[AI Investment Lab #7] | 15 Investment Styles × MBTI | I had 3 AIs create the 'Ultimate Analysis Prompt' and tested them

When you start looking into investing, you get confused by all the katakana terms popping up one after another, like "value investing," "swing trading," and "day trading."
Moreover, lately, it has become standard practice to have AI do that analysis.

So, in this article, I will do two things at once.

1. Explain 15 investment styles in a way that is easy for beginners to understand(including "who does it = famous people," "what personality is suitable = MBTI," and "numerical benchmarks for judgment")

2. The main verification: Have three AIs—ChatGPT, Claude, and Gemini—design 'analysis prompts for 15 styles,' grade them, and create the best one.

The motto of this lab is "Verification > Vibe." We won't end with 'AI seems kind of amazing'; we will actually get our hands dirty and compare with numbers.

Part 1 | How has AI changed investing in the first place?

First, let's start with the premise. As of 2026, there is a survey showing that about 70% of people who invest in individual stocks use generative AI (ChatGPT, Claude, Gemini, etc.) to research stocks. AI is now a standard tool for investing.

However, I will state the most important point first.
AI is not a 'magic crystal ball that predicts stock prices.'

In a study by a university, even when AI was made to analyze a massive amount of financial news, the accuracy rate for short-term predictions of individual stocks was about 51%. It was almost the same as flipping a coin. AI cannot answer, 'Will this stock go up tomorrow?'

So what is it used for? The answer is not 'prediction' but 'analysis, summarization, and reflection.' Having it summarize financial statements in seconds, comparing indicators with competitors, pointing out oversights in your own judgment. The correct way to use it is as an "excellent research assistant" that organizes vast amounts of information.

There are two main ways to involve AI.

• 'Hands-off' type AI: Robo-advisors (WealthNavi, ROBOPRO, etc.) automate everything from stock selection to allocation and accumulation. It is a perfect match for index investing, and beginners should start here.

• 'Partner' type AI: Have ChatGPT or Claude research financial results, industries, and competitors to help with your own judgment. It shows its strength in styles where you choose stocks yourself..

And this is the crux of this article.AI answers can casually contain 'plausible-sounding lies' (hallucinations). In one survey, only a little over 60% of people properly verify AI information with primary sources (financial statements, IR materials, etc.) every time. 'Ask AI -> Verify with primary sources yourself' is one set.

Part 2 | 15 Investment Styles by 'Famous People, Personality, and Numbers'

There is no official number of investment styles, but when organized, there are about 15 types.
They can be divided into three major axes. For each style, I will add 'representative famous people (Japanese if possible),' 'suitable personality (MBTI tendency),' and 'numerical benchmarks for judgment.'

*MBTI and numbers are 'references/benchmarks.'

Axis ① Styles divided by holding period

1. Scalping (Ultra-short term of a few seconds to a few minutes) A method of buying and selling dozens to hundreds of times a day, accumulating small price differences. The most difficult level requiring reflexes and iron discipline. This is the style that Testa (cumulative profit over 7 billion yen) mastered in his early days. 📊 Dozens to hundreds of times a day | 🧭 ESTP/ISTP (Reflex type) | 🤖 Almost no role (too fast)

2. Day Trading (Completes within the day) Do not hold stocks overnight; sell everything by the end of the day. Instead of being affected by overnight news, you are glued to the screen during the day. This is the legendary style where BNF earned about 2 billion yen in a few minutes during a misorder incident. 📊 No overnight holdings | 🧭 ESTP/ISTP | 🤖 Previous day analysis/review only

3. Swing Trading (Several days to several weeks) Ride the waves for a few days to a few weeks. You don't need to be glued to the screen, and it is easy to do even for office workers or as a side hustle. Trend following, which cis says he started winning with after switching to short-term, is the basis. 📊 Assumed range ±5-15% | 🧭 ISTP/ESTP + analytical types | 🤖 Effective for organizing materials

4. Position Trading (Several months to 1 year) Look at the flow of the economy and business performance and hold for several months to a year; it is the middle ground between short-term and long-term. 📊 Spanning quarterly earnings | 🧭 INTJ/ENTJ | 🤖 Effective for organizing macro trends

5. Long-term Investment (5 years or more) Bet on the growth of the company itself and keep holding. It requires little effort and is suitable for beginners. Represented by Wahei Takeda (died 2016), who was called the 'Warren Buffett of Japan'. 📊 Check stock prices only a few times a year | 🧭 ISTJ and other patient types | 🤖 Long-term competitiveness check

Axis ② Styles divided by mindset (philosophy)

6. Value Investing (Buying undervalued stocks) Buy stocks that are cheaper than their intrinsic value and wait for them to return to their fair price. Toshiya Imura, who turned from a comedian to an investor and has accumulated profits of around 8.5 billion yen, is famous (overseas, Buffett and Graham). 📊 PBR under 1.0x, ROE 8% or higher | 🧭 INTJ (analytical/independent) | 🤖 AI's main battlefield

7. Growth Investing (Buying growth stocks) Bet on companies that will grow significantly from now on, even if they are currently expensive. AI, semiconductors, and SaaS are the main battlefields. Cathie Wood (ARK) is the representative overseas. 📊 Sales growth rate 20% or more per year | 🧭 ENTJ/ENTP/INTJ | 🤖 Scenario verification

8. Quality Investing (Holding super-excellent companies) Hold 'exceptionally good companies' with high profit margins and sound finances for a long time. Focus on 'company strength' rather than cheapness. Represented by the later years of Buffett and Terry Smith. 📊 ROE 15% or higher, operating profit margin 20% or higher | 🧭 INTJ/INTP | 🤖 Financial comparison

9. Index Investing (Investing in indices as a whole) Invest in the entire index such as the S&P 500 or All Country World Index. It is the most popular method for beginners and the royal road. In Japan, 'Hottarakashi Investment Technique' by Hajime Yamazaki (died 2024) and Kenichi Mizuse popularized it. Atsugiri Jason is also famous. 📊 Trust fee 0.2% or less, monthly accumulation | 🧭 ISTJ/ENFJ | 🤖 Fully automated with robo-advisors

10. High Dividend Investing (Aiming for dividend income) Aim for regular income called dividends. Blogger Pelican achieved FIRE in his 40s with 5-6 million yen in annual dividends. Testa has also been focusing on dividends in recent years. 📊 Dividend yield 3.5% or higher | 🧭 ESTJ/ISFJ | 🤖 Dividend cut risk confirmation

11. Dividend Growth Investing (Aiming for consecutive dividend increase stocks) Aim for 'companies that increase dividends every year' rather than current yield. The US 'Dividend Aristocrats' (25+ years of consecutive dividend increases) are the standard. 📊 10+ years of consecutive dividend increases | 🧭 ISTJ/ESTJ | 🤖 Confirmation of dividend increase track record

12. Shareholder Benefit Investing (Aiming for shareholder perks - unique to Japan) A culture unique to Japan where you receive 'things you can use in daily life' such as meal vouchers and shopping coupons. Former professional shogi player Hiroto Kiritani is the synonym for this. 📊 (Perks + Dividend) Total yield 3% or higher, under 100,000 yen | 🧭 ESFJ/ESFP | 🤖 Perk comparison

Axis ③ Special styles

13. Event Investing (Aiming for events such as earnings, TOB, splits, etc.) Aim for price movements triggered by specific events. The focus is on past patterns and risk organization rather than prediction. 📊 A few days before and after the event | 🧭 INTP/INTJ | 🤖 Pattern analysis

14. Momentum Investing (Riding the upward momentum) Trend following based on the idea that 'stocks that are rising will continue to rise for a while'. The opposite idea to value. 📊 Top 6-month return, near 52-week high | 🧭 ISTP/ESTP | 🤖 Screening

15. Theme Investing (Investing in groups of stocks with hot themes) Bulk buying based on seasonal themes such as AI, semiconductors, and space. Easy to ride the news, but be careful of 'fads'. 📊 Stocks with high sales dependence on the theme | 🧭 ENFP/ENTP | 🤖 Identifying stocks

Part 3 | MBTI is not a 'prophecy' but a 'mirror of yourself'

I have paired MBTI with styles, but here is an important warning. MBTI cannot tell you 'you will make money with this stock'. This is because stock prices do not move based on personality.

Then what is it useful for? It is to know your 'losing patterns' in advance. Even in the professional world, it is said that 'investment results are determined by mentality,' so what determines victory or defeat is action rather than knowledge. Even if you cannot move the market, you can change your own actions.

Common pitfalls and countermeasures by 4 groups:

• Analysts (INTJ / INTP / ENTJ / ENTP): They tend to trust their own analysis too much, leading to delayed stop-losses when they are wrong. → Decide on a mechanical rule to 'sell immediately if wrong' in advance.

• Sentinels (ISTJ / ISFJ / ESTJ / ESFJ): Because they are diligent, they tend to panic at news of a market crash and stop their accumulation plans at the bottom. → The solution is 'not to look at the market'. Do not open the app.

• Explorers (ISTP / ISFP / ESTP / ESFP): They tend to trade too much in search of speed and excitement, often losing money to transaction fees. → '90% of assets in accumulation, only 10% for high-risk challenges'.

• Diplomats (INFJ / INFP / ENFJ / ENFP): They choose based on philosophy or empathy, and emotions easily influence their judgment. → Separate 'the desire to support' from 'investment decisions'.

Thinking 'I'm the only one who will be fine' is the most dangerous thing. People who know their own weaknesses are better able to avoid major failures.

Part 4 | [Main Topic/Verification] I Had 3 AIs Create 'Analysis Prompts for 15 Styles' and Scored Them

This is where the real work of this lab begins. Even if we call it 'AI investment analysis,' each AI should have its own strengths and weaknesses. I verified this across all 15 styles.

What was verified (what is being measured is not 'accuracy')

What was verified was 'which AI can design more rigorous and reproducible analysis prompts'. Prediction accuracy is not measured (short-term AI predictions are essentially coin flips—see Part 1). What is being evaluated is faithfulness to primary information, comprehensiveness, reusability, resistance to hallucinations, and the structure that leads to investment decisions.

Method

1. Prepare one common prompt (a task specifying the perspectives, numerical benchmarks, and judgments for the 15 styles)

2. Submit it to ChatGPT, Claude, and Gemini under the same conditions

3. Each AI generates 'analysis prompts for 15 styles' (= 15 prompts × 3 companies)

4. Score based on 9 criteria (0-2 points each, weighted) and convert to a 100-point scale

Results

Rankings and Observations

1st Place: Claude (96 points) It allows for immediate copy-pasting via code blocks, clearly specifies 'replace only this part,' and fixes the output into a 5-part structure plus exit conditions. However, most of the winning margin was due to
'usability', and the depth of the analysis itself was on par with Gemini. Points were deducted for leaving ambiguous expressions (e.g., 'sufficient liquidity').

2nd Place: Gemini (88 points)
The highlight was forcing the output table to be structured with 'Source/Reference Date' as a column. Designing it to require a source for every single number is the best fit for verification. The identification of weaknesses was also specific. However, each style came with a massive amount of theoretical explanation, and external SEO article links were mixed in, so it requires extraction and verification for copy-pasting. This 'over-stuffing & external inclusion' resulted in a point deduction.

3rd Place: ChatGPT (80 points)
It is faithful, concise, and easy to handle, but the instructions for primary information were 'vague,' and it stopped at instructions without creating an output template. It had the thinnest finish of the three companies.

⚠️ Important limitations of this verification (disclosed honestly)

The scoring was performed by Claude, which is one of the subjects of the verification.
AIs tend to rate their own output highly, and the result of Claude being in 1st place may contain self-evaluation bias. In fact, Claude's winning margin is skewed toward 'ease of use,' a perspective prone to subjectivity.

To make it fairer, we should have ChatGPT and Gemini anonymously score the same three items and determine the result by the average of the four. (This will be done in the next issue, #8). The only high-confidence conclusion that can be drawn at this point is that 'Gemini and Claude are neck-and-neck, with ChatGPT one step behind'.

Meta-discovery (the most interesting part this time)

Rather than a single winner, the accurate conclusion was that 'each AI has different strengths'.

• Gemini ... The creation of the output table and 'columnar sources' are the best. If you want rigor, this is the one.

• Claude ... The best for immediate copy-paste use and obedience to instructions. If you want to use it right away, this is the one.

• ChatGPT ... The most concise but the shallowest. Easy for a starting point.

In other words, in practice, it seems effective to have Gemini create the framework and then use Claude to refine it for immediate use.

The 'Definitive Prompt' integrating all 3 companies

I took the best parts of all three companies to create a definitive version. The foundation is Claude's structure, into which I transplanted Gemini's 'Source/Reference Date' column (forcing a source for every number), and trimmed the redundancy with ChatGPT's conciseness.

I will provide the definitive version for value investing as a representative example (the remaining 14 styles can also be created using the same template).

# 役割
あなたはバリュー投資の分析アシスタントです。指標の割安度を「同業平均」「自社の過去5年平均」と比較し、割安の理由とバリュートラップ（万年割安）の可能性まで検証して判定します。

# 対象（ここだけ差し替える）
・銘柄：______（証券コード：____）
・基準日：______

# 根拠のルール
・参照の優先順位：決算短信 ＞ 有価証券報告書 ＞ 決算説明資料 ＞ 適時開示 ＞ IR資料 ＞ 会社HP ＞ 公的・業界統計 ＞ 信頼できる報道。
・すべての数値に〈資料名・年度・基準日（可能ならページ）〉を併記する。
・各項目を「事実（資料で確認）／推測（論理的推論）／不明（資料で確認できず）」に分類する。取得できない値は「不明」と書き、推測で埋めない。

# 分析観点（すべて評価）
1. 指標：PER・PBR・ROE・自己資本比率・営業利益率・FCF・配当利回り
2. 比較：同業種平均、および過去5年の自社平均との対比
3. 割安の理由（一時的か、構造的か）
4. カタリスト（割安が是正される具体的なきっかけ）
5. バリュートラップの検証（「安いまま」で終わる要因の有無）

# 数値サマリー（この表を必ず埋める。空欄は「不明」と記入）
| 指標 | 割安の目安 | 当社実測値 | 同業平均／自社5年平均 | 参照元・基準日 | 判定(◯/△/×) |
|---|---|---|---|---|---|
| PBR | 1.0倍未満 | | | | |
| PER | 業種平均以下 | | | | |
| ROE | 8%以上 | | | | |
| 自己資本比率 | 40%以上 | | | | |
| 営業利益率 | プラス・安定 | | | | |
| FCF | プラス（創出できているか） | | | | |
| 配当利回り | （参考値） | | | | |

# 必ず複数挙げるリスク
バリュートラップ（万年割安）／構造的な業界衰退／減益で割安が解消（EPS低下でPERだけ上昇）／ガバナンス・資本効率の低さ／カタリストの不在。

# 出力（この5部構成・数値と表を優先）
①結論：真の割安か、バリュートラップか（一言）
②数値サマリー：上の表（参照元・基準日まで必ず記入）
③根拠：「事実／推測／不明」に分けて記述。特に「なぜ割安なのか」を明示
④リスク：上記から複数
⑤判断：【買い】／【様子見】／【見送り】＋ 根拠を表の充足状況で2〜3行

And, the 'next step' for this verification

Just in case, here is the most important caveat. This three-company comparison is strictly an evaluation of the 'quality of the blueprint (prompt).' I have not yet run the definitive version with any real stocks even once.

The real verification is ahead—running the definitive version with companies like Mitsubishi Corporation and measuring ① quality of output, ② how much 'unknown' appears, and ③ whether it is correct when cross-referenced with primary information. Only by doing this is the 'verification' complete. Next time, I plan to perform the four-party scoring along with the actual run.

Summary | How should beginners start? + How to get along with AI

We have looked at 15 styles, but the answer to 'which one, in the end?' is clear.

First, let's start with 'index investing'. You simply invest small amounts regularly into the entire market. If you use NISA, you get tax benefits, and if you want to leave it to AI, you can start automated management with robo-advisors as well. It is the classic approach that many professionals agree is the 'optimal solution for beginners'.

Once you get used to it, take your second step according to your personality and lifestyle.

• You like researching companies (you want to make AI your partner) → Value investing / Growth investing

• You want regular cash flow → High-dividend investing

• You appreciate things you can use in daily life → Shareholder benefit investing

• You want to make decisions in a few days to a few weeks → Swing trading

Conversely, day trading and scalping require an environment where you can stay glued to the market all weekday long and significant experience, so they are not suitable as a first step.

Finally, here are the two core principles of this article. AI is a 'research assistant,' not a 'prediction machine'. And MBTI is a 'mirror to know your own losing patterns,' not a 'personality test for making money'. If you master these two as tools, you will surely find a style that suits you.

And above all—Always verify AI answers yourself by checking primary sources. This is the most important rule of 'Verification > Vibe'.

[Appendix] 'Image' of suitable investments by MBTI type (16 types)

I have summarized all 16 types so you can roughly grasp 'what kind of investment might suit you' based on your type.

Conclusion

We have looked at 15 investment styles through 'celebrities, MBTI, and numbers,' grasped the correct way to use AI, and even verified the 'ultimate analysis prompt' using 3 AIs. You can copy and use the definitive value investing prompt in the article as is, and you can create your own for the other 14 styles using the same template.

By now, you are ready to find a style that suits you and start using AI while 'verifying' it. All that's left is to try one style that interests you with a small amount. And remember, always verify AI answers by checking primary sources yourself—as long as you don't forget that, you'll be fine.

Next preview: The paid magazine 'AI Investment Lab Premium' starts from #8

Free articles (AI Investment Lab) will continue to be published as before.

On the other hand, in the paid magazine 'AI Investment Lab Premium', I will publish content that can be used repeatedly as a reference, after spending more time verifying and refining it.

In the paid version, I plan to sequentially from #1, creating content that delves even deeper based on the free articles.

The planned content includes, for example, the following:

• A definitive collection of prompts for all 15 investment styles

• Prompt tuning and verification by MBTI type

• Execution logs of running the definitive prompts on actual stocks

• Detailed verification including AI 'unknown' rates and cross-referencing with primary sources

• Analysis and supplementary data that could not fit in the free version

I will continue to value the content that can be learned from the free version alone.

Furthermore, for those who want to 'use the completed prompts as they are,' 'efficiently utilize verified analysis,' or 'read deeper insights,' I will produce articles for the paid version with quality as the top priority.

If you think it might be even a little bit useful, I would be happy if you could check out AI Investment Lab Premium.

I will continue to share valuable content in both the free and paid versions with the theme of 'Verification > Vibe'.

List of SNS links

• X: @EBE_Ryuki

• note: ebe0911

• Zenn: ebe_ryuki

• Qiita: ryukiebe0911

• LinkedIn: Ryuki Ebe

• Instagram (AI Investment Lab exclusive): @aitoushilabo