This repository contains a Python-based quantitative research project built as part of an inter-hostel competition. The problem statement was provided by Beyond IRR (see Kriti2026_quant.pdf). The project focuses on predicting market movement and generating long-term Buy/Sell signals.
The work explores three approaches:
- Pure Quantitative (rule/indicator-based) approach
- Machine Learning (ML) approach
- Integrated approach combining Quant + ML
The repository is organized so that file names clearly indicate the approach used, and all evaluation outputs/metrics are stored in the performance_stats/ folder with descriptive names.
- Analyze index/market data
- Build long-term trading signals (Buy/Sell)
- Evaluate strategies using relevant performance metrics (returns, drawdowns, risk-adjusted measures, etc.)
- Compare outcomes across:
- Quant-only logic
- ML-only model
- Integrated Quant+ML pipeline
-
quantitative_approach.py- Implements a traditional quantitative strategy, typically based on engineered indicators, rules, trend logic, or statistical signals.
- Produces Buy/Sell signals and/or positions for long-term holding periods.
-
ML.py- Implements the machine learning approach.
- Includes feature engineering and model training/inference logic to predict market direction/movement and translate predictions into signals.
-
integrated_approach.py- Implements the hybrid approach, integrating quantitative features/signals with ML predictions (or using quant logic as filters/confirmations).
- Designed to test whether combining both improves robustness and performance.
-
indexes.csv- Primary dataset used for the analysis (index/market values and related fields).
- Used across quant, ML, and integrated workflows.
-
Kriti2026_quant.pdf- Official problem statement shared by Beyond IRR for the competition.
-
performance_stats/- Contains all performance evaluation metrics and outputs, stored with meaningful filenames.
- This folder is the primary place to look for backtest summaries, metric tables, and evaluation artifacts.
-
PERFORMANCE_AND_ML_REPORT.md- Consolidated write-up/report describing performance results and ML observations.
-
plot_performance.py- Utility for generating plots/visualizations of performance and/or equity curves from computed results.
ML_README.md- ML-specific notes and documentation (features, models, pipeline details, etc.)
Exact commands may vary depending on how each script is implemented, but a common workflow is:
-
Run Quant strategy
python quantitative_approach.py
-
Run ML strategy
python ML.py
-
Run Integrated strategy
python integrated_approach.py
-
Generate plots
python plot_performance.py
After running, check:
performance_stats/for saved metrics/resultsPERFORMANCE_AND_ML_REPORT.mdfor summarized findings
Depending on the approach, the scripts typically generate:
- Predicted market movement labels/scores (ML / integrated)
- Buy/Sell signals or position series
- Backtest performance metrics (saved under
performance_stats/) - Visualizations (via
plot_performance.py)
This project is for educational/research purposes (inter-hostel competition). It does not constitute financial advice. Trading and investing involve risk, and past performance does not guarantee future results.