Go from a view of the world to a decision about your money, and know when to change your mind. Each path below starts with one decision and shows what AI can research for you.
01
Investment Framework
Decisions you can trace to your portfolio’s job are easier to review
Turn your view of the world into specific calls, and use AI to research each one.
Start with what the money is for.
A theme is too broad to invest in until you break it down.
Take AI capital spending. Break it into the measurable variables that drive revenue, margins, and valuation, such as rates, demand, and lead times, and ask which of them you can invest in. A thesis names the call, what’s already priced in, and what would prove it wrong. An AI agent in ChatGPT or Claude can research each piece and pull the economic data behind it. It can also lean toward the consensus, so use it to test your view rather than to supply one.
Which allocation fits what you need the money for?
Have AI lay out how much risk each mix carries before you choose one.
Your goals and your time horizon come first.
Allocation shapes the risks and returns you experience.
Your mix of stocks, bonds, and cash decides which risks your portfolio carries. Research on diversified funds found that policy allocation explained much of each fund’s return variation over time, though not every difference between funds. The Investor.gov allocation overview and the Ibbotson–Kaplan study explain the difference.
Let AI compare the mixes, and decide for yourself how big a drop you could sit through.
If you pick stocks, start from screens you’ve tested
Find candidates with screens that hold up, let AI read their filings, and set your price targets before you buy.
Skip this path if you only own funds. Otherwise it covers the screen, the company research, and the cost of trading.
Test the screen before you trust the stock.
Factor research has found traits, such as cheapness and momentum, that went with higher returns in the past, and a ranked screen turns them into a list of candidates. Many of those published factors weaken or fail under stricter tests. And if you rerun a tweaked screen over the same history often enough, one version will look good by chance.
This path shows you how to tell the difference before you put money on it.
Use your own return, volatility, and correlation forecasts to compare allocations on an efficient frontier, which shows the lowest volatility for each expected return. Keep a comparison and see what changes when you revise an assumption.
Each issue will open with what’s interesting in the markets, without predictions, then cover my themes and theses, every trade and why I made it, and where the whole portfolio stands. The system behind it all is versioned, so you’ll see every change I make to it.