Random Forest

from sklearn.ensemble import RandomForestRegressor
from sklearn.metrics import mean_squared_error

rf_strategy_returns = {}

for pair in top_10_pairs:
    stocks = pair[0]
    data = train_test_split_dict[stocks]

    train_X = data['train_X']
    train_y = data['train_y']
    test_X = data['test_X']
    test_y = data['test_y']
    spread = test_y

    # Create and train a random forest model
    rf_model = RandomForestRegressor(n_estimators=100, random_state=42)
    rf_model.fit(train_X, train_y)

    # output RMSE
    train_pred = rf_model.predict(train_X)
    test_pred = rf_model.predict(test_X)
    print(f"{stocks}:")
    print("  Train RMSE:", np.sqrt(mean_squared_error(train_y, train_pred)))
    print("  Test RMSE :", np.sqrt(mean_squared_error(test_y, test_pred)))


    final_value, cum_returns = score_fn(rf_model, test_X, spread, type="non_neural_net")


    n_days = len(cum_returns)
    annualized_return = (final_value ** (252 / n_days)) - 1
    annualized_return_percent = annualized_return * 100

    rf_strategy_returns[stocks] = annualized_return_percent
    print(f"  Annualized Return: {annualized_return_percent:.2f}%\n")

# Compute the mean annualized return for all 10 pairs
mean_annualized_return = np.mean(list(rf_strategy_returns.values()))
print(f"Mean Annualized Return of the Random Forest Model for the Top 10 Pairs: {mean_annualized_return:.2f}%")

pairs = list(rf_strategy_returns.keys())
returns = list(rf_strategy_returns.values())
pair_labels = [f"{p[0]}-{p[1]}" if isinstance(p, (list, tuple)) else str(p) for p in pairs]


mean_return = np.mean(returns)

# Visualization - Random Forest
plt.figure(figsize=(12, 6))
bars = plt.bar(pair_labels, returns, color='lightgreen', edgecolor='black')
plt.axhline(mean_return, color='red', linestyle='--', label=f'Mean = {mean_return:.2f}%')

for bar in bars:
    height = bar.get_height()
    plt.text(bar.get_x() + bar.get_width()/2, height + 0.5, f'{height:.2f}%', ha='center', va='bottom')

plt.title('Annualized Return of Random Forest Strategy for Top 10 Pairs')
plt.ylabel('Annualized Return (%)')
plt.xticks(rotation=45)
plt.legend()
plt.tight_layout()
plt.show()