Good questions tend to connect. These reading paths move from the mechanics of a decision to the evidence needed to support it.
Money, understood.
Costs, cash flows, and risk. Start with what can be measured before debating what might happen.
Explore the foundations →Useful, inspectable AI.
Build an evidence trail, verify the numbers, and measure the economics of completed work.
Explore the workflows →Where the ideas meet.
Financial research with sharper tools and the same need for sound assumptions.
Explore the intersection →Read the original.
Investor education and original research behind the essays, with notes on why each source matters.
Open the reading list →Financial foundations
- The Compounding Cost of Small Fees — Work through fee drag and compare the full cost of ownership.
- Duration Is a Risk Budget, Not a Forecast — Translate a yield change into a price sensitivity, with the limitations in view.
The connecting idea: turn a familiar percentage into an explicit scenario with units, a time horizon, and assumptions.
Applied AI
- Retrieval Is Not Verification — Why finding a source and supporting a claim are different jobs.
- Let Language Models Read. Let Code Calculate. — Separate extraction, interpretation, and arithmetic.
- The Real Cost of an AI Workflow — Measure accepted work rather than token expenditure alone.
The connecting idea: a system is useful when the whole chain can be tested, including its failures and review requirements.
AI and markets
- The Backtest You Should Not Trust — Audit the information clock before believing the equity curve.
- Valuing AI Without Valuing the Story — Trace a capability to cash flows and an investment price.
- An Evidence-First Workflow for AI Financial Research — Bring the source ledger, calculations, and counterargument together.
The connecting idea: better tools can accelerate research; the investment claim still has to earn its evidence.