Adaptive Moving Average Strategy – Finance Hackathon

Our Approach

We developed an Adaptive Moving Average Strategy that dynamically decides whether to use a Simple Moving Average (SMA) or an Exponential Moving Average (EMA) based on market behavior parameters.

The model uses three key indicators:

• Volatility – measures how much the stock price fluctuates.
• Trend Strength – measures how strongly the market is moving in one direction.
• Noise – measures randomness or choppiness in price movements.

Decision Logic

• EMA is used in volatile and trending markets where a faster response is beneficial.
• SMA is used in calmer or sideways markets to smooth short-term fluctuations.

This makes the algorithm adaptive and capable of handling different market regimes.

---

Libraries Used

Library	Purpose
pandas	Data handling and manipulation
numpy	Numerical operations
matplotlib	Data visualization
streamlit	Building an interactive dashboard
yfinance	Fetching updated stock data
fastapi, uvicorn	Optional: for API endpoints

---

File Overview

1. src/backtest.py

This file implements a basic Moving Average Crossover Backtest.

Core logic:

• Entry: When the fast MA crosses above the slow MA (Bullish crossover → Buy next day).
• Exit: When the fast MA crosses below the slow MA (Bearish crossover) or when one of the following triggers occur:
  • Stop Loss: 3% below entry price
  • Take Profit: 5% above entry price
  • Maximum Holding Period: 7 trading days

Outputs:

• Total Return (%)
• Maximum Drawdown (%)
• Sharpe Ratio
• Win Rate (%)
• Number of Trades

---

2. src/optimize_dynamic_trend_noise.py

This file builds on backtest.py and optimizes the logic by repeatedly calling the backtest for different moving average configurations.

It calculates:

• Volatility using the standard deviation of returns.
• Trend Strength using directional percentage movement.
• Noise Ratio using the difference between cumulative and total absolute moves.

Based on these, it dynamically selects whether to use SMA or EMA and identifies the best MA window pair (e.g., 10/20, 12/26, 20/50, etc.) for each stock.

All results and performance metrics are stored in the /reports/ folder.

---

3. dashboard/app.py

An interactive Streamlit dashboard that displays:

• The selected stock’s price, moving averages, and crossover signals.
• Key metrics (Return, Win Rate, Sharpe Ratio, Max Drawdown, Trades).
• Market condition parameters (Volatility, Trend Strength, Noise).
• The final decision of whether SMA or EMA was chosen and why.

The dashboard uses only three months of data (August 1, 2025 – November 7, 2025) as required by the hackathon.

---

Data Source

All stock data was fetched using the Yahoo Finance (yfinance) library, which provides reliable and frequently updated price data.

Data folders:

• data/raw/ – Unprocessed data from Yahoo Finance
• data/processed/ – Data after moving averages and signal computation
• data/trimmed/ – Final filtered data (Aug–Nov 2025) used for backtesting and dashboard

---

Stocks Analyzed

Symbol	Company
RELIANCE.NS	Reliance Industries
TCS.NS	Tata Consultancy Services
INFY.NS	Infosys Ltd
HDFCBANK.NS	HDFC Bank
ICICIBANK.NS	ICICI Bank
ADANIENT.NS	Adani Enterprises
ITC.NS	ITC Limited
MARUTI.NS	Maruti Suzuki
TATASTEEL.NS	Tata Steel
LT.NS	Larsen & Toubro
SBI.NS	State Bank of India

Each stock’s optimized strategy and performance metrics are saved in /reports/.

---

Output Summary

Each report (e.g., reports/RELIANCE_NS_dynamic_trend_noise_optimization.csv) includes:

Symbol	Volatility	TrendStrength	Noise	MA_Type	MA_Pair	Return	WinRate	Sharpe	MaxDD	Trades
RELIANCE.NS	1.22	6.97	60.10	EMA	10/20	-1.93	33.3	0.97	-4.98	3

---

How to Run

1. Install dependencies

pip install -r requirements.txt