sec-edgar-skill

SEC EDGAR 技能 - 文件分析

前置条件

重要：在进行任何 EdgarTools 操作之前，务必运行此设置：

from edgar import set_identity
set_identity("Your Name [email protected]")  # SEC requires identification

这是 SEC 法律要求。未进行设置，操作将会失败。

---

安装

必须安装 EdgarTools：

pip install edgartools

---

Token 效率策略

始终优先使用 .to_context() —— 它提供的摘要可减少 56%–89% 的 Token 消耗：

对象	repr() Token 数	.to_context() Token 数	节省比例
Company	~750	~75	90%
Filing	~125	~50	60%
XBRL	~2,500	~275	89%
Statement	~1,250	~400	68%

**规则：**先调用 .to_context() 了解可用内容，再深入查看。

---

访问文件的三种方式

1. 已发布文件 - 跨公司批量分析

from edgar import get_filings

# Get recent 10-K filings
filings = get_filings(form="10-K")

# Filter by date range
filings = get_filings(form="10-K", year=2024, quarter=1)

# Multiple form types
filings = get_filings(form=["10-K", "10-Q"])

2. 最新文件 - 实时监控

from edgar import get_current_filings

# Get today's filings from RSS feed
current = get_current_filings()

# Filter by form type
current_10k = get_current_filings().filter(form="10-K")

3. 公司文件 - 单一实体分析

from edgar import Company

# By ticker
company = Company("AAPL")

# By CIK
company = Company("0000320193")

# Get company's filings
filings = company.get_filings(form="10-K")
latest_10k = filings.latest()

---

财务数据访问

方法 1：Entity Facts API（快速、多期间）

适用于比较各期间的趋势：

company = Company("AAPL")

# Get income statement for multiple periods
income = company.income_statement(periods=5)
print(income)  # Shows 5 years of data

# Get balance sheet
balance = company.balance_sheet(periods=3)

# Get cash flow
cashflow = company.cash_flow_statement(periods=3)

方法 2：文件 XBRL（详细、单一期间）

适用于全面分析单份文件：

company = Company("AAPL")
filing = company.get_filings(form="10-K").latest()

# Get XBRL data
xbrl = filing.xbrl()

# Access financial statements
statements = xbrl.statements
income_stmt = statements.income_statement
balance_sheet = statements.balance_sheet
cash_flow = statements.cash_flow_statement

---

常见工作流

工作流 1：比较不同公司的收入

from edgar import Company

companies = ["AAPL", "MSFT", "GOOGL"]
for ticker in companies:
    company = Company(ticker)
    income = company.income_statement(periods=3)
    print(f"\n{ticker} Revenue Trend:")
    print(income)

工作流 2：分析最新的 10-K

from edgar import Company

company = Company("NVDA")
filing = company.get_filings(form="10-K").latest()

# Get filing metadata
print(filing.to_context())

# Get full text (expensive - 50K+ tokens)
# text = filing.text()

# Get specific sections
# items = filing.items()  # Risk factors, MD&A, etc.

工作流 3：追踪内部人士交易

from edgar import Company

company = Company("TSLA")
insider_filings = company.get_filings(form="4")  # Form 4 = insider trades

for filing in insider_filings[:10]:
    print(filing.to_context())

工作流 4：监控各行业近期提交的文件

from edgar import get_filings

# Get recent tech 10-Ks (use SIC codes)
# SIC 7370-7379 = Computer Programming, Data Processing
filings = get_filings(form="10-K", year=2024)
# Filter by company characteristics after retrieval

工作流 5：多年财务趋势

from edgar import Company

company = Company("AMZN")

# 5-year income statement
income = company.income_statement(periods=20)  # 20 quarters = 5 years

# 5-year balance sheet
balance = company.balance_sheet(periods=20)

print("Income Statement Trend:")
print(income)
print("\nBalance Sheet Trend:")
print(balance)

---

在文件中搜索

重要区别：

filing = company.get_filings(form="10-K").latest()

# Search WITHIN the filing document (finds text in the 10-K)
results = filing.search("climate risk")

# Search API DOCUMENTATION (finds how to use EdgarTools)
docs_results = filing.docs.search("how to extract")

请勿混淆！

---

关键对象参考

Company

company = Company("AAPL")
company.to_context()  # Summary with available actions
company.name          # Company name
company.cik           # CIK number
company.sic           # SIC code
company.industry      # Industry description
company.get_filings() # Access filings

Filing

filing.to_context()   # Summary
filing.form           # Form type (10-K, 10-Q, etc.)
filing.filing_date    # Date filed
filing.accession_number
filing.text()         # Full document text (EXPENSIVE)
filing.markdown()     # Markdown format
filing.xbrl()         # XBRL financial data
filing.items()        # Document sections

XBRL（财务数据）

xbrl = filing.xbrl()
xbrl.to_context()     # Summary
xbrl.statements       # All financial statements
xbrl.facts            # Individual facts/metrics

Statement（财务报表）

stmt = xbrl.statements.income_statement
print(stmt)           # ASCII table format
stmt.to_dataframe()   # Pandas DataFrame

---

反模式（请避免）

不要：从原始文本中解析财务数据

# BAD - expensive and error-prone
text = filing.text()
# try to regex parse revenue from text...

应当：使用结构化 XBRL 数据

# GOOD - structured and accurate
income = company.income_statement(periods=3)

不要：只需元数据时仍加载整份文件

# BAD - wastes tokens
text = filing.text()  # 50K+ tokens

应当：先使用上下文

# GOOD - minimal tokens
print(filing.to_context())  # ~50 tokens

---

表格类型速查

表格	描述	用途
10-K	年度报告	全年财务数据、业务描述
10-Q	季度报告	季度财务数据
8-K	当前报告	重大事件（并购、管理层变动）
DEF 14A	委托书	高管薪酬、董事会信息
4	内部人士交易	内部人士的股票交易
13F	机构持仓	对冲基金持有哪些资产
S-1	IPO 注册文件	IPO 前提交的文件
424B	招股说明书	债券／股票发行

---

错误处理

from edgar import Company

try:
    company = Company("INVALID")
except Exception as e:
    print(f"Company not found: {e}")

# Check if filings exist
filings = company.get_filings(form="10-K")
if len(filings) == 0:
    print("No 10-K filings found")

---

性能提示

1. 检索前先筛选：使用表格类型和日期筛选条件
2. 趋势分析使用 Entity Facts API：比解析多份文件更快
3. 批量操作：通过循环处理多家公司
4. 缓存结果：存储经常访问的数据

---

参考文档

详细文档请参阅：

• EdgarTools 工作流
• 对象参考
• 表格类型参考

也可以使用内置文档：

from edgar import Company
company = Company("AAPL")
company.docs.search("how to get revenue")