Nov 9, 2026 · 1h 2m
Kevin Schoovaerts builds an AI investing system without surrendering judgment
The AI Masterclass for Microcap Investors | Kevin Schoovaerts
The episode examines how microcap investors can automate research and monitoring while preserving the human responsibility behind every decision.
- 1AI workflows become more reliable when high-quality data, reusable skills, Python, and focused context work together.
- 2Specialized agents can generate ideas, research companies, challenge theses, and track holdings across a shared knowledge vault.
- 3Automation cannot supply conviction or accountability; investors still need fundamentals, explicit exit rules, and independent judgment.
Don't miss
Schoovaerts describes how thesis tracking challenged his reluctance to sell by testing a holding against explicit exit rules.
The brief
Kevin Schoovaerts, a microcap investor with a physics background, describes a pipeline that moves from idea generation to research, contrarian testing, and thesis tracking.
The system combines Claude, financial data, filings, Python, and reusable skills; its value depends less on clever prompts than on reliable inputs and captured process.
Specialized agents triage companies, produce research, and monitor holdings against their original theses, flagging positions as intact, pressured, or broken.
A shared knowledge vault gives the system memory, allowing it to surface past mistakes and challenge an investor’s reluctance to sell when exit rules are triggered.
Schoovaerts returns to Richard Feynman’s warning about fooling oneself: AI can accelerate learning and push back, but conviction and accountability remain human work.
What was said on this episode
17 statements · 11 positive · 4 negative · 2 neutral
Early ChatGPT produced polished but unreliable equity analysis
“for equity analysis, the conclusion was it's a bit beautifully written nonsense.”
Listen at 8:06
- Kevin Schoovaertson LLM outputsNegative9:27
Investors should not accept LLM outputs without independent judgment
“you can never take a result at point-blank range.”
Listen at 9:27
- Kevin Schoovaertson AI investing toolsPositive10:33
AI investing tools remain early enough for investors to learn
“we're still very early, so it's never too late to learn all these things”
Listen at 10:33
- Kevin Schoovaertson LLM PDF extractionNegative20:12
LLM extraction from PDF tables can introduce errors
“if you give the LLM a PDF and you just think, okay, it'll extract the data correctly, there can be errors in it”
Listen at 20:12
Anthropic's financial skill can produce initiating coverage reports
“it produces an initiating coverage report”
Listen at 21:22
- Kevin Schoovaertson AI-generated research reportsPositive22:05
Independent quality control can screen AI reports for reliability issues
“we added some sort of an independent quality control”
Listen at 22:05
AI agents can analyze 100–200 companies simultaneously
“with an agent, you can do sector analysis, so you can like analyze 100 or 200 companies in one go”
Listen at 25:36
- Kevin Schoovaertson SaaS companiesPositive28:05
AI can rank 100 SaaS companies for AI resilience
“you can rank 100 companies and then dig deeper into, for example, the top ones”
Listen at 28:05
- Kevin Schoovaertson AI scoring and stock performanceNeutral28:54
The observed AI scores and stock performance do not establish true correlation
“I'm not going to say that there is a true correlation between the AI scoring and the performance itself.”
Listen at 28:54
- Kevin Schoovaertson EODHDPositive42:12
EODHD currently offers Kevin the broadest data coverage
“For the moment, for me, they have the broadest set of data.”
Listen at 42:12
- Kevin Schoovaertson Investment dataPositive43:01
Investment analysis requires high-quality and recent data
“You need to use quality data, but it also needs to be the most recent as possible”
Listen at 43:01
- Kevin Schoovaertson Friend's portfolio exposure to AI CapEx spendingNegative45:03
A friend's portfolio was nearly 60% exposed to AI capital spending
“almost 60% of his portfolio is exposed to AI CapEx spending”
Listen at 45:03
- Kevin Schoovaertson Claude and Yahoo FinancePositive46:47
Claude can usually obtain a near-current price through Yahoo Finance
“normally if you ask it to use Yahoo Finance, it should be able to get you a price that is closest to reality”
Listen at 46:47
Opus 5.5 produced a checked valuation model without errors
“it did it in one go and I checked the quality and there were no errors in it”
Listen at 53:46
- Kevin Schoovaertson DCF valuationNeutral54:18
Investors without DCF experience must learn the underlying valuation steps
“If you've never created a DCF before, you will have to go through the steps.”
Listen at 54:18
- Kevin Schoovaertson AI learning toolsPositive57:39
AI learning tools can accelerate learning accounting and company fundamentals
“he learns it a lot faster than I did in the past because he's using all these tools to learn”
Listen at 57:39
- Kevin Schoovaertson Automated investment pipelinesPositive57:47
Learning fundamentals should precede building automated investment pipelines
“that always should be the first step”
Listen at 57:47
Statements are attributed to the speaker as said on the episode and reflect their view at the time, not PodLume's. They are not advice.
Weekly email
State of the Pod, every Monday
The week's most-discussed names and stories from podcasts, with the moments behind them. Free, five minutes, unsubscribe in one click.