app.py

import gradio as gr
import pandas as pd
import yfinance as yf
from datetime import timedelta,datetime
import pytz
import matplotlib.pyplot as plt
from PIL import Image
import numpy as np

nifty_list = ["ADANIENT","ADANIPORTS","APOLLOHOSP","ASIANPAINT","AXISBANK","BAJAJ-AUTO","BAJFINANCE","BAJAJFINSV","BPCL","BHARTIARTL","BRITANNIA","CIPLA","COALINDIA","DIVISLAB","DRREDDY","EICHERMOT","GRASIM","HCLTECH","HDFCBANK","HDFCLIFE","HEROMOTOCO","HINDALCO","HINDUNILVR","ICICIBANK","ITC","INDUSINDBK","INFY","JSWSTEEL","KOTAKBANK","LTIM","LT","M&M","MARUTI","NTPC","NESTLEIND","ONGC","POWERGRID","RELIANCE","SBILIFE","SBIN","SUNPHARMA","TCS","TATACONSUM","TATAMOTORS","TATASTEEL","TECHM","TITAN","UPL","ULTRACEMCO","WIPRO","%5ENSEI"]

class Stocks:
    def __init__(self, symbol):
        self.symbol = symbol
        self.data = self.fetch_data()

    def fetch_data(self):
        try:
            
            ticker_symbol = self.symbol if self.symbol[0] == '%' else f"{self.symbol}.ns"

            
            data = yf.Ticker(ticker_symbol).history(period="10y", auto_adjust=True)
            return data
        except Exception as e:
            print(f"Error fetching data for {self.symbol}: {e}")
            return None

    def currentdateavailability(self, curDate):
        if curDate in self.data.index:
            return curDate
        else:
            
            curDate_dt = datetime.strptime(curDate, "%Y-%m-%d")
            newcDate_dt = curDate_dt - timedelta(days=1)

            
            newcDate_str = newcDate_dt.strftime("%Y-%m-%d")
            return self.currentdateavailability(newcDate_str)

    def CurPrice(self, curDate=None):
        curDate = self.currentdateavailability(curDate)
        return self.data.loc[curDate, 'Close'] if curDate is not None else self.data.iloc[-1]['Close']

    def NDayRet(self, N, curDate):
        curDate = self.currentdateavailability(curDate)
        NDate = self.data.index[self.data.index.get_loc(curDate) - N]
        return self.data.loc[curDate, 'Close'] - self.data.loc[NDate, 'Close']

    def DailyRet(self, curDate):
        curDate = self.currentdateavailability(curDate)
        return self.data.loc[curDate, 'Close'] - self.data.loc[curDate, 'Open']


    def Last30daysPrice(self, curDate=None):
        if curDate is not None:
            curDate = self.currentdateavailability(curDate)
            curDate_index = self.data.index.get_loc(curDate)
            return self.data.iloc[curDate_index - 30:curDate_index]['Close'].values
        else:
            return self.data.iloc[-30:]['Close'].values


    # This below function returns last 30 calender days close prices i.e. 30 days including holidays so less than 30 days close values are returned. Above fuction gives last 30 trading day close prices.
    # def Last30daysPrice(self, curDate=None):
    #     curDate = self.currentdateavailability(curDate)

    #     if curDate is not None:
    
    #         curDate_dt = datetime.strptime(curDate, "%Y-%m-%d")
    #         days_ago_30 = curDate_dt - timedelta(days=30)
    #         thirty_days_ago_date = days_ago_30.strftime("%Y-%m-%d")

    
    #         thirty_days_ago_date = self.currentdateavailability(thirty_days_ago_date)

    
    #         curDate_index = self.data.index.get_loc(curDate)
    #         thirty_days_ago_index = self.data.index.get_loc(thirty_days_ago_date)

    
    #         return self.data.iloc[thirty_days_ago_index:curDate_index + 1]['Close'].values
    #     else:
    #         return self.data.iloc[-30:]['Close'].values


stocks_dict = {symbol: Stocks(symbol) for symbol in nifty_list}

nifty_stocks = {symbol: stocks_dict[symbol] for symbol in nifty_list[:-1]}

nifty50 = {"nifty50": stocks_dict[nifty_list[-1]]}
def dateoffset(input_date_str):
    
    input_date_dt = datetime.strptime(input_date_str, "%Y-%m-%d")

    
    new_date_dt = input_date_dt - timedelta(days=1)

    
    new_date_str = new_date_dt.strftime("%Y-%m-%d")

    return new_date_str


def setdates(startdate, enddate):
    while startdate not in nifty50["nifty50"].data.index:
        startdate = dateoffset(startdate)

    while enddate not in nifty50["nifty50"].data.index:
        enddate = dateoffset(enddate)

    return startdate, enddate


def organisedata(startdate, enddate):
    
    startdate, enddate = setdates(startdate, enddate)

    
    symbols = list(nifty_stocks.keys())

    
    common_index = nifty50["nifty50"].data.loc[startdate:enddate].index

    
    data_frame = pd.DataFrame(index=symbols, columns=common_index)

    
    for symbol, stock_object in nifty_stocks.items():
        stock_data = stock_object.data.loc[startdate:enddate, 'Close']
        data_frame.loc[symbol] = stock_data.reindex(common_index).values

    return data_frame

def previoustimeframedata(n, startdate):
    
    startdate_dt = pd.to_datetime(startdate)

    
    ndaysagodate = startdate_dt - timedelta(days=int(n))

    
    ndaysagodate_str = ndaysagodate.strftime("%Y-%m-%d")
    startdate_str = startdate_dt.strftime("%Y-%m-%d")

    
    return organisedata(ndaysagodate_str, startdate_str)

def portfoliooperations(equity,startdate,ndaywindow,portfolio):
    
    startdate_dt = pd.to_datetime(startdate)
    windowenddate = startdate_dt + timedelta(days=int(ndaywindow))
    windowenddate_str = windowenddate.strftime("%Y-%m-%d")

    startdate,windowenddate = setdates(startdate,windowenddate_str) 
    
    window_data = organisedata(startdate,windowenddate) 

    differences = window_data.iloc[:, -1] - window_data.iloc[:, 0]  

    next_portfolio = differences[differences > 0].index.tolist() 

    
    portfolio_sum = window_data.loc[portfolio, window_data.columns[0]].sum()

    
    multiplier = equity / portfolio_sum if portfolio_sum != 0 else 0


    portfolio_value = pd.DataFrame(index=window_data.columns, columns=['value'])

    for date in window_data.columns:
    
      portfolio_sum = window_data.loc[portfolio, date].sum()
    
      portfolio_value.loc[date, 'value'] = portfolio_sum * multiplier


    return next_portfolio,portfolio_value

def mainfunction (equity,startdate,enddate,ndaywindow):
    
    pastwindow = previoustimeframedata(n=ndaywindow,startdate=startdate) # No Errors untill here

    differences = pastwindow.iloc[:, -1] - pastwindow.iloc[:, 0]

    portfolio = differences[differences > 0].index.tolist() # No Errors untill here

    portfolio,portfolio_value = portfoliooperations(equity=equity,startdate=startdate,ndaywindow=ndaywindow,portfolio=portfolio)



    enddate_tz = datetime.strptime(enddate,"%Y-%m-%d").replace(tzinfo=pytz.timezone('Asia/Kolkata'))

    while portfolio_value.index[-1] < pd.to_datetime(enddate_tz) - timedelta(days=int(ndaywindow)):

      portfolio,new_portfolio_value = portfoliooperations(equity=equity,startdate=startdate,ndaywindow=ndaywindow,portfolio=portfolio)

      portfolio_value = pd.concat([portfolio_value, new_portfolio_value])

      startdate = (pd.to_datetime(startdate)+ timedelta(days=int(ndaywindow))).strftime("%Y-%m-%d")
      
      equity = portfolio_value.iloc[-1, 0]

    return portfolio_value

def calculate_cagr(series):
    total_return = (series.iloc[-1] / series.iloc[0]) - 1
    num_years = len(series) / 252  
    cagr = (1 + total_return) ** (1 / num_years) - 1
    return cagr * 100


def calculate_volatility(series):
    return series.pct_change().std() * np.sqrt(252) * 100


def calculate_sharpe_ratio(series, risk_free_rate=0):
    cagr = calculate_cagr(series)
    volatility = calculate_volatility(series)
    sharpe_ratio = (cagr - risk_free_rate) / volatility
    return sharpe_ratio


def final_function(equity,startdate,enddate,ndaywindow):

    equity = int(equity)
    ndaywindow = int(ndaywindow)

    portfolio_value = mainfunction(equity=equity,startdate=startdate,enddate=enddate,ndaywindow=ndaywindow)
    nifty_data = nifty50["nifty50"].data
    subset_data = nifty_data[startdate:enddate]
    initial_nifty = subset_data['Close'][0]
    nifty_dataseries = (equity/initial_nifty)*subset_data['Close']
    plt.figure(figsize=(10, 6))
    plt.plot(portfolio_value['value'], label='Strategy')
    plt.plot(nifty_dataseries, label='Nifty50 as Benchmark')
    plt.title('Benchmark vs Strategy')
    plt.xlabel('Date')
    plt.ylabel('Equity')
    plt.legend()


    image_path = "output_plot.png"
    plt.savefig(image_path)
    plt.close()


    image = Image.open(image_path)

    strategy_cagr = calculate_cagr(portfolio_value['value'])
    strategy_volatility = calculate_volatility(portfolio_value['value'])
    strategy_sharpe_ratio = calculate_sharpe_ratio(portfolio_value['value'])

    benchmark_cagr = calculate_cagr(nifty_dataseries)
    benchmark_volatility = calculate_volatility(nifty_dataseries)
    benchmark_sharpe_ratio = calculate_sharpe_ratio(nifty_dataseries)


    return image, strategy_cagr, strategy_volatility, strategy_sharpe_ratio, benchmark_cagr, benchmark_volatility, benchmark_sharpe_ratio
title = "Portfolio tracking Nifty50 Stocks"
description = """
This App Demo is made for an Assignment. This Demo takes Initial Equity, Start Date, End Date, Time Window as inputs. Due to COVID 19 causing the fall of almost all Stock Prices, some Dates might result in strategy falling to zero at March 2020. Hence please try other dates 
"""


iface = gr.Interface(
    fn=final_function,
    inputs=[
        gr.Textbox(label="Equity",placeholder="Enter Equity Number"),
        gr.Textbox(label="Start Date",placeholder="YYYY-MM-DD"),
        gr.Textbox(label="End Date",placeholder="YYYY-MM-DD"),
        gr.Textbox(label="N-day Window",placeholder="Enter Window in Days")
    ],
    outputs=[
        gr.Image(type="pil"),
        gr.Textbox(label="Strategy CAGR (%)"),
        gr.Textbox(label="Strategy Volatility (%)"),
        gr.Textbox(label="Strategy Sharpe Ratio"),
        gr.Textbox(label="Benchmark CAGR (%)"),
        gr.Textbox(label="Benchmark Volatility (%)"),
        gr.Textbox(label="Benchmark Sharpe Ratio")
    ],
    title=title,
    description=description,
    
)

if __name__ == "__main__":
    iface.launch()