Hadi Safari Katesari is Developing Tools for Smarter Economic Predictions

Every day, the stock market and overall economy top the news headlines — but while the raw numbers are easily accessible, the deeper meanings and complex interactions of these fundamental financial pillars are often obscure. 

Mathematical Sciences Teaching Assistant Professor Hadi Safari KetasariHis newly released research paper, co-authored with Samira Zaroudi, John Jay College of Criminal Justice, and S. Yaser Samadi, Southern Illinois University, blends statistical rigor with real-world usability through powerful new tools: the Bayesian Copula Factor Autoregressive (BCFAR) and Time Series Bayesian Factor Regression (TS-BFR) models.

These tools are designed to uncover obscure patterns in complicated data that comes in mixed forms, such as numbers and categories, and evolves over time — with results available for real-time applications.

“Real-world economic data isn’t just numbers,” explained Safari Katesari.  “It’s also categories like recession indicators or credit ratings. Our models combine that information into a single framework, so we can understand it better and predict trends more accurately.”

These models explore the hidden forces, such as investor confidence or economic stability, that help explain why economic indicators might move together. Movements in the stock market can affect investor confidence, while economic changes like interest rate adjustments or employment trends can influence market performance. Although these factors can’t be directly measured, they influence many other indicators. Understanding their interaction over time allows experts to make better economic forecasts, shape policy and manage risk.

Using advanced techniques, Safari Katesari's team can estimate and analyze how these invisible drivers shift over time.

The groundbreaking BCFAR model takes this research a step further by combining those hidden forces with a smart mathematical method that connects data like numbers and categories. This makes it possible to detect subtle, dynamic relationships among gross domestic product, stock indices, interest rates, policy decisions and other factors.

“We created these models from the ground up,” he said. “They’re not just an improvement — they’re a whole new way of bringing together complex economic data.”

“We created these models from the ground up,” he said. “They’re not just an improvement — they’re a whole new way of bringing together complex economic data.”

The results are promising. In testing with both simulations and real U.S. macroeconomic data, the models proved both accurate and fast. They show broad potential to help policymakers, financial analysts and business leaders quickly discern viable, actionable insights. 

For governments and central banks, this could mean smarter, more responsive decisions on interest rates or inflation. For businesses, it offers better forecasting tools to guide investment, staffing and resource planning. And for investors, it provides early clues about market trends and risks.

“These tools help us track both visible trends and invisible drivers,” he said. “That’s deeply valuable for planning, whether you’re managing a private portfolio or a national economy.”

The models could also be adapted for fields such as healthcare, where data might include lab results and diagnostic categories, or social science research that involves tracking behaviors or survey responses over time.

Safari Katesari is especially proud that the models go beyond theory to work in the real world.

“Our model helps make sense of economic signals,” he said. “It gives us better tools to predict change, respond to uncertainty and make smarter decisions — whether in Washington or on Wall Street.”