Multi-Source Financial Dataset for Regime-Aware Asset Allocation: NSE Sectoral Indices, Stock Fundamentals, and Macro-Financial Indicators (2013-2024)

This study introduces a regime-aware asset allocation framework based on a novel macro-financial signal — the Market Health Index (MHI) — and evaluates its performance across dynamic stock-bond-cash allocations. The hypothesis driving this research is that market conditions can be effectively captured using macroeconomic, technical, and sentiment-based indicators, and that portfolios dynamically reallocated according to market regimes (Bullish, Neutral, Bearish) can achieve better downside protection and performance stability than static allocation strategies.
Data Components and Sources:
The dataset used in this research integrates several layers:
1. Sectoral Independence Data:
Weekly closing prices for 15 sectoral indices listed on the NSE from January 2022 to December 2024 were collected directly from NSE’s historical data repository. Weekly log returns were computed and then transformed to uniform margins, enabling copula-based dependency analysis using Kendall’s Tau correlation to identify the least-dependent sectors.
2. Stock Selection Data:
For each of the top 10 least-correlated sectors, a list of 12–20 constituent stocks was compiled. The top 10 stocks per sector were shortlisted using 2-year absolute return , based on daily price data. From these, one representative stock per sector was selected using a scoring mechanism that combines:
o Stochastic Risk (risk-adjusted efficiency),
o Bayesian Impact (causal strength of financial indicators), and
o Cosine Similarity Drift (behavioral alignment with sector index).
Monthly prices (2013–2024) for the final 10 stocks were sourced from Yahoo Finance and used to calculate quarterly returns.
3. Market Health Index (MHI) Data:
MHI is constructed using 12 indicators, grouped into six categories:
o Micro: Real GDP growth and interest rate (RBI repo), Consumer Confidence Index
o Liquidity: FII, DII, SPP flows
o Risk: Nifty Historical Volatility and Global Uncertainty Index
o Technical: 200-day Moving Average and RSI
o Fundamental: Nifty P/E ratio
o Macro: Global Risk Sentiment Index
These indicators were normalized using Min-Max scaling.
Labeling and Regime Detection:
Using HMM, the quarterly MHI time series was classified into three latent regimes — Bullish, Neutral, Bearish — based on inferred state transitions. These regimes were used to guide dynamic allocation of assets via Bayesian Optimization, targeting maximum expected return adjusted for volatility.
Interpretation of the Data:
The dataset supports dynamic portfolio reallocation conditioned on inferred market regimes. It allows analysts to compare performance across allocation strategies (e.g., 60/40, all-equity, conservative, risk-parity) under various macro-financial conditions.
Usage Note: The dataset is suitable for researchers interested in regime modeling, macro-financial integration, optimization under uncertainty, and portfolio construction strategies that emphasize resilience during market stress.