Backtesting
12_2 momentum strategy (nifty50)

LINK TO GOOGLE COLAB NOTEBOOK
LINK TO GITHUB

import yfinance as yf

import numpy as np

import pandas as pd

import matplotlib.pyplot as plt

# Get NIFTY 50 stock symbols

stocks = pd.read_html('https://en.wikipedia.org/wiki/NIFTY_50')[2]['Symbol']

stocks = stocks + '.NS'

# Download historical data

df = yf.download(stocks.to_list(), start='2014-01-01')

df_close = df['Adj Close']

df_dret = df_close.pct_change()

df_mret = (df_dret + 1).resample('M').prod() - 1

# Calculate 12-month rolling returns

df_mret_12 = df_mret.rolling(12).apply(lambda x: (x + 1).prod() - 1).dropna()

# Function to calculate top performers for a given date

def top_performers(date):

    top = df_mret_12.loc[date].nlargest(10)

    return df_mret.loc[date:][1:3][top.index].mean(axis=1).values

# Calculate bimonthly returns of the strategy

bimonthly_ret = []

for date in df_mret_12.index[:-2:2]:  # Adjust index for 2-month rebalancing

    try:

        bimonthly_ret.extend(top_performers(date)) #we use extend not append

    except IndexError:

        continue

print(bimonthly_ret[0:3])

# Calculate portfolio value over time

portfolio_value = [1]

for r in bimonthly_ret:

    portfolio_value.append(portfolio_value[-1] * (1 + r))

# Create a DataFrame for portfolio value

portfolio_dates = df_mret_12.index[1:-1:1]  # Adjust dates for 1-month periods

portfolio_df = pd.DataFrame({

    'Date': portfolio_dates,

    'Portfolio Value': portfolio_value[1:]

})

portfolio_df.set_index('Date', inplace=True)

# Calculate drawdown

portfolio_df['Drawdown'] = portfolio_df['Portfolio Value'] / portfolio_df['Portfolio Value'].cummax() - 1

# Calculate Sharpe ratio

annualized_return = np.prod([1 + r for r in bimonthly_ret]) ** (12 / len(bimonthly_ret) ) - 1

annualized_std = np.std(bimonthly_ret) * np.sqrt(12)  

sharpe_ratio = annualized_return / annualized_std

# Calculate Beta

# Use NIFTY 50 index as benchmark

nifty50 = yf.download('^NSEI', start='2014-01-01')['Adj Close']

nifty50_mret = nifty50.pct_change().resample('M').apply(lambda x: (x + 1).prod() - 1).dropna()

cov_matrix = np.cov(nifty50_mret[df_mret_12.index[1]:df_mret_12.index[-2]], bimonthly_ret)

beta = cov_matrix[0, 1] / cov_matrix[0, 0]

print(f"Annualized Return: {annualized_return:.2%}")

print(f"Annualized Volatility: {annualized_std:.2%}")

print(f"Sharpe Ratio: {sharpe_ratio:.2f}")

print(f"Beta: {beta:.2f}")

print(f"Maximum Drawdown: {portfolio_df['Drawdown'].min():.2%}")

# Plot portfolio value over time

plt.figure(figsize=(10, 7))

plt.plot(portfolio_df['Portfolio Value'], label='Portfolio Value')

plt.title('Portfolio Value Over Time')

plt.xlabel('Date')

plt.ylabel('Portfolio Value')

plt.legend()

plt.grid(True)

plt.show()

# Plot drawdown over time

plt.figure(figsize=(10, 7))

plt.plot(portfolio_df['Drawdown'], label='Drawdown', color='red')

plt.title('Drawdown Over Time')

plt.xlabel('Date')

plt.ylabel('Drawdown')

plt.legend()

plt.grid(True)

plt.show()

Result: