import pandas as pd from sklearn.model_selection import train_test_split from sklearn.ensemble import RandomForestRegressor from sklearn.metrics import mean_squared_error import numpy as np # Step 1: Load the data related to Largecap mutual funds mf_data = pd.read_csv("MFLargeMetrics.csv") mf_data.dropna(inplace=True) # Step 2: Feature Engineering # Assign higher rewards to good metrics and penalties to bad metrics # Normalize the features to bring them to a comparable scale # Good metrics: Returns, 3Y_CAGR, 5Y_CAGR, Sharpe_Ratio mf_data['Normalized Return (1Y)'] = mf_data['Return (1Y)'] / mf_data['Return (1Y)'].max() mf_data['Normalized_3Y_CAGR'] = mf_data['CAGR (3Y)'] / mf_data['CAGR (3Y)'].max() mf_data['Normalized_5Y_CAGR'] = mf_data['CAGR (5Y)'] / mf_data['CAGR (5Y)'].max() mf_data['Normalized_Sharpe'] = mf_data['Sharpe ratio'] / mf_data['Sharpe ratio'].max() # Bad metrics: PE_Ratio, Drawdown, Risk mf_data['Normalized_PE'] = mf_data['P/E'] / mf_data['P/E'].max() mf_data['Normalized_Expense_Ratio'] = mf_data['Expense ratio'] / mf_data['Expense ratio'].max() mf_data['Normalized_Volatality'] = mf_data['3-year volatility'] / mf_data['3-year volatility'].max() mf_data['Normalized_Asset_under_management'] = mf_data['Asset under management'] / mf_data['Asset under management'].max() mf_data['Normalized_Max_Drawdown'] = mf_data['Max Drawdown']/mf_data['Max Drawdown'].min() # Step 3: Calculate score for each mutual fund. Here I have given more weightage to 3 year returns # and Sharpe Ratio while on the other hand, I have penalized funds with higher drawdown and higher P/E ratio mf_data['Score'] = 0.4 * mf_data['Normalized_3Y_CAGR'] + \ 0.2 * mf_data['Normalized_5Y_CAGR'] + \ 0.3 * mf_data['Normalized Return (1Y)'] - \ 0.2 * mf_data['Normalized_Volatality'] - \ 0.05 * mf_data['Normalized_Expense_Ratio'] + \ 0.3 * mf_data['Normalized_Sharpe'] - \ 0.2 * mf_data['Normalized_PE'] - \ 0.3 * mf_data['Normalized_Max_Drawdown'] # Step 4: Define features and target X = mf_data[['CAGR (3Y)', 'CAGR (5Y)', 'P/E', 'Return (1Y)', '3-year volatility', 'Expense ratio', 'Alpha (3 years)', 'Sharpe ratio', 'Max Drawdown', 'Asset under management']] y = mf_data['Score'] # Split the data into training and testing sets X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=42) # Step 5: Initialize the model, train the model and predict the scores rf = RandomForestRegressor(n_estimators=200, random_state=42) # Train the model rf.fit(X_train, y_train) # Predict on the test set y_pred = rf.predict(X_test) # Evaluate the model mse = mean_squared_error(y_test, y_pred) print(f'Mean Squared Error: {mse}') print(rf) # Step 6: Get feature importance importances = rf.feature_importances_ feature_importance_df = pd.DataFrame({ 'Feature': X.columns, 'Importance': importances }).sort_values(by='Importance', ascending=False) print("\n\n") print(feature_importance_df) # Step 7: Select top 5 mutual funds based on predicted performance mf_data['Predicted Score'] = rf.predict(X) top_5_funds = mf_data.sort_values(by='Predicted Score', ascending=False).head(5) top_5_funds.style.set_properties(**{'border': '1.3px solid green', 'color': 'magenta'}) print("\n\n") top_5_funds[['Name', 'Predicted Score']].style.set_properties(**{'border': '1.3px solid green', 'color': 'blue'})

🧠 Python's Smart Approach to Mutual Fund Selection!

Subscribe to our premium content to read the rest.

Keep Reading

AlgoEdge Insights