calibrate

Review and update DCF assumptions for $ARGUMENTS.

Python Environment

When running Python code, set PYTHONPATH so stock_analyzer is importable:

PYTHONPATH="${CLAUDE_PLUGIN_ROOT:-.}" python3 -c "from stock_analyzer import ..."

Canonical Run (reference)

Minimal, copy-pasteable call chain that bypasses the three common pitfalls (staticmethod on FinancialMetrics, dcf_inputs vs raw cached shape, zero-arg get_summary()):

from stock_analyzer import StockManager, FinancialMetrics, DCFModel

manager = StockManager()
cached = manager.load_financial_data("$ARGUMENTS")          # raw cached JSON payload
assumptions, _ = manager.get_or_create_assumptions("$ARGUMENTS")

dcf_inputs = FinancialMetrics.calculate_dcf_inputs(         # staticmethod, NO instance
    income_statement=cached["data"]["income_statement_annual"]["reports"],
    balance_sheet=cached["data"]["balance_sheet"]["reports"],
    cash_flow=cached["data"]["cash_flow"]["reports"],
    overview=cached["data"]["overview"],
    income_annual=cached["data"]["income_statement_annual"]["reports"],  # enables R&D-adjusted metrics
)

shares = float(cached["data"]["overview"]["SharesOutstanding"])
price  = float(cached["data"]["quote"]["Global Quote"]["05. price"])

model = DCFModel(assumptions)
model.calculate_fair_value(dcf_inputs, shares, price, verbose=True)  # dcf_inputs, NOT cached
print(model.get_summary())                                           # zero args
implied = model.reverse_dcf(dcf_inputs, shares, price)               # also dcf_inputs

Gotchas:

• FinancialMetrics.calculate_dcf_inputs(...) is a staticmethod. The class has no useful instance — do NOT write FinancialMetrics(data).calculate_dcf_inputs(...).
• DCFModel.calculate_fair_value and DCFModel.reverse_dcf both take the dcf_inputs dict, not the raw cached JSON. Passing raw cached data raises ValueError: dcf_inputs missing required fields: revenue, operating_income, market_cap.
• DCFModel.get_summary() takes zero args — it reads from self.results set by the prior calculate_fair_value() call.
• Credit-spread lookup is stock_analyzer.metrics.get_spread_for_rating(rating) → Optional[float]; there is no stock_analyzer.credit_spreads module. Returns None for unknown ratings (including compound strings like "Aaa/AAA" — split to "Aaa" or "AAA" first), so guard before arithmetic or rf + spread raises TypeError: unsupported operand type(s) for +: 'float' and 'NoneType'.