Creating Financial Models: Advanced Valuation and Risk Assessment
The creating financial models skill from Anthropic Cookbooks provides AI assistants with comprehensive financial modeling capabilities including DCF analysis, sensitivity testing, Monte Carlo simulations, and scenario planning. Rather than simple ratio calculations, this skill builds complete valuation models using industry-standard methodologies that investment professionals rely on.
What This Skill Does
This skill enables creation of sophisticated financial models for investment analysis, valuation, and risk assessment. It combines multiple analytical techniques—discounted cash flow projections, sensitivity analysis identifying key value drivers, Monte Carlo simulations modeling uncertainty, and scenario planning comparing alternative futures.
Discounted cash flow analysis builds complete models with multiple growth scenarios, calculates terminal values using both perpetuity growth and exit multiple methods, determines weighted average cost of capital, and generates enterprise and equity valuations. This goes far beyond simple present value calculations to create comprehensive valuation frameworks.
Sensitivity analysis tests how changes in assumptions impact valuations, creates data tables for multiple variables, generates tornado charts ranking sensitivity, and identifies critical value drivers. This reveals which assumptions matter most, guiding where to focus due diligence efforts.
Monte Carlo simulation runs thousands of scenarios with probability distributions for uncertain inputs, generating confidence intervals for valuations and calculating probabilities of achieving target returns. This quantifies uncertainty in ways single-point estimates cannot.
Getting Started
The skill requires historical financial statements (typically 3-5 years), revenue growth assumptions, operating margin projections, capital expenditure forecasts, working capital requirements, and discount rate components. For DCF models specifically, you'll need terminal growth rates or exit multiples.
Two core scripts power the functionality: dcf_model.py serves as the complete DCF valuation engine handling projections, terminal value calculations, and discounting. Meanwhile, sensitivity_analysis.py provides the framework for testing assumption impacts and ranking value drivers.
Models automatically perform quality checks including balance sheet balancing verification, cash flow reconciliation, circular reference resolution, sensitivity bound checking, and statistical validation of Monte Carlo results. These catches prevent common modeling errors before they impact analysis.
Key Features
Complete DCF Modeling: Build full discounted cash flow models with multi-year projections, free cash flow calculations, terminal value computation via perpetuity growth or exit multiples, and derivation of both enterprise and equity values. Models include implied valuation multiples for comparison against trading comps.
Sensitivity Analysis: Create data tables showing how value changes across assumption ranges, generate tornado charts identifying most impactful variables, conduct break-even analysis, and visualize relationships between assumptions and outcomes.
Monte Carlo Simulation: Run 1,000 to 10,000 iterations with probability distributions for uncertain variables, model correlations between variables, generate probability distributions of valuations, calculate confidence intervals (90%, 95%), and compute risk metrics like value-at-risk.
Scenario Planning: Define best/base/worst case scenarios with distinct assumptions, model different economic environments, test strategic alternatives, assign probability weights, and calculate probability-weighted expected values.
Multiple Model Types: Support corporate valuations (mature companies, growth companies, turnarounds), project finance (infrastructure, real estate, energy), M&A analysis (acquisition valuations, synergy modeling, accretion/dilution), and LBO models (leveraged buyouts, IRR/MOIC returns, debt capacity).
Usage Examples
When evaluating a technology company acquisition, the skill builds a DCF model from historical financials, projecting revenue growth with product adoption curves, modeling operating margin expansion as scale increases, forecasting capital expenditure declining as percentage of revenue, and determining enterprise value. Sensitivity analysis reveals valuation dependence on terminal growth rate and discount rate assumptions.
For infrastructure project financing, the skill creates project finance models with construction timelines, operating revenue ramps, debt service schedules, and investor return calculations. Monte Carlo simulation models uncertainty in construction costs, demand projections, and operating expenses, generating probability distributions for equity IRRs.
When assessing leveraged buyout opportunities, the skill models acquisition financing structures, projects operational improvements and margin expansion, calculates free cash flow available for debt repayment, determines exit valuations at various multiples, and computes sponsor equity returns under different scenarios.
Best Practices
Document assumptions clearly and separate them from calculations. Good models distinguish inputs (assumptions users modify), calculations (formulas deriving outputs from inputs), and outputs (results presented to decision-makers). This structure enables easy sensitivity testing and scenario comparison.
Use multiple valuation methods for triangulation. Run DCF analysis, compare against trading multiples, check acquisition precedents. When methods converge on similar values, confidence increases. When they diverge significantly, investigate why and determine which method is most reliable given company characteristics.
Apply appropriate risk adjustments rather than artificially conservative assumptions. Don't haircut revenue growth to compensate for uncertainty—model reasonable growth and capture uncertainty through Monte Carlo simulation or scenario probabilities. This produces more rigorous analysis than arbitrary conservatism.
Validate models against market reality. Check whether implied multiples from DCF analysis align with trading comparables. If your model produces P/E of 50x for a mature industrial company, revisit assumptions before concluding the market is wrong.
Consider correlation effects in Monte Carlo simulations. Revenue and margins often correlate positively (strong demand enables pricing power). Construction delays and cost overruns correlate (same factors cause both). Modeling these correlations produces more realistic distributions than assuming independence.
When to Use This Skill
Use this skill when performing comprehensive investment analysis requiring rigorous valuation. Evaluating acquisition targets, assessing investment opportunities, supporting lending decisions, or valuing complex securities all benefit from the modeling sophistication this skill provides.
The skill is particularly valuable when uncertainty is material and single-point estimates insufficient. Real estate developments, technology ventures, project finance—situations with uncertain cash flows and long time horizons justify Monte Carlo simulation and scenario analysis.
It's ideal for professional-quality deliverables. When presenting to investment committees, negotiating acquisitions, or supporting fairness opinions, the modeling standards and documentation quality this skill produces meet professional expectations.
When NOT to Use This Skill
Don't use sophisticated modeling when simpler analysis suffices. Valuing a small acquisition with reliable cash flows doesn't require 10,000 Monte Carlo iterations. Match analytical sophistication to decision importance and uncertainty magnitude.
Avoid over-relying on models when qualitative factors dominate. Financial models quantify assumptions, but they don't determine whether assumptions are reasonable. Management quality, competitive dynamics, regulatory risks—these require judgment beyond what models capture.
It's not appropriate for trading decisions driven by market timing or technical factors. This skill creates fundamental valuation models useful for long-term investment decisions, not short-term price movements or momentum strategies.
Don't expect models to compensate for bad data. Garbage in, garbage out applies forcefully to financial modeling. Unreliable financial statements, inflated projections, or incorrect discount rate components produce misleading valuations regardless of modeling sophistication.
Related Skills
This skill complements analyzing-financial-statements for calculating financial ratios, spreadsheet for manipulating model data, and xlsx for working with Excel-formatted financial models.
Source
This skill is maintained by Anthropic Cookbooks. View on GitHub