimization and false positives/negatives, monitoring the macroeconomic environment for model longevity, and managing operational risks.
Real-time data acquisition
Obtaining real-time or minimally delayed data.
Minimizing over-optimization risks
Reducing the risks of over-optimization and false positives or negatives.
Macroeconomic monitoring for model longevity
Monitoring the macroeconomic environment to determine the lifespan of the models.
Operational risks
Solution
Factor models typically rely on linear models, which, while capable of creating models that generate higher returns than indices, fail to capture certain anomalies in companies, such as hidden risks on the balance sheet or unused data in the model.
The goal is to eliminate positions using artificial intelligence models, as they allow us to process massive amounts of data and non-linear techniques to identify potential risks. By reducing the number of assets that lose value while remaining in the portfolio, overall performance will improve, as well as key metrics like the Sharpe ratio and drawdown.
Tech stack
Results
Reduction of permanent drawdowns in the portfolio.
