Using generative AI to find "tenbaggers" (stocks that increase in value by 10 times) is not about predicting the future, but rather a mundane task of offloading information volumes that the human brain cannot process to machines.
If you simply ask AI, "Tell me stocks that will go up 10x," it won't return useful answers.
However, this doesn't mean AI is useless for investing. If used as a tool to extract "signs," "structural changes," and "common conditions," individuals can gain an industry-wide perspective. Here is how to do it in five steps.
3 Prerequisites (Adjusting Expectations)
The term "tenbagger" was popularized by US fund manager Peter Lynch in his 1989 book One Up on Wall Street.
Before diving into methods, let's confirm three unavoidable prerequisites.
【Never use living expenses】
Money needed within a few months or emergency funds are out of scope. Investing with the goal of 10x returns involves high volatility, requiring you to endure periods where the price might halve. Using only money whose loss wouldn't disrupt your life is the sole entry point.
【Adopt a medium-to-long-term timeline】
Many examples cited by Lynch show stock prices growing over units of several years to ten years. Seeking results in six months is fundamentally incompatible with finding tenbaggers.
【Assume most candidates will fail】
A JP Morgan Asset Management report, "The Agony and the Ecstasy" (2014, Michael Batty), reports that since 1980, about 40% of Russell 3000 constituents ended with negative returns, and only about 7% were "extreme winners" significantly outperforming the index. Research by financial economist Hendrik Bessembinder also shows that wealth creation in the US stock market is extremely concentrated in a few stocks.
Hunting for tenbaggers is a game where you must be prepared to miss nine times without disrupting your life. Whether you hold this mindset first determines how you use AI subsequently.
3 Reasons Humans Miss Tenbaggers
According to Japan Exchange Group data, there are approximately 3,900 listed companies on the Tokyo Stock Exchange. Including US stocks and global industry trends far exceeds what an individual can visually track.
【Information Volume Limits】
Earnings releases, securities reports, industry analyses. Reading one company carefully takes hours. Comparing thousands of companies alone is not a realistic use of time.
【Confirmation Bias】
People tend to invest in companies they know or products they've used. This tendency, known in behavioral economics as confirmation bias or home bias, fixes one's vision to their "living sphere." The possibility that the next 10x stock is growing in an industry you don't know is structurally overlooked.
【Inefficient Analysis of Past Data】
Ending past tenbaggers with the memory of "I should have bought then." The work of verbalizing common financial metrics and business structures of companies that achieved 10x growth as conditions is often missing.
All three are information processing problems. This is exactly where AI becomes the breakthrough.
Which AI Should You Use?
Conclusion: Choose AI for stock hunting based on whether it "can search with citations" and "can read long disclosure documents directly." No single service is perfect; realistic operation involves using different tools for each step. The items below are arranged in order of workflow.
➊ Step 1: "Search-Integrated Type"
When mapping industries, open an AI that returns source links. OpenAI released "Deep Research" in 2025, which creates research reports by cross-referencing multiple sources. Google's Gemini also offers "grounding," matching answers with Google Search results. Treat answers without links as baseless and discard them immediately.
❷ Step 2: "Long-Text Reading Type"
Once candidates are narrowed down to a few companies, throw in securities reports or earnings presentation materials directly for summarization and issue extraction. Anthropic's Claude is a model emphasizing long-context processing, with web search added in 2025. Check the official page for character limits before use to reduce splitting efforts.
Regarding free vs. paid versions: Many wonder which is better. Free versions suffer from outdated training data cutoffs, lack of search capabilities, and limited reading volume.
Therefore, the criterion for switching to paid is how many primary sources you need to read. Once candidates exceed five and you're reading more than 10 disclosure documents monthly, moving to a paid version costing a few thousand yen per month is reasonable.
However, regardless of the service, output precision depends on question design. What matters isn't which AI you choose, but asking the same question to two or more AIs and looking for discrepancies. Note down conflicting numbers or company names to treat as issues for checking primary sources in the next step.
4 Questions to Dig Out Treasure Stocks
AI output is largely determined by question design. Ask AI for conditions, not stock names. Here are four ready-to-use questions:
【Identify Growth Industries】
Ask: "Rank rapidly growing global markets by revenue growth rate and investment amount." This maps industries. The areas identified here become the subsequent search scope.
【Extract Structural Changes】
Ask: "Tell me areas where the market will expand in the next 5 years due to regulatory changes, technological innovation, or demographic shifts." The goal is to verbalize irreversible underlying changes, not stock prices.
【Extract Common Winning Traits】
Ask: "List common financial, business structure, and competitive advantages of past tenbagger companies." Elements like high gross margins, recurring revenue, and barriers to entry appear as conditions.
【Convert to Conditions】
Follow up with: "Create criteria for companies meeting the above characteristics. Tell me only the features of likely matches."
Update Candidate List Every Two Months
If you try prompts once and stop, AI remains just a fancy search bar. Turning discovery into assets requires recording and updating. What information should be recorded?
【5 Pieces of Information to Record】
① Company Name
② Noted Structural Change
③ Reason for Belief in Growth (one sentence)
④ Numbers (Market Cap, Revenue Growth Rate, Operating Profit, Years Since Listing, Top Shareholder Ownership %)
⑤ Date Recorded
Record these in Excel or similar. Candidates with blank fields are those you don't yet understand.
Update timing is primarily every two months, with mandatory checks during earnings announcements. Verify if numbers fall within specified regulations; drop outliers with written reasons.
Once this habit sticks, you shift from vaguely watching news to gathering information with purpose. Intuitive "looks good" is replaced by "can explain why it will grow." The standard of not keeping stocks on the list if you can't write the reason for growth in one sentence becomes a bulwark for investment decisions.
Summary
What supports AI-driven tenbagger hunting is not the illusion that machines predict the future, but the design of a mundane, repeatable task: using AI as a net to scoop only "signs" from the vast sea of disclosure information from thousands of companies, while humans retain final judgment and responsibility.
After closing this article, one question suffices to start. Throw: "Tell me three industries whose market size will rapidly expand in the next 5 years, with evidence," and write the returned three into the first rows of your candidate list.
Bookmark this article as a reference for finding treasure stocks.
References
[1] Peter Lynch, John Rothchild, One Up on Wall Street (1989 / Japanese translation: Diamond Inc.)
[2] JP Morgan Asset Management, Michael Batty, "The Agony and the Ecstasy: The Risks and Rewards of a Concentrated Stock Position" (2014)
[3] Hendrik Bessembinder, "Do Stocks Outperform Treasury Bills?" Journal of Financial Economics (2018)
[4] Financial Services Agency / Investment Advisory Registration System under the Financial Instruments and Exchange Act (Article 29)
[5] Japan Exchange Group "Number of Listed Companies/Shares" Public Data (2024)
