Predictive analytics for market volatility, trading algorithms, and liquidity risk management

Predictive analytics can be defined as a fact-based decision-making process to gain insights about expected future events and decision outputs. It involves building statistical models using data and implementing simulations and scenarios to predict outcome probabilities. It is one of the three perspectives of business analytics, the other two being performance or descriptive analytics and exploratory or prescriptive analytics (Zhang et al., 2005; Foucault et al., 2013). Predictive analytics basically answers the questions: What is likely to happen? What can be the reasonable outcomes for a decision? What are the expected probabilities for all possible alternative outcomes? Why do I need to predict?

The answer to the first question is easy. For any decision we make now, there will be consequences in the future. Many decisions in industries, such as financial services, insurance, healthcare and marketing depend on future events and their impact. For instance, to improve risk management, a loan officer might want to know the likelihood of default during the term of a loan before approving a loan application. To improve profitability on credit cards, a bank may want to know who is likely to use up their accumulated rewards points before expiration. For guiding capital allocation decisions, an insurance executive may want to know the expected claims for each insurance policyholder in next twelve months (Cont, 2001; Aldridge, 2013; Avramov et al., 2021).