Tirna Chakraborty

[ RESEARCH ] R / TIME SERIES / SIMILARITY SCORING

Economic Regime Research

finance_mode

Tirna Chakraborty

Macro Research / Quantitative Finance

This research project explores how changing macroeconomic conditions influence asset returns and whether these shifts can be used to improve return forecasting. Rather than relying solely on traditional asset pricing frameworks such as the Fama-French Five Factor Model, the study adopts a dynamic, data-driven approach that incorporates a broader set of economic indicators.

A comprehensive set of macro-financial state variables is constructed, including equity market performance, the yield curve, commodity prices such as oil and copper, monetary policy indicators, market volatility and stock-bond correlation. These variables are systematically transformed using rolling statistical techniques, including standardisation over a 10-year window and winsorisation, to ensure consistency and robustness across time.

Macro shot of digital stock tickers and financial data streams

The core methodology is built around a distance-based similarity framework. For each point in time, the current economic environment is compared with historical periods using Euclidean distance. The most similar historical regimes are identified, and their subsequent returns are used to generate forward-looking expectations. This enables the model to adapt to evolving macroeconomic conditions rather than assuming static relationships between risk factors and returns.

"The model treats macro context as a living state, comparing today's conditions with the historical periods they most closely resemble."

The framework is applied across multiple levels of financial analysis. At the factor level, timing strategies are implemented across key risk factors, including momentum, to evaluate whether regime-based signals can enhance performance. The analysis is further extended to industry portfolios and individual stocks, with predictive accuracy assessed using out-of-sample metrics such as mean squared error (MSE) and out-of-sample R^2.

Overall, the project highlights the importance of macroeconomic context in asset pricing and demonstrates how regime-aware modelling can improve forecasting performance. It also reflects strong capabilities in quantitative analysis, financial modelling and the application of advanced data techniques using R.

Method Snapshot

Variables

Yield curve, commodities, volatility, policy indicators and stock-bond correlation.

Similarity

Euclidean distance used to identify historical macro regimes most similar to the present.

Testing

Out-of-sample MSE and R^2 used to evaluate forecasting accuracy.