Turn annual reports into a rough valuation memo
A poster tried an AI assistant that reads 10-Ks and estimates earnings quality, returns on capital, cash flow, and implied growth.
- What they did
- They set up a custom investing tool to work through annual filings. The system first drew on EDGAR and XBRL Company Facts, then on simpler APIs, and it was asked to build earnings estimates, measure capital returns, approximate owner cash flow, and run a rough reverse DCF. As the thread developed, people pointed out problems with custom tags, note disclosures, and look-ahead bias, and suggested pairing structured figures with text retrieval while keeping source passages and claim history outside the conversation for later checking.
- How it went
- The post did not include any results from a backtest or live use. The process ran into several dead ends: full filing histories were too large for the model window, smaller chunks caused it to skip evidence and invent details, and historical tests looked too strong because filings were restated or arrived later than the data being used. Using retrieved passages, citations, and point-in-time inputs improved trustworthiness, but it still did not solve comparing footnotes across different companies.
- Tips
- Extract each filing once, preserve the supporting text location, and treat later filings as new inputs instead of reprocessing the whole archive. Have a person check anything that depends on disclosure wording, restatements, or timing.
Try it yourself with Claude
Read [company name or ticker]’s latest 10-K from [file or URL]. From that filing, gather the figures and disclosure details needed to estimate earnings power after adjustments, capital returns on the business, and owner cash flow. Show the source line or note for every number you use, then estimate a rough reverse DCF from [current price] at a [discount rate]% discount rate. Stop and ask before any step that would send, submit, delete, or pay for something. End with a short memo covering assumptions, questionable items, and any disclosure quirks that affect comparability.
Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.
Get 3 agent uses like this every morning
With the day's top AI news. Free, and one click to unsubscribe.
Related agent uses
Use your agent to get umbrella coverage estimates
Coverage Cat lets a personal agent collect details, size umbrella protection, and surface carrier quotes through its API and MCP setup.
Use AI to push back on an insurance denial
They had an AI agent examine a homeowners policy and state rules, then used the findings to challenge a water-damage denial.
Research crypto assets and stage trades for review
A custom agent studied crypto markets, reviewed a full portfolio, and prepared swaps and perp orders on Hyperliquid.
Draft appeal letters from denied insurance claims
An automation watches for denied claims, gathers patient details, and drafts a formal appeal for staff to review.
Prepare a flight-delay compensation claim safely
An agent checked a flight’s timing, estimated the payout, and drafted the airline claim without submitting it.
Use an agent to push a warranty claim by email
A user let an agent handle a mattress warranty thread, pressing the company with legal points until the back-and-forth stalled.