Financial markets are not random.
Empirical analysis of the NIFTY 50 index reveals strong evidence of:
• persistent structural regimes
• volatility clustering
• tail-risk concentration
This repository publishes the daily output of a proprietary regime-detection engine applied to:
➡️ NIFTY 50
The system analyzes market structure and classifies the environment into distinct behavioral regimes.
This repository does not contain the source code of the model.
Instead, it serves as a public archive of model outputs for transparency and research.
The system analyzes market data along three structural axes:
| Axis | Purpose |
|---|---|
| 📈 Trend | Detect directional persistence |
| 🌪 Volatility | Measure risk and clustering behavior |
| 💧 Liquidity | Identify structural fragility |
These axes combine to produce eight macro market regimes.
| Regime | Description |
|---|---|
| 🟢 Bull Grind | Slow, persistent upward trend with low volatility |
| 🔵 Structured Trend | Strong directional trend with sustained momentum |
| 🟡 Premium Harvest | Low-volatility range where option decay dominates |
| 🟠 Chaotic | Whipsaws, false breakouts, unstable structure |
| 🔴 Panic Crash | High-volatility downside shock regime |
| ⚪ Compression | Extremely low volatility before expansion |
| 🟣 Short Squeeze | Rapid upward volatility spike |
| 🟤 Thin Trend | Directional move with weak liquidity |
These regimes capture how the market behaves, not just where it moves.
Analysis of historical NIFTY data reveals:
Regimes tend to persist for multiple days rather than changing randomly.
Example persistence probabilities:
| State | P(stay) |
|---|---|
| Bull Grind | ~0.83 |
| Structured Trend | ~0.77 |
| Panic Crash | ~0.75 |
Large price moves tend to be followed by large moves.
Evidence:
• |Return| autocorrelation ≈ 0.27 at lag-1
• GARCH persistence ≈ 0.9 in calm regimes
• High-volatility states persist ~15 days
Extreme losses cluster in specific regimes.
Most downside tails occur during:
• Chaotic states
• Panic Crash regimes
These states represent <20% of time but dominate extreme events.
Each day the engine publishes:
- detected macro regime
- volatility estimates
- transition probabilities
- structural risk indicators
- market interpretation
Example output: