The hour before a trade is where most mistakes get baked in: when the research is done and the decision is about to be made. Five questions force the reasoning into the open. If anything's shaky, now's when to find out.
// On this page
Search “questions to ask before buying a stock” and you get self-reflection frameworks: are you ready to invest, what’s your risk tolerance, have you thought about your time horizon. Search “questions to ask AI before buying a stock” and you get generic research prompts: the same ones you’d run a week before you were anywhere near committing capital. Neither serves the moment that matters: the hour before you place the order, when you’ve done the work and you’re about to act on it.
Five questions, in order. First the bet itself, last the blind spot you didn’t think to ask about. Each one is something AI is good at: structuring the reasoning, flagging what you’ve assumed, articulating the bear case you’ve quietly discounted. None of them ask AI to do the thing it can’t: tell you whether to buy. These assume you’ve already done the background research. If you haven’t, that’s a different job: the free AI tools worth using for it are covered separately. Come back here once you’ve got a view.
The five questions: (1) What am I actually betting on? (2) What’s the strongest case against buying this today? (3) What does the market already believe? (4) What would have to change for me to be wrong? (5) What am I not asking?
What am I actually betting on?
The story you tell yourself about why you’re buying (strong brand, good management, big market) is almost never the actual bet. The bet is one specific thing that has to remain true for the returns to materialise. Most buyers can’t state it cleanly, and the comfortable reason is that the story does the work of hiding it. The vague version is easier to live with. It never has to be right about anything in particular. This question forces AI to peel the narrative back to its load-bearing claim. When the answer comes back as a vague paragraph, I push again until it collapses to one testable sentence. That sentence is what you’re buying.
What’s the strongest case against buying this today?
There’s a difference between a bad company and a bad entry. A company can be a good long-term hold and still be the wrong thing to buy this week: earnings two days out, a sector rotation underway, a competitor about to launch. Most bear-case prompts ask for the long-term reasons not to own a stock, which produces a generic risk register. This one asks for the timing case: the reason a cautious buyer might wait three to six months even if the thesis is right. When I run it, I’m not looking to be talked out of the buy; I’m looking for the one near-term thing I’d kick myself for ignoring.
What does the market already believe?
This is a question about the story behind the price, not the live price itself. AI can’t tell you what the stock is worth right now. But it can tell you the story the current price appears to be telling: what investors must be assuming about growth, margins, and competitive position for the price to make sense. That’s a reasoning task, and it’s one AI does well. The value isn’t in the answer itself. I read it to see whether the assumptions baked into the price match my own. If the market is assuming 25% revenue growth for five years and I think it’ll do 15%, that gap is the trade.
What would have to change for me to be wrong?
A thesis you can’t falsify isn’t a thesis. This question is sharper than the version most research checklists include. It’s not “what could go wrong” but “what specific number or event, appearing in the next quarter, would tell me clearly that the core assumption isn’t holding.” A threshold, not a direction, fixed in advance, while you can still be honest with yourself. I write mine down before I buy, because I know that without one I’ll spend the following months quietly moving the goalposts to keep a losing thesis alive, and I’ll have a perfectly good reason each time.
What am I not asking?
This is the question no other list includes, and the one that earns its place last. The other four are structured versions of things a disciplined buyer would do anyway. This one uses AI for something only it can do: spot the dimension of the decision you’ve left out of the frame entirely.
Confirmation bias in buy decisions is rarely about being wrong on what you know. It’s about omission: the question you didn’t think to ask because the thesis already felt complete. AI isn’t emotionally invested in your thesis being right. Give it a clean three-sentence summary of what you believe, ask it what’s missing, and it will name the dimension your reading has been quietly avoiding. Sometimes the answer is uncomfortable. That’s the point.
I ran this on a META thesis recently, framed around AI-driven ad optimisation and capital discipline. Gemini flagged the regulatory dimension, which wasn’t in my framing at all: specifically, whether the EU’s Digital Markets Act could formally decouple Meta’s cross-platform data synthesis and erode the proprietary data moat the whole thesis was built on.
// What Gemini said
Dimension flagged: Regulatory
If antitrust mandates or privacy legislation (such as the EU’s Digital Markets Act) formally decouple Meta’s ability to synthesise data across Instagram, WhatsApp, and Facebook, by what specific percentage does the “proprietary data” moat degrade, and can the AI engine maintain its targeting superiority without that cross-platform signal?
How would a structural decoupling of Meta’s data silos change your assessment of a 40% operating margin floor?
Where these fall short
The questions structure the reasoning. They do not replace it, and they cannot reach the data the moment of decision depends on. In a standard chat session (no live data feed wired in) AI doesn’t have the live price, the current options chain, real-time volume, or the order book in the seconds before you place the order. Any specific number it gives you (a P/E, a margin figure, a price target) needs verifying against a live source before you act on it. The models are better at the shape of the analysis than at the arithmetic. Use the answers to test your own thinking, not as inputs to the trade. Ask the model to make the call itself and the confident narrative hides which numbers it actually checked, which is the whole reason these questions put you, not the AI, in the chair.
The decisions these questions protect against are the ones you’d already half-made before you sat down: the buy you were looking for permission to make, not advice on whether to make it. Fifteen minutes is a cheap price for finding that out before the money’s on the table. Once you’ve bought, the Prompt Stack Field Guide covers the decisions that come after the trade, with five copy-paste prompts for the post-entry moments that usually get made on instinct. For the broader four-stage framework these questions draw from, see the Prompt Stack.
The pair to this one runs in the opposite direction. These five questions guard the buy. There is a matching prompt for the sell decision. It does the same job in reverse, forcing the exit trigger to be written down before the conviction has somewhere to escape to.
Field Report
What worked: The sequence. Each question is something AI is good at: structuring the bet, arguing timing, articulating what’s priced in, defining a falsification trigger, naming the blind spot.
What didn’t: Anything that needs a live number. Verify every figure the model produces before it touches the trade. The questions are for thinking, not for pricing.
Time required: Fifteen minutes, less if you already know the thesis well enough to summarise it in three sentences for Q5.
These five questions ran against three real buys before being written down: NFLX, GOOGL, TXRH. The trades (entry windows, hold durations, exit returns) are at /trades. The sell side of the same discipline follows the same pattern.
Ben tests ways of getting reliable answers from AI on his own investing: documenting what each model got wrong, what each one caught, and the prompts that survived the cuts. About Ben →
The site runs AI on real investing decisions. Start with the Prompt Stack for the four-stage framework, or the Field Guide PDF for the condensed version, free, no email.