Skip to content

Institutional Holdings: Analyze Hedge Fund Portfolios with Python

Every quarter, institutional investment managers with over $100 million in assets must disclose their equity holdings to the SEC via 13F filings. EdgarTools parses these filings into structured Python objects, giving you a window into what the biggest funds are buying and selling -- in a few lines of code.

from edgar import Company

berkshire = Company("BRK.A")
report = berkshire.get_filings(form="13F-HR")[0].obj()
report

13F institutional holdings report for Berkshire Hathaway parsed with Python edgartools

One call to filing.obj() and you get a fully parsed report: the investment manager, total portfolio value, every position, and the subsidiary managers involved.

Find 13F Filings

Search by Fund

If you know the fund's ticker or name, start with Company:

from edgar import Company

# By ticker
berkshire = Company("BRK.A")
filings = berkshire.get_filings(form="13F-HR")

# Latest filing
report = filings[0].obj()
print(report.management_company_name)  # "Berkshire Hathaway Inc"
print(f"${report.total_value:,} across {report.total_holdings} holdings")

Search Across All Filers

To browse 13F filings market-wide, use get_filings:

from edgar import get_filings

# All 13F holdings reports filed in Q3 2024
filings = get_filings(form="13F-HR", year=2024, quarter=3)

# Filter to a specific date range
filings = get_filings(
    form="13F-HR",
    filing_date="2024-11-01:2024-11-15"
)

Access Holdings Data

The .holdings property returns a pandas DataFrame with one row per security, sorted by value:

report.holdings_view()

Berkshire Hathaway 13F holdings table parsed with Python edgartools

Every position includes the issuer name, resolved ticker symbol, CUSIP, market value, share count, and voting authority breakdown.

Key Columns

Column What it is
Issuer Company name ("APPLE INC")
Ticker Resolved ticker symbol ("AAPL")
Value Market value in thousands of dollars
SharesPrnAmount Share count or principal amount
Cusip 9-character CUSIP
Type "Shares" or "Principal"
PutCall "PUT", "CALL", or empty

Values are in thousands -- the SEC's reporting unit. A Value of 60,656,116 means approximately $60.7 billion.

Work with the DataFrame Directly

h = report.holdings

# Top 5 positions
h.head(5)

# Filter by ticker
apple = h[h['Ticker'] == 'AAPL']

# Total portfolio value in dollars
total = h['Value'].sum() * 1000

Compare Holdings Quarter-over-Quarter

One call to see what changed since last quarter:

report.compare_holdings()

13F holdings quarter-over-quarter comparison for Berkshire Hathaway in Python edgartools

Every position gets a status: NEW, CLOSED, INCREASED, DECREASED, or UNCHANGED. Results are sorted by absolute value change so the biggest moves appear first.

Dig into the Comparison Data

comparison = report.compare_holdings()
df = comparison.data

# New buys this quarter
new_buys = df[df['Status'] == 'NEW']

# Positions that were sold entirely
exits = df[df['Status'] == 'CLOSED']

# Positions where shares increased
increases = df[df['Status'] == 'INCREASED']

The comparison DataFrame includes Shares, PrevShares, ShareChange, ShareChangePct, Value, PrevValue, ValueChange, ValueChangePct, and Status.

See how positions evolve over multiple quarters with sparkline visualizations:

history = report.holding_history(periods=4)
history

Each row shows share counts per quarter and a Unicode sparkline so you can spot trends at a glance. Access the underlying data with history.data.

Analyze Portfolio Concentration

h = report.holdings
total = h['Value'].sum()
h['Weight'] = (h['Value'] / total * 100).round(2)

# Top 10 holdings by portfolio weight
h[['Ticker', 'Issuer', 'Value', 'Weight']].head(10)

Find Options Positions

options = report.holdings.query("PutCall in ['PUT', 'CALL']")
puts = options.query("PutCall == 'PUT'")
calls = options.query("PutCall == 'CALL'")

Access the Previous Quarter

previous = report.previous_holding_report()  # Returns a ThirteenF or None
if previous:
    print(f"Previous period: {previous.report_period}")
    print(f"Previous holdings: {previous.total_holdings}")

Understand Multi-Manager Filings

Large institutions like Berkshire Hathaway file consolidated 13F reports covering multiple subsidiary managers. The .holdings property automatically aggregates across all managers. If you need per-manager detail, use .infotable instead:

report.infotable    # Disaggregated: one row per manager-security pair
report.holdings     # Aggregated: one row per security (recommended)

# See who the other managers are
for mgr in report.other_managers:
    print(f"{mgr.name} (File: {mgr.file_number})")

Use View Objects in Your Own App

holdings_view(), compare_holdings(), and holding_history() all return view objects. They render beautifully in the terminal via Rich, but also support iteration, indexing, and direct DataFrame access:

view = report.holdings_view()

# Iterate rows as dicts
for row in view:
    print(row['Ticker'], row['Value'])

# Index a single row
view[0]

# Slice (returns DataFrame)
view[:10]

# Access the full DataFrame
view.data

Things to Know

Values are in thousands. The SEC requires 13F values in $000s. A Value of 135,364 means $135.4 million.

holdings vs infotable. Use holdings (aggregated by CUSIP) for portfolio analysis. Use infotable only when you need per-manager detail in multi-manager filings.

Ticker resolution. Tickers are resolved from CUSIPs. Most resolve correctly, but delisted or obscure securities may show as blank.

13F-NT means no holdings. Notice filings indicate the manager had nothing to report. Check with report.has_infotable().

Report period vs filing date. The report_period is the quarter end (e.g., 2025-09-30). The filing_date can be up to 45 days later.

Historical filings work too. 13F filings before Q3 2013 used fixed-width TXT format instead of XML. EdgarTools parses both transparently -- the same filing.obj() call works for filings back to 2005. The TXT parser handles spaced CUSIPs, multi-line company names, paginated tables, and multi-manager continuation rows. About 93% of TXT-era filings parse successfully; a small number with unusual formatting may return an empty infotable.