import yfinance as yf
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import gradio as gr

def get_data(symbol, start_date, end_date):
    data = yf.download(symbol, start=start_date, end=end_date)
    return data['Close']

def monte_carlo_simulation(data, num_simulations, num_days):
    log_returns = np.log(1 + data.pct_change())
    mean = log_returns.mean()
    variance = log_returns.var()
    drift = mean - (0.5 * variance)
    stdev = log_returns.std()

    daily_returns = np.exp(drift + stdev * np.random.normal(size=(num_days, num_simulations)))
    price_paths = np.zeros_like(daily_returns)
    price_paths[0] = data.iloc[-1]

    for t in range(1, num_days):
        price_paths[t] = price_paths[t - 1] * daily_returns[t]

    return price_paths

def sigma_confidence_interval(price_paths, confidence_level=0.95):
    final_prices = price_paths[-1]
    lower_bound = np.percentile(final_prices, (1 - confidence_level) / 2 * 100)
    upper_bound = np.percentile(final_prices, (1 + confidence_level) / 2 * 100)
    return lower_bound, upper_bound

def plot_simulation(price_paths, title):
    plt.figure(figsize=(10, 5))
    plt.plot(price_paths)
    plt.title(title)
    plt.xlabel('Days')
    plt.ylabel('Price')
    plt.grid(True)
    plt.show()

def plot_correlation(nifty_data, sensex_data):
    correlation = nifty_data.corr(sensex_data)
    plt.figure(figsize=(10, 5))
    plt.scatter(nifty_data, sensex_data)
    plt.title(f'Correlation between NIFTY and SENSEX: {correlation:.2f}')
    plt.xlabel('NIFTY')
    plt.ylabel('SENSEX')
    plt.grid(True)
    plt.show()

def simulate(start_date, end_date, num_simulations, num_days):
    nifty_data = get_data('^NSEI', start_date, end_date)
    sensex_data = get_data('^BSESN', start_date, end_date)

    nifty_paths = monte_carlo_simulation(nifty_data, num_simulations, num_days)
    sensex_paths = monte_carlo_simulation(sensex_data, num_simulations, num_days)
    nifty_lower_bound, nifty_upper_bound = sigma_confidence_interval(nifty_paths)
    sensex_lower_bound, sensex_upper_bound = sigma_confidence_interval(sensex_paths)

    fig, axs = plt.subplots(1, 3, figsize=(40, 10))

    for path in nifty_paths.T:
        axs[0].plot(path)
    axs[0].set_title('Monte Carlo Simulation - NIFTY')
    axs[0].set_xlabel('Days')
    axs[0].set_ylabel('Price')
    axs[0].grid(True)

    for path in sensex_paths.T:
        axs[1].plot(path)
    axs[1].set_title('Monte Carlo Simulation - SENSEX')
    axs[1].set_xlabel('Days')
    axs[1].set_ylabel('Price')
    axs[1].grid(True)

    correlation = nifty_data.corr(sensex_data)
    axs[2].scatter(nifty_data, sensex_data)
    axs[2].set_title(f'Correlation between NIFTY and SENSEX: {correlation:.2f}')
    axs[2].set_xlabel('NIFTY')
    axs[2].set_ylabel('SENSEX')
    axs[2].grid(True)

    plt.tight_layout()
    return fig, f'NIFTY Sigma Confidence Interval at 0.95: [{nifty_lower_bound:.1f} : {nifty_upper_bound:.1f}], SENSEX Sigma Confidence Interval at 0.95: [{sensex_lower_bound:.1f} : {sensex_upper_bound:.1f}]'

def gradio_app(start_date, end_date, num_simulations, num_days):
    fig, interval = simulate(start_date, end_date, num_simulations, num_days)
    return fig, interval

iface = gr.Interface(
    fn=gradio_app,
    inputs=[
        gr.Textbox(placeholder="YYYY-MM-DD", label="Start Date"),
        gr.Textbox(placeholder="YYYY-MM-DD", label="End Date"),
        gr.Slider(minimum=100, maximum=10000, step=100, label="Number of Simulations"),
        gr.Slider(minimum=10, maximum=365, step=10, label="Number of Days for Forecast")
    ],
    outputs=[gr.Plot(label="Simulation Plots and Correlation Plot"), gr.Textbox(label="Sigma Confidence Interval")],
    title="Monte Carlo Simulation for NIFTY and SENSEX"
)

iface.launch()