Older trading software followed rules a person wrote in advance. Newer systems pull in far more: funds have used satellite photos of retail car parks to read sales before earnings, and a UC Berkeley study of 4.8 million parking lot images found that edge was worth 4% to 5% in the three days around earnings announcements (Berkeley Haas). Notice the window: three days.
The risk regulators worry about is sameness. The Bank of England warned that more advanced AI trading "could lead to firms taking increasingly correlated positions and acting in a similar way during a stress, thereby amplifying shocks" (Bank of England, April 2025). IOSCO's May 2026 supervisory toolkit lists "Flash crashes from AI-driven trading" and "Herding behavior" among its systemic risks (IOSCO report, page 26).
I hold (or am considering) [company name and ticker, or ETF]. Using the annual report and quarterly results in this project, plus web search for anything more recent, write a 5-year research thesis for it. Cite the document or link for every factual claim, and say clearly when something is your inference rather than a stated fact.
Structure it as: what the business actually sells and who pays for it, how it makes money and whether that is getting better or worse over the last 3 years, the 2 or 3 things that must be true for this business to be meaningfully bigger in 5 years, and what the current price already seems to assume.
Finish with one paragraph I could reread in a year that says, in plain language, why I own this. Do not tell me to buy or sell, and do not predict a share price.
Assume it is 5 years from now and the thesis we just wrote turned out to be wrong. Work backwards and give me the 5 most likely reasons it failed, ranked by how likely they are, using evidence from the filings and recent news rather than generic risks.
For each reason, give me one specific, measurable warning sign I could check each quarter (for example "gross margin falls below X% two quarters in a row" or "the largest customer drops below Y% of revenue"), and where in the company's reports I would find that number.
Then tell me which one of these risks the market seems to be ignoring right now, and which one is probably already priced in, and explain why in two sentences each.
This is the whole long-horizon habit. When new results come out, add them to the Project and run this. It takes five minutes and stops you reacting to headlines.
New quarterly results for [company] are now in this project. Compare them against my saved thesis and the warning signs from my pre-mortem.
For each warning sign, tell me the current number, the previous number, and whether it is fine, worth watching, or triggered, with a citation to where you found it. Then tell me whether anything in these results changes one of the 2 or 3 things that must be true for the thesis, and separate real business changes from short-term noise like one-off costs or share price moves.
End with one of three verdicts: thesis intact, thesis weakened (say exactly what would need to happen next quarter to break it), or thesis broken. Do not recommend buying or selling.
IOSCO: Supervisory Toolkit for AI Use in Capital Markets, May 2026
Bank of England: AI in the financial system, April 2025
Berkeley Haas: how hedge funds use satellite images
SPIVA U.S. Mid-Year 2025 scorecard
Yahoo Finance: an AI-powered flash crash is coming