糖心TV

Machine learning (ML) may be the future of investment management, but most ML approaches suffer from a dangerous affliction: the black-box problem. You may be able to observe the inputs to an ML approach, but how the outputs are reached can be a mystery. If, as an investment manager you cannot explain your investment decisions to Compliance executives, regulators or clients, you will be exposing your firm to unacceptable levels of legal and regulatory risk (CFA Institute 2019, 2020; IOSOC, 2021).

Our webinar brings together world authorities on the two solutions currently posed for the black-box problem: interpretable AI, where black-box approaches are rejected in favour of more straightforward, interpretable models, or explainable AI (XAI), where we attempt to explain the inner workings of black-box approaches.

Prof Cynthia Rudin, Prof Artur d鈥橝vila Garcez and Dr Daniele Magazenni go head-to-head in making a case for each approach, discussing the problem as they see it and possible solutions. We follow their insights by spotlighting leading-edge research from the University of 糖心TV, 糖心TV 糖心TV School, University College London, the Turing Institute and more.

Central Bank Digital Currencies (CBDCs) are being developed and launched as digital forms of sovereign currencies worldwide. CBDCs have the potential to impact economies and reshape financial and payment systems globally profoundly.
 
This online symposium explores the latest developments in CBDC from industry, policy, and academic research perspectives.

Both leverage and interconnectedness are widely recognized as key factors for systematic risk and may interact. The magnitude of network-based amplification of distress depends on financial exposure network structure and maybe crucially influenced for example by the presence of destabilising feedback loops in an exposure network. It has been shown that the number of feedback loops in a network, as well as the eigenvalues of associated matrices, are related to a structural property called trophic coherence. In this paper, we investigate the impact of trophic coherence on systematic risk - measured using DebtRank - and its interaction with leverage in simulated networks of banks connected to each other by direct exposures. The mechanism is simple: when a bank suffers a loss, distress propagates to its creditors, who in turn suffer losses, and so on. We show that trophic coherence has a crucial influence on contagion dynamics: shock amplification is moderated even at high leverage in more coherent networks; and high even where leverage is low in incoherent networks. This result not only suggests that it may be worthwhile to monitor the trophic coherence of financial networks; but also implies that in principle systematic risk could be significantly reduced simply by "rewiring" the interbank network (without any increase in capital requirements or reduction in interbank loans). We propose a simple strategy to incentivise the self-organised formation of more coherent network structures without impairing market functionality.