Skip to content

Form 497K: Parse Mutual Fund Summary Prospectus Filings with Python

Form 497K is the SEC-mandated summary prospectus for mutual funds and ETFs. EdgarTools parses these filings into structured Python objects, giving you fee tables, expense examples, and performance returns without touching any XBRL (there isn't any). With about 20,000 filings per year, the HTML structure is tightly standardized under Form N-1A rules, making extraction reliable across fund families.

from edgar import find

filing = find("0001683863-25-002784")  # Vanguard California Long-Term Tax-Exempt Fund
p = filing.obj()
p

Vanguard 497K summary prospectus parsed with Python edgartools

You get the fund name, series ID, tickers, portfolio turnover, fees, expense examples, and average annual returns in one shot.

Access Fee Data

The .fees property returns a DataFrame with one row per share class, showing the breakdown of annual operating expenses:

p.fees

497K annual operating expenses fee table

Column What it is
class_name Share class name ("Admiral Shares")
ticker Ticker symbol ("VCLAX")
management_fee Management fee as a percentage
twelve_b1_fee 12b-1 distribution fee
other_expenses Other operating expenses
total_annual_expenses Gross total before any waivers
fee_waiver Fee waiver or expense reimbursement
net_expenses Net total after waivers

All fee values are percentages of net assets. A management_fee of 0.08 means 0.08%.

See Expense Examples

The .expense_example property shows the hypothetical cost of holding $10,000 for 1, 3, 5, and 10 years:

p.expense_example
Column What it is
class_name Share class name
ticker Ticker symbol
1yr Cost after 1 year
3yr Cost after 3 years
5yr Cost after 5 years
10yr Cost after 10 years

These are dollar amounts computed by the SEC formula: assume 5% annual return and no change in expense ratios. They are the same hypothetical numbers printed in the paper prospectus.

Read Performance Returns

The .performance property returns average annual total returns across all share classes and any benchmark indices:

p.performance

497K average annual returns with benchmark comparisons

Column What it is
label Return type label (e.g. "Return Before Taxes")
section Share class section header
1yr 1-year return as a percentage
5yr 5-year return as a percentage
10yr 10-year return as a percentage
since_inception Return since inception (if available)

Benchmark rows (like "Bloomberg Municipal Bond Index") appear alongside fund rows and can be filtered by checking whether the label matches a share class ticker or name.

Work with Individual Share Classes

The .share_classes property gives you a list of ShareClassFees objects, one per share class, with all fee data as attributes:

for sc in p.share_classes:
    print(sc.class_name, sc.ticker, f"{sc.total_annual_expenses}%")
# Investor Shares VCITX 0.14%
# Admiral Shares VCLAX 0.09%

Access individual share class attributes directly:

sc = p.share_classes[0]
sc.management_fee        # Decimal('0.13')
sc.twelve_b1_fee         # None (Vanguard has no 12b-1 fees)
sc.total_annual_expenses # Decimal('0.14')
sc.expense_1yr           # 14  (dollars)
sc.max_sales_load        # None (no front-end load)

Compare Fund Families

Using a Fidelity filing from the same form type:

fidelity = find("0000035341-25-000099").obj()
fidelity

Fidelity Equity-Income Fund 497K summary prospectus

The same structure works regardless of fund family because Form N-1A mandates the section layout. Vanguard, Fidelity, BlackRock, and others all produce the same table structure.

Search for 497K Filings by Fund Company

from edgar import Company

# Get recent 497K filings for a fund company
vanguard = Company("VANGUARD CALIFORNIA TAX-FREE FUNDS")
filings = vanguard.get_filings(form="497K")
filings.head(5)

To look up a specific fund series across all its filings:

# Multiple filings from the same company CIK will often cover different fund series
for filing in filings.head(10):
    p = filing.obj()
    print(p.fund_name, p.series_id, p.tickers)

Common Analysis Patterns

Find the cheapest share class

df = p.fees
cheapest = df.loc[df['net_expenses'].fillna(df['total_annual_expenses']).idxmin()]
print(cheapest['class_name'], cheapest['ticker'])

Check for fee waivers

funds_with_waivers = p.fees[p.fees['fee_waiver'].notna()]

Check quarterly performance extremes

# Best and worst quarters are parsed from the bar chart section
print(p.best_quarter)   # (Decimal('8.80'), 'December 31, 2023')
print(p.worst_quarter)  # (Decimal('-6.78'), 'March 31, 2022')

best_pct, best_date = p.best_quarter

Extract investment objective

print(p.investment_objective)
# "The Fund seeks to provide a moderate and sustainable level of current
#  income that is exempt from federal and California personal income taxes..."

Filter performance to fund rows only (exclude benchmarks)

fund_tickers = set(p.tickers)
# Keep rows where the label contains a ticker or class name
perf = p.performance
fund_rows = perf[perf['label'].str.contains('Return')]

Quick Reference: Properties

Property Type Description Example
fund_name str Fund name from header or HTML "Vanguard California Long-Term Tax-Exempt Fund"
tickers list[str] All share class tickers ["VCITX", "VCLAX"]
series_id str Series ID from SGML header "S000002567"
class_ids list[str] Class/Contract IDs ["C000006936", "C000044247"]
cik str Filing entity CIK "783401"
num_share_classes int Number of share classes 2
share_classes list[ShareClassFees] Per-class fee models see ShareClassFees below
best_quarter tuple[Decimal, str] Best quarterly return and period (Decimal('8.80'), 'December 31, 2023')
worst_quarter tuple[Decimal, str] Worst quarterly return and period (Decimal('-6.78'), 'March 31, 2022')
portfolio_turnover Decimal Annual portfolio turnover rate Decimal('81')
investment_objective str Text of the investment objective "The Fund seeks..."
prospectus_date str Date printed on the prospectus "March 28, 2025"
portfolio_managers list[str] Portfolio manager names ["James M. D'Arcy"]

Quick Reference: DataFrame Properties

Property Columns What it contains
fees class_name, ticker, management_fee, twelve_b1_fee, other_expenses, total_annual_expenses, fee_waiver, net_expenses Annual operating expenses per share class
expense_example class_name, ticker, 1yr, 3yr, 5yr, 10yr Hypothetical $10K investment costs
performance label, section, 1yr, 5yr, 10yr, since_inception Average annual total returns

ShareClassFees Fields

Field Type Description
class_name str Share class name
ticker str Ticker symbol
class_id str SEC class/contract ID (C000xxxxx)
management_fee Decimal Management fee percentage
twelve_b1_fee Decimal 12b-1 fee percentage
other_expenses Decimal Other expenses percentage
acquired_fund_fees Decimal Acquired fund fees (for funds of funds)
total_annual_expenses Decimal Gross total expenses
fee_waiver Decimal Fee waiver/expense reimbursement
net_expenses Decimal Net expenses after waiver
expense_1yr int $10K hypothetical cost, 1 year
expense_3yr int $10K hypothetical cost, 3 years
expense_5yr int $10K hypothetical cost, 5 years
expense_10yr int $10K hypothetical cost, 10 years
max_sales_load Decimal Maximum front-end sales load
max_deferred_sales_load Decimal Maximum deferred sales charge
redemption_fee Decimal Redemption fee percentage

Things to Know

No XBRL. 497K filings contain no XBRL data. All structured data comes from HTML table parsing. The SEC mandates the table layout under Form N-1A, which is why parsing is reliable across fund families.

Fee percentages are small numbers. A management_fee of Decimal('0.13') means 0.13%, not 13%. Do not multiply by 100 before displaying.

fee_waiver may be None. Many funds have no active waivers. Check net_expenses first; fall back to total_annual_expenses if net_expenses is None.

Performance rows include benchmarks. The .performance DataFrame mixes fund share class rows with index/benchmark rows. Use the label and section columns to distinguish them.

best_quarter and worst_quarter may be None. Some 497K filings (particularly those for new funds with less than a calendar year of performance) omit the bar chart section, leaving both as None.

Series ID vs CIK. The series_id (e.g., S000002567) identifies the specific fund series within the filing entity. The cik identifies the registrant (fund company). A single registrant CIK can have dozens of series IDs.