Accessing and Analyzing Form 13F Reports: Using Python with FMP API and Selenium for Data Retrieval

				
					import requests
import pandas as pd

# Define the base URL for the API endpoint
url = 'https://financialmodelingprep.com/api/v4/institutional-ownership/portfolio-holdings'

# List of quarter-end dates for 2023
quarter_end_dates = [
    '2022-03-31', '2022-06-30', '2022-09-30', '2022-12-31',
    '2023-03-31', '2023-06-30', '2023-09-30', '2023-12-31',
    '2024-03-31'  
]

# List of CIKs you're interested in
ciks = [
        '0001067983', 
        '0000019617', 
        '0000070858', 
        '0000895421', 
        '0000886982'
        ]

# Initialize an empty DataFrame to append each quarter's data for all CIKs
all_data_df = pd.DataFrame()

# Your API key (replace with your actual API key)
api_key = ''

for cik in ciks:
    for date in quarter_end_dates:
        # Parameters for the GET request
        params = {
            'date': date,  # Use the current quarter-end date
            'cik': cik,  # Use the current CIK
            'page': 0,  # Assuming you're only interested in the first page of results
            'apikey': api_key
        }
        
        # Make the GET request for the current quarter and CIK
        response = requests.get(url, params=params)
        
        if response.status_code == 200:
            # If the request was successful, append the data to the DataFrame
            data = response.json()
            quarter_cik_df = pd.DataFrame(data)
            all_data_df = pd.concat([all_data_df, quarter_cik_df], ignore_index=True)
        else:
            print(f"Failed to retrieve data for CIK {cik} on {date}: {response.status_code}")

# Process the fetched data
all_data_df['date'] = pd.to_datetime(all_data_df['date'])
all_data_df['year'] = all_data_df['date'].dt.year
all_data_df['quarter'] = all_data_df['date'].dt.quarter
transformed_df = all_data_df[['year', 'quarter', 'investorName', 'symbol', 'weight', 'marketValue']].copy()
transformed_df.rename(columns={'symbol': 'ticker', 'marketValue': 'value'}, inplace=True)
transformed_df['value'] = pd.to_numeric(transformed_df['value'])

# Calculate and merge total portfolio values
total_market_values = transformed_df.groupby(['year', 'quarter', 'investorName'])['value'].sum().reset_index(name='total_portfolio_value')
transformed_df = pd.merge(transformed_df, total_market_values, on=['year', 'quarter', 'investorName'])
transformed_df['computed_weight'] = transformed_df['value'] / transformed_df['total_portfolio_value']
transformed_df.drop('weight', axis=1, inplace=True)
transformed_df.rename(columns={'computed_weight': 'weight'}, inplace=True)

# Filter and prepare final_df for plotting
final_df = transformed_df[transformed_df['weight'] > 0.05].copy()
final_df['period'] = final_df['year'].astype(str) + ' Q' + final_df['quarter'].astype(str)
final_df.sort_values(by=['period', 'investorName', 'ticker'], inplace=True)
total_values = final_df.groupby(['period', 'investorName'])['value'].sum().reset_index(name='totalValue')
final_df = pd.merge(final_df, total_values, on=['period', 'investorName'], how='left')
final_df['totalValueText'] = '$' + (final_df['totalValue'] / 1e9).round(2).astype(str) + 'B'

fig, ax = plt.subplots(figsize=(24, 12))

periods = final_df['period'].unique()
investorNames = final_df['investorName'].unique()
tickers = final_df['ticker'].unique()

# Assign a unique color to each ticker
colors = {ticker: plt.cm.tab20(i/len(tickers)) for i, ticker in enumerate(tickers)}

# Position index for each period and investorName
period_indices = {period: i for i, period in enumerate(periods)}
investor_positions = {investorName: i for i, investorName in enumerate(investorNames)}

bar_width = 0.2
spacing = 0.02

for period in periods:
    period_position = period_indices[period]
    
    for investorName in investorNames:
        investor_df = final_df[(final_df['period'] == period) & (final_df['investorName'] == investorName)]
        if not investor_df.empty:
            bottom = 0
            totalValueText = investor_df['totalValueText'].iloc[0]  # Get the total value text for the investor
            
            for _, row in investor_df.iterrows():
                bar_position = period_position + investor_positions[investorName]*(bar_width + spacing)
                weight = row['weight']
                ax.bar(bar_position, weight, bottom=bottom, width=bar_width, color=colors[row['ticker']], edgecolor='white')
                bottom += weight
                
                # Retaining the original segment labels within the bar
                ax.text(bar_position, bottom - (weight / 2), f"{row['ticker']}\n{row['weight']*100:.1f}%\n${row['value']/1e9:.1f}B", ha='center', va='center', fontsize=9, rotation=90)

            # Adding total portfolio value next to the investor name
            label_with_value = f"{investorName} ({totalValueText})"
            ax.text(bar_position, bottom + 0.03, label_with_value, ha='center', va='bottom', fontsize=12, rotation=90, transform=ax.transData)

ax.set_xticks([i + (bar_width + spacing) * len(investorNames) / 2 - bar_width/2 for i in range(len(periods))])
ax.set_ylim(0, ax.get_ylim()[1] * 1.4)
ax.set_xticklabels(periods)
ax.set_ylabel('Weight (%)')
ax.set_title('Quarterly Portfolio Composition for Investors (Assets Exceeding 5% Allocation)', fontsize = 20)

plt.xticks(rotation=45)
plt.tight_layout()
plt.show()