valuation-skill

name	valuation-skill
description	Performs end-to-end intrinsic valuation for a company. Extracts financial data from US (10-K/10-Q) and Indian (Annual/Quarterly reports) financial statements using scrapers or XBRL downloaders. Builds a bottom-up cost of capital (WACC), runs a base case DCF model using historical growth, performs a Monte Carlo simulation, and generates an HTML report. Use this skill when the user asks to value a company, extract valuation inputs, or run a discounted cash flow (DCF) model.
license	MIT
compatibility	opencode
metadata	{"audience":"developers","workflow":"valuation"}

Valuation End-to-End Workflow

## Step 1: Document Retrieval Before extracting inputs, retrieve the necessary financial statements: - **For US Companies:** Download the latest 10-K and the latest 10-Q using the appropriate scraper. - **For Indian Companies (Primary Focus):** Use the NSE scraper or the XBRL downloader to download: 1. The latest Annual Report. 2. The most recent Quarterly Report. 3. The *previous* Quarterly Report as well. *Note on Indian Companies:* Balance sheet estimates are typically only published twice a year (usually Q2 and Q4). If the most recent quarterly report lacks a balance sheet, you must refer to the previous quarterly report that contains one. *Tooling:* You may use the included `scripts/download_xbrl.py` script. If it requires authentication, ask the user to provide the XBRL cookie.

Step 2: Input Extraction

• Do NOT use any automated AI pipelines (like ai_pipeline.py) to extract or guess inputs.
• You must manually figure out the inputs yourself by analyzing the retrieved financial statements.
• Follow the detailed extraction rules defined in references/extraction_rules.md to parse the retrieved documents.
• Reference references/business_list.md to properly map businesses and R&D amortization periods.
• The final output must conform strictly to the schema defined in assets/output_schema.json.

Step 3: Valuation Execution (Base Case)

Once inputs are extracted and formatted, you MUST ask the user for the following key valuation assumptions using the question tool before proceeding:

1. Revenue growth rate for next year
2. Operating Margin for next year
3. Compounded annual revenue growth rate - years 2-5
4. Target pre-tax operating margin
5. Year of convergence for margin

After receiving these inputs from the user:

• Run the valuation model located inside this repository, or utilize the provided scripts/calculate_valuation.py.
• Cost of Capital (WACC): Must be calculated using the bottom-up philosophy utilizing the scripts/wacc.py module. This module uses Damodaran's CSV lookups located in references/ (e.g., industry_averages_us.csv, country_equity_risk_premiums.csv, industry_averages_global.csv) to find the industry unlevered beta, weighted equity risk premium, and synthetic rating for the cost of debt.
• Growth and Margin Estimates: Use the user-provided growth and margin estimates to drive the base case scenario instead of relying solely on historical trailing averages.

Step 4: Monte Carlo Simulation

After establishing the base case:

• Run a Monte Carlo simulation.
• Vary the growth rates and operating margins to generate a distribution of possible intrinsic values.

Step 5: Reporting

• Create a final HTML report detailing the intrinsic valuation results.
• The report must include the base case valuation, the simulation distributions, the computed cost of capital (bottom-up build), and all the primary inputs used.

Step 6: Stories to Numbers (Valuation as a Picture)

• You must create a visual representation of the valuation that maps narrative "stories" to the quantitative assumptions in the model.
• Include a comprehensive HTML table structure in the final report that mirrors the "Valuation as a Picture" framework.
• The top section of this table must include narrative descriptions explicitly linking to the assumptions for:
  • Growth Story: Linked to revenue growth rates.
  • Profitability Story: Linked to operating margins.
  • Growth Efficiency Story: Linked to the Sales-to-Capital ratio.
  • Risk Story & Competitive Advantages: Linked to the Cost of Capital and Terminal Value assumptions.
• The middle section must show a 10-year time-series (plus a terminal year column) displaying the year-by-year outputs for: Revenue Growth, Revenue, Operating Margin, Operating Income, EBIT(1-t), Reinvestment, FCFF, Cost of Capital, and Cumulated WACC.
• The left sidebar must display the final value bridge: PV of Terminal Value, PV of Cash Flows, Probability of Failure, Value of Operating Assets, less Debt/Minority Interests, plus Cash/Non-Operating Assets, to arrive at the Value of Equity and Estimated Value per Share.

Step 7: Valuation Diagnostic

• Use the Task Tool to spawn a general sub-agent to perform a rigorous diagnostic of the valuation outputs.
• Pass the base outputs (Current Price, Estimated Value, Growth Rate, Operating Margin, Sales to Capital, ROC, Cost of Capital) and ask the sub-agent to evaluate the model using the following 6-step framework:
  1. Check revenue growth rate: Compare forecasted growth against industry averages and recent history. Is it realistic?
  2. Check dollar revenues: Analyze total market size and the implied market share by Year 10.
  3. Check your margins: Compare the target margin to industry averages and historical unit economics.
  4. Check how much you are reinvesting: Evaluate the Sales-to-Capital ratio and whether the terminal ROC makes sense. Is reinvestment consistent with forecasted growth?
  5. Risk Metrics: Compare the Cost of Capital to the industry median. Assess the transition of the discount rate over time and the failure probability.
  6. Price versus Value: Analyze the delta between calculated value and current price. Identify which inputs (Growth, Margin, Sales/Capital, Cost of Capital) might need adjustment if the value seems too high or too low.
• Append the sub-agent's qualitative diagnostic report into the final HTML report under a "Valuation Diagnostic Report" section to provide a narrative critique of the model's assumptions.

<system_instructions> You are an expert financial analyst. Your task is to extract specific financial metrics from the provided 10-K and 10-Q documents strictly based on the defined accounting rules in references/extraction_rules.md. </system_instructions> <input_context> Input Context See attached recent annual report. See attached recent quarterly report. </input_context>

<execution_instructions> Execution Instructions Read the footnotes to verify terms, accumulated NOLs, debt maturity profiles, and R&D figures. Do not output conversational text. Output must strictly conform to the JSON schema defined in assets/output_schema.json. </execution_instructions>