correlations.py

pandas as pd
import numpy as np

# Assuming returns is your dataset with returns data
def calculate_correlations(returns, symbols, window=50):
    """
    Calculate rolling correlations between ETFs
    
    Parameters:
    returns: numpy array or pandas DataFrame of returns
    symbols: list of symbol names
    window: correlation lookback period
    """
    # Create DataFrame if not already
    df = pd.DataFrame(returns, columns=symbols)
    
    # Dictionary to store correlation pairs
    corr_pairs = {}
    
    # Calculate rolling correlation for each pair
    for i in range(len(symbols)):
        for j in range(i+1, len(symbols)):
            pair_name = f'{symbols[i]}_{symbols[j]}'
            corr_pairs[pair_name] = df[symbols[i]].rolling(window=window).corr(df[symbols[j]])
    
    # Combine all correlations into one DataFrame
    corr_df = pd.DataFrame(corr_pairs)
    
    return corr_df

# Example usage:
# corr_df = calculate_correlations(returns, symbols)

# To get the most recent correlation matrix:
def get_current_corr_matrix(df, symbols):
    """Get the most recent correlation matrix between all symbols"""
    return df[symbols].corr()

symbols = ['XLF','XLI','XLK', 'XLY']
prices = yf.download(symbols)['Adj Close'].dropna()
returns = prices.pct_change()
returns.head()