Joint Research projects

Working together on this collaboration between UT and ING has personally highlighted the incredible outcomes of blending academic research with industry expertise, especially through our successful joint projects in AI for finance.

Research Projects

This collaboration aims to explore and develop advanced AI applications in the fields of data handling, risk management, and business operations within the finance sector

Main Research Topics:

1. Applications of synthetic data generation for Finance
   1. Testing trading strategies robustness, comparing portfolio construction methods, estimating the risk of a portfolio or a strategy, alternative pricing and hedging of options and other derivatives, generating trading signals, detecting anomalies in fundamental data, with a particular focus on using generative adversarial networks.
   2. Synthetic generator for (arbitrage-free) volatility surfaces
   3. Synthetic data generators that are differentially private, i.e. do not leak information about the original data, and still have enough features
2. Early warning systems for credit risk. Despite many years of research into credit risk, large and unexpected losses still happen frequently. Research on the causal relationships between market prices and external ratings as well as applying machine learning techniques and using new datasets for predicting downgrading and default  of loans is beneficial to reduce credit losses.
3. Research on risk management related topic
4. Privacy-enhancing techniques for storing and analysing confidential data
5. Federated Learning. This is a machine learning technique that trains an algorithm across multiple servers holding local data samples, without exchanging them. Research is needed into how this can be used in Finance applications, especially those that use confidential data.
6. Applications of Reinforcement learning in Finance. Existing applications include portfolio optimization and optimal trade execution. Further research is needed to extend this technique to other areas in finance.
7. The value of innovation projects in Finance. Innovative projects have a high-risk of failure and are often also focused on cost reduction and loss-avoidance topics. Therefore the impact on the P&L of the company is not immediately clear. The project is supposed to find ways of measuring the cost/benefit ratio and provide a conceptual approach.
8. Explainable Artificial Intelligence in Finance. Many empirical studies provide evidence of the numerous advantages AI can bring to the financial sector, offering a new approach to credit risk management. One highly relevant barrier for wider adoption of AI in the financial sector is related with the concept of explainability. AI solutions are often referred to as “black boxes” because typically it is difficult to trace the steps the algorithm took to arrive at its decision. The lack of XAI techniques tailored to risk models disincentivizes the use of machine learning in financial modelling and hampers progress. Main objective: Run a comprehensive list of global and local explainability techniques on ML-based credit risk models.
9. The use of "meta labeling" technique (tailored to non-HFT strategies). The approach consists of building a secondary ML model that learns how to use a primary exogenous model. It can help build an ML system on top of a white box (like a fundamental model founded on economic theory). The advantages of the approach is that it uses a way higher signal to noise ratio than when applying ML directly to (very noisy) traditional financial data. 
10. Machine learning for credit scoring. The objective is to train and test different ML models for the purpose of building an adaptive rating system applicable to both SME and personal lending, able to cope with low-quality, sparse data.