Credit Finance

The credit risk model landscape is heavily regulated. Using our explainable AI and advanced ML models which look into transaction, performance , social media, etc. we provide high accuracy, preferred degree of conservatism and keeps models in compliance with regulatory requirements

Given a default scenario, our AI models can help to model or predict the Loss (LGD) which traditional models can't find due to the bimodal nature of distribution. We take into view long and short term customer and economic conditions to model the loss under BASEL guidelines helping to optimise capital

The last block in the expected loss models is the exposure that banks have at default. Using both IRB and AIRB models, we can predict the exposure and aid in the institution's capital management. This helps in further growth of the banks' business yet maintaining requisite capital in case of downturn

Our VaR forecast use Artificial Neural Networks (ANNs) instead of HS or GARCH. Based on Mean Loss Comparison, Violation Ratio and Christofferson’s conditional coverage test, both the simulation and real data results prove that the ANN combinations have superior forecast performance than the individual VaR models

Based on a derivation of Smets Wouters models, this is an elaborate New Keynesian model where Wages and Prices are Calvo-sticky and numerous sources of shocks and interventions are used to correctly model or simulate the economy. This helps in stress testing the banks' capital in different economic conditions.

With the changing landscape of the amount of data available, new drivers like social media, sustainability information heavily impacts credit risk model especially in micro finance and wholesale models. While these create very accurate systems, we should also heed to making the models GDPR compliant.