Machine learning cab be used to design and understand problem and trying to derive solution using data. Being working as FinTech professional, I am seeing the change happening.

I have extensively worked in Market Risk. As per definition, Market Risk is the risk of loss due to factors that affect an entire market or asset class. Market risk is also referred as undiversifiable risk as it affects all asset class and is unpredictable. There are number of bank regulations regarding the same to keep value of risk of par, however, it is always tough to reach the outcome.

The recent example was in 2007 due to sub-prime crisis and now due to COVID market became predictable again. To mitigate the risk and find a solution, bank has started using existing portfolio and had created synthetic portfolios as well and is giving shocks to them. The shocks are nothing but kind of predictable behaviour if specific conditions happens. Example - stay at home order is placed by government. In that case, airlines, gas, hospitality, and related industries will be impacted. If it stays for say 3 months, what should be value of stock or value at risk if these related holdings will be in portfolio. This scenario can be repeated for multiple type of factors and for different given times keeping other historical data factors like economy, inflation of country etc. in place. This is helping to give some idea what bank can do if such conditions arise and goes for long. This is helping derive solution to bank and customer and helping government as well so that regulations can be met.

The system is somewhat in mid-stage and proving to be efficient as it is able to predict and hence helping bank to think about solution to minimize the loss. Bank has enough data to come up with different set of models. Things are getting settled in terms of deriving a settled solution. The model/algorithms are gradually getting smart to learn from external patterns like market trends (getting feeds from Bloomberg, Reuters and other systems) to catch the key attributes and implement the shocks to predict the outcome for future based upon given time frame.

This is helping bank to first minimize the risk, secondly gives the liberty to utilize the data in efficient way to design better products and thirdly helping to be in threshold limits to comply the financial laid guidelines.

Rome was not made in a day, but it will be made soon. Things are evolving and I still believe things can’t be predicated 100% but for sure there will be enough indications and as we say - precaution is better than cure. Based upon the indications, we will be able to plan and act in better way.