ISI is restructuring its organization chart, for such reason some information is not available at the moment.

Industrial research
AI Lab for Finance & Business

Head of operations
Paolo Bajardi

Scientific Leaders
Paolo Bajardi, Francesco Bonchi, André Panisson

Created in the context of the new Industrial Research area of the ISI Foundation, the AI Lab for Finance & Business is a joint effort of ISI Foundation and Intesa Sanpaolo Innovation Center (owned by Intesa Sanpaolo).
This Lab is an industrial research lab aimed at developing novel cutting-edge data science and machine learning techniques for specific real-world cross-industries challenges with special focus on the financial domain. The Lab consists of a team of senior and junior data science researchers, hired by ISI Foundation, and in collaboration with Innovation Center Lab domain experts.

Read more

In collaboration with

Projects

Advanced Early Warning System
L

Leveraging the fully anonymised temporally-resolved networked data of firm-to-firm transactions, the project aims at investigating cascading effects on credit risk. The goal is to devise an actionable pipeline able to extract meaningful features from non-conventional data to outline novel predictive models of economic-financial performance.

Read more
Asset Allocation
T

The goal of the project is to develop analytical models that give a temporal representation of both assets and liabilities of life and non-life insurance business.

Read more
Counterparty Exposure Esploraton - Phase 2
H

Hedging XVAs (X Valuation Adjustments) represents a formidable challenge in risk management. XVAs are generated from any non-collateralized derivative, then the various positions are aggregated and hedged as a sole complex portfolio. Thus, the resulting portfolio is subject to various risk factors such as credit, interest rates, inflation, equity, foreign exchange and commodities. In order to appropriately manage all these risks, it is fundamental to visualize them in a condensed and effective way.

Read more
Counterparty Exposure Exploration
T

The research activities aim to design and prototype a decision support tool for monitoring and exploring derivative contracts. Financial institutions stipulate a wide variety of derivative contracts with different counterparties (central counterparties, financial counterparties, and corporate counterparties), different underlying instruments, and different risk factors.

Read more
Cybersecurity
T

The research project aims at designing and prototyping a digital system that allows different parties to compute a global assessment for a common individual without sharing the local. The system will be based on cryptographic tools (e.g. homomorphic encryption techniques) and accomplish with current data protection laws (e.g., GDPR).

Read more
DVA
H

Hedging the DVA (Debt Valuation Adjustment) represents a formidable challenge in risk management, since the bank (or hedger) should sell protection on itself to cover his own credit risk. Moreover, DVA is a hybrid risk, subject to numerous and possibly correlated risk drivers. A popular strategy when facing this problem is hedging the credit risk by proxy: e.g. selling protection on a basket of names which are highly correlated with the hedger.

Read more
Early warning algorithms
W

Widely used protection models that determines the allocation between “risky” and “non-risky” assets are not optimal for low-frequency tradable portfolios adjustments. The objective is to develop a proprietary early warning algorithm capable of spotting sudden changes in market dynamics.

Read more
Risk Overlay
T

The project will investigate the application of machine learning techniques to the problem of tail risk hedging for investment portfolios.  Given the challenging and open-ended vision of this project, it is articulated in several correlated activities with the final goal of developing a coherent and innovative algorithmic pipeline to identify risk drivers in financial portfolios and to devise optimal strategies to mitigate such risks.

Read more
XVA Calibration
M

Modelling valuation adjustments (collectively known as XVA) represents one of the toughest challenges in derivatives pricing and financial engineering at large, as it involves modelling net future exposure with portfolio effects, based on consistent specifications of market price dynamics under the risk neutral measure across several currencies and asset classes.

Read more