Valuation and Scenario Analysis
Valuation is the question every investment ultimately turns on: what is this actually worth, and is the current price a good deal? It's also where false precision is most seductive — a model can output '$147.32 per share' and make a pile of guesses look like a fact. This module covers the main valuation approaches, how AI helps you think through them, and the intellectual honesty that separates real valuation from elaborate guessing.
The core valuation approaches
Valuation is not the search for a single correct number — it's the construction of a defensible range under explicit assumptions. Anyone who gives you a precise valuation to the penny without showing the assumptions and the range is either naive or selling something. The honest output of valuation is "somewhere between X and Y, depending on these key drivers, with this scenario most likely."
How AI genuinely helps with valuation
Selecting the right approach
"I'm valuing a pre-profit high-growth software company. Which valuation approaches make sense and which don't, and why?" AI is excellent at matching method to situation — for instance, explaining why a DCF is shaky for a company with no stable cash flows and why revenue multiples might dominate.
Building the comp set
"What publicly traded companies are the best comparables for [Company], and what makes them comparable or not?" Combined with a live-search tool to pull current multiples, this accelerates the most tedious part of comps analysis. Verify the multiples against a real data source.
Pressure-testing the DCF assumptions
This is where AI earns its keep. "My DCF assumes 20% revenue growth tapering to 4% terminal growth, 25% operating margins, and a 10% discount rate. Walk through whether each assumption is reasonable and where I'm most likely being too optimistic." AI as a skeptical reviewer of your assumptions is genuinely valuable — especially Claude, which is trained to push back rather than flatter.
Structuring the scenario analysis
"Help me design bull, base, and bear scenarios for this valuation, with the key assumptions that differ in each." Then you run the actual numbers in a spreadsheet. The output is a range with probabilities — honest valuation.
An LLM will happily produce "$147.32 per share" with total confidence. That number is the product of a dozen assumptions, each uncertain, compounded through a model. The decimal places are theater. Always force the analysis back to a range and the assumptions driving it. A valuation is only as solid as its weakest assumption, and there are always several weak ones.
The DCF sensitivity reality
A DCF's output is extraordinarily sensitive to two assumptions: the discount rate and the terminal growth rate. A one-percentage-point change in either can move the valuation 20-30%. This is why a single-point DCF output is nearly meaningless and a sensitivity table is essential.
Putting it together: an AI-assisted valuation
(1) Use AI to select appropriate methods for the situation. (2) Use live search to gather current comps and market data, verified against primary sources. (3) Build the actual model in a spreadsheet — real computation. (4) Use AI as a skeptical reviewer of every key assumption. (5) Run bull/base/bear scenarios with a sensitivity table. (6) Use AI to write up the thesis honestly, including what would have to be true for the valuation to hold and what would break it. The output is a defensible range, not a false-precise number.
The best thing AI-assisted valuation can do is force intellectual honesty: make every assumption explicit, test how much each one matters, and present a range rather than a false-precise answer. If your AI workflow makes you more rigorous about assumptions rather than letting you skip them, you're using it right. If it lets you generate confident-looking numbers faster without thinking harder, you're using it dangerously.
Next
Module 8 confronts the part of finance where AI carries the most risk: regulation, compliance, and fiduciary responsibility. Using AI carelessly here isn't just a quality problem — it can be a legal and ethical one.