AAdevloper/nifty50-market-regime

📊 NIFTY 50 Market Regime Dataset

Dataset for training machine learning models to predict market regimes (RISK_ON/RISK_OFF) in Indian financial markets.

📋 Dataset Description

This dataset contains technical indicators and corresponding market regime labels for NIFTY 50, used to train binary classification models for regime prediction.

Market Regimes

• RISK_ON (1): Favorable market conditions - lower volatility, bullish momentum
• RISK_OFF (0): Cautious conditions - higher volatility, defensive positioning

📊 Features

Feature	Type	Description	Range
india_vix	float	India VIX volatility index	0-100+
rsi_14	float	14-day Relative Strength Index	0-100
ma_50	float	50-day moving average	> 0
ma_200	float	200-day moving average	> 0
regime	int	Target label (0=RISK_OFF, 1=RISK_ON)	0 or 1

Feature Descriptions

1. India VIX: Measures expected volatility in NIFTY 50

   • Low values (< 15): Low fear, potentially RISK_ON
   • High values (> 25): High fear, potentially RISK_OFF

2. RSI-14: Momentum indicator

   • < 30: Oversold (potentially bullish)

   • 70: Overbought (potentially bearish)

   • 40-60: Neutral

3. MA-50: Short-term trend indicator

4. MA-200: Long-term trend indicator

📈 Statistics

Load the dataset to see:

• Total samples
• Class distribution (RISK_ON vs RISK_OFF)
• Feature correlations
• Summary statistics

💻 Usage

Load with Pandas

import pandas as pd

# Load from local file
df = pd.read_csv("market_regime_data_nifty50.csv")

# View first few rows
print(df.head())

# Check class distribution
print(df['regime'].value_counts())

Load from Hugging Face

from datasets import load_dataset

# Load dataset
dataset = load_dataset("AAdevloper/nifty50-market-regime")

# Convert to pandas
df = dataset['train'].to_pandas()

Example Training Code

from sklearn.model_selection import train_test_split
from xgboost import XGBClassifier

# Prepare data
X = df[['india_vix', 'rsi_14', 'ma_50', 'ma_200']]
y = df['regime']

# Split
X_train, X_test, y_train, y_test = train_test_split(
    X, y, test_size=0.2, random_state=42, stratify=y
)

# Train
model = XGBClassifier(max_depth=6, learning_rate=0.1, n_estimators=100)
model.fit(X_train, y_train)

# Evaluate
score = model.score(X_test, y_test)
print(f"Accuracy: {score:.2%}")

🎯 Use Cases

1. Market Regime Classification: Train models to predict current regime
2. Trading Strategy Development: Use regime predictions for position sizing
3. Risk Management: Adjust portfolio based on predicted regime
4. Feature Engineering Research: Experiment with technical indicators
5. MLOps Pipeline Development: Practice model deployment and monitoring

🔧 Data Collection

The dataset includes:

• Historical technical indicators for NIFTY 50
• Calculated from OHLCV (Open, High, Low, Close, Volume) data
• Regime labels based on market behavior patterns

📊 Data Quality

• ✅ No missing values
• ✅ Features normalized/scaled appropriately
• ✅ Balanced class distribution (or document imbalance)
• ✅ No data leakage in feature engineering
• ✅ Temporally consistent

🏗️ Related Projects

This dataset is part of a complete MLOps pipeline:

• Model: market-regime-classifier
• Live Demo: MLOps Finance Pipeline Space
• Source Code: GitHub Repository

📄 Citation

If you use this dataset, please cite:

@misc{nifty50_market_regime,
  author = {AAdevloper},
  title = {NIFTY 50 Market Regime Dataset},
  year = {2025},
  publisher = {Hugging Face},
  howpublished = {\url{https://huggingface.co/datasets/AAdevloper/nifty50-market-regime}}
}

📄 License

MIT License - Free to use for research and commercial applications

---

Part of the MLOps Finance Pipeline project 🚀

For questions or improvements, visit the GitHub repository