stock-oracle/tests/test_yfinance_comparison.py

"""
Test script to compare Stock Oracle financial data with direct yfinance data
"""

import asyncio
import requests
import yfinance_plus as yf
from datetime import datetime, timezone
import pandas as pd
import json
from typing import Dict, Any

# API configuration
API_URL = "http://localhost:18001/api/v1"

def get_yfinance_data(ticker: str) -> Dict[str, Any]:
    """Get financial data directly from yfinance"""
    print(f"\n=== Getting yfinance data for {ticker} ===")
    
    # Create ticker object
    stock = yf.Ticker(ticker)
    
    # Get basic info
    info = stock.info
    print(f"Company: {info.get('longName', 'N/A')}")
    print(f"Sector: {info.get('sector', 'N/A')}")
    print(f"Industry: {info.get('industry', 'N/A')}")
    
    # Get quarterly financial statements
    quarterly_income = stock.quarterly_income_stmt
    quarterly_balance = stock.quarterly_balance_sheet
    quarterly_cashflow = stock.quarterly_cashflow
    
    # Get current price data
    current_price = info.get('currentPrice', info.get('regularMarketPrice', 0))
    market_cap = info.get('marketCap', 0)
    
    print("\n=== Quarterly Income Statement ===")
    if not quarterly_income.empty:
        print(quarterly_income.iloc[:, 0:2])  # Show latest 2 quarters
        
    print("\n=== Quarterly Balance Sheet ===")
    if not quarterly_balance.empty:
        print(quarterly_balance.iloc[:, 0:2])  # Show latest 2 quarters
        
    print("\n=== Quarterly Cash Flow ===")
    if not quarterly_cashflow.empty:
        print(quarterly_cashflow.iloc[:, 0:2])  # Show latest 2 quarters
    
    # Extract specific metrics for comparison
    yfinance_data = {
        'info': info,
        'quarterly_income': quarterly_income.to_dict() if not quarterly_income.empty else {},
        'quarterly_balance': quarterly_balance.to_dict() if not quarterly_balance.empty else {},
        'quarterly_cashflow': quarterly_cashflow.to_dict() if not quarterly_cashflow.empty else {},
        'current_price': current_price,
        'market_cap': market_cap
    }
    
    return yfinance_data

def get_stock_oracle_data(ticker: str, start_date: str, end_date: str) -> Dict[str, Any]:
    """Get financial data from Stock Oracle API"""
    print(f"\n=== Getting Stock Oracle data for {ticker} ===")
    
    # Request data from API
    response = requests.post(
        f"{API_URL}/financial/data",
        json={
            "ticker": ticker,
            "start_date": start_date,
            "end_date": end_date,
            "period_type": "all",
            "include_metrics": True,
            "force_refresh": False
        }
    )
    
    if response.status_code == 200:
        data = response.json()
        print(f"Company: {data['company']['name']}")
        print(f"Sector: {data['company']['sector']}")
        print(f"Industry: {data['company']['industry']}")
        print(f"Number of periods: {len(data['financial_data'])}")
        return data
    else:
        print(f"Error: {response.status_code} - {response.text}")
        return None

def compare_financial_metrics(yfinance_data: Dict, oracle_data: Dict):
    """Compare financial metrics between yfinance and Stock Oracle"""
    print("\n=== COMPARISON RESULTS ===")
    
    if not oracle_data or 'financial_data' not in oracle_data:
        print("No Stock Oracle data to compare")
        return
    
    # Get the latest quarter from Stock Oracle
    latest_oracle = oracle_data['financial_data'][-1] if oracle_data['financial_data'] else None
    
    if not latest_oracle:
        print("No financial data in Stock Oracle")
        return
    
    print(f"\nStock Oracle Latest Period: {latest_oracle['period_date']}")
    
    # Compare basic metrics
    print("\n--- Revenue Comparison ---")
    oracle_revenue = latest_oracle.get('revenue', 0)
    print(f"Stock Oracle Revenue: ${oracle_revenue:,.2f}")
    
    # Try to get revenue from yfinance
    if 'quarterly_income' in yfinance_data and yfinance_data['quarterly_income']:
        quarters = list(yfinance_data['quarterly_income'].keys())
        if quarters:
            latest_quarter = quarters[0]
            yf_revenue = yfinance_data['quarterly_income'][latest_quarter].get('Total Revenue', 0)
            print(f"YFinance Revenue: ${yf_revenue:,.2f}")
            if oracle_revenue > 0:
                diff_pct = ((oracle_revenue - yf_revenue) / yf_revenue) * 100
                print(f"Difference: {diff_pct:.2f}%")
    
    print("\n--- Market Cap Comparison ---")
    oracle_market_cap = latest_oracle.get('market_cap', 0)
    yf_market_cap = yfinance_data.get('market_cap', 0)
    print(f"Stock Oracle Market Cap: ${oracle_market_cap:,.2f}")
    print(f"YFinance Market Cap: ${yf_market_cap:,.2f}")
    if oracle_market_cap > 0 and yf_market_cap > 0:
        diff_pct = ((oracle_market_cap - yf_market_cap) / yf_market_cap) * 100
        print(f"Difference: {diff_pct:.2f}%")
    
    print("\n--- P/E Ratio Comparison ---")
    oracle_pe = latest_oracle.get('pe_ratio', 0)
    yf_pe = yfinance_data['info'].get('trailingPE', 0)
    print(f"Stock Oracle P/E: {oracle_pe:.2f}")
    print(f"YFinance P/E: {yf_pe:.2f}")
    if oracle_pe > 0 and yf_pe > 0:
        diff_pct = ((oracle_pe - yf_pe) / yf_pe) * 100
        print(f"Difference: {diff_pct:.2f}%")
    
    print("\n--- Other Metrics from Stock Oracle ---")
    print(f"EPS: ${latest_oracle.get('eps', 0):.2f}")
    print(f"Total Assets: ${latest_oracle.get('total_assets', 0):,.2f}")
    print(f"Total Equity: ${latest_oracle.get('total_equity', 0):,.2f}")
    print(f"ROE: {latest_oracle.get('roe', 0):.2%}")
    print(f"Net Margin: {latest_oracle.get('net_margin', 0):.2%}")
    
    # Check if we have real yfinance quarterly data
    if 'quarterly_income' in yfinance_data and yfinance_data['quarterly_income']:
        print("\n--- Available YFinance Metrics ---")
        quarters = list(yfinance_data['quarterly_income'].keys())
        if quarters:
            latest_quarter = quarters[0]
            print(f"Latest YFinance Quarter: {latest_quarter}")
            
            # Show some key metrics from yfinance
            income_data = yfinance_data['quarterly_income'][latest_quarter]
            for key in ['Total Revenue', 'Net Income', 'Operating Income', 'Gross Profit']:
                if key in income_data:
                    print(f"{key}: ${income_data[key]:,.2f}")

def main():
    """Main comparison function"""
    # Test with Apple
    ticker = "AAPL"
    start_date = "2024-01-01"
    end_date = "2024-12-31"
    
    print(f"Comparing financial data for {ticker}")
    print("=" * 60)
    
    # Get data from both sources
    yfinance_data = get_yfinance_data(ticker)
    oracle_data = get_stock_oracle_data(ticker, start_date, end_date)
    
    # Compare the data
    compare_financial_metrics(yfinance_data, oracle_data)
    
    # Test with another ticker
    print("\n" + "=" * 60)
    ticker = "MSFT"
    print(f"\nComparing financial data for {ticker}")
    print("=" * 60)
    
    yfinance_data = get_yfinance_data(ticker)
    oracle_data = get_stock_oracle_data(ticker, start_date, end_date)
    compare_financial_metrics(yfinance_data, oracle_data)

if __name__ == "__main__":
    main()