How to Spot Red Flags in Chinese & HK Markets

AI Prompts to Avoid Value Traps and Streamline Stock Research

Jul 26, 2025

Introduction to AI Stock Screening: Avoid Value Traps with Smart AI Prompts

I'll show you how AI-powered stock screening can transform your investment research, helping you quickly eliminate value traps in Chinese and HK markets by spotting hidden red flags like regulatory risks and balance sheet issues. In this guide, I'll share proven AI prompts for stock analysis that save hours of manual work, focusing on business models, competitive moats, and forensic accounting to streamline your process and uncover genuine opportunities

But let's start with a brief historical anecdote:

Paul Erdős once said, "A mathematician is a machine for turning coffee into theorems." In 1979, Ron Graham bet him $500 he couldn't quit amphetamines for a month. Erdős won but complained: "You've set mathematics back a month. I'd stare at blank paper with no ideas, just like an ordinary person."

As a mathematician (with Erdős number 2, in case you care about those things), I've built an AI screening system that's like having that amphetamine-fueled Erdős for financial analysis—constantly alert, pattern-matching across data to flag disasters before you waste time on them.

I learned this the hard way after spending two days analyzing what looked like an undervalued stock, only to discover massive regulatory red flags buried deep in the footnotes.

Traditional stock screeners can find low P/E ratios, but they miss the nuanced warning signs hidden in footnotes that separate genuine opportunities from elaborate value traps. This isn't about replacing fundamental analysis—it's about creating a smarter filter so you stop wasting time on companies that were financial disasters from day one.

The Art of Asking AI the Right Questions

Most people ask AI about stocks like they're asking Siri for directions. The trick is being specific about what matters for investment decisions.

The Setup:

You're a veteran equity analyst known for spotting red flags and cutting through corporate nonsense. Focus on material business drivers, balance sheet quality, and corporate governance issues. Provide balanced analysis with specific evidence from verifiable sources like company filings, industry reports, or credible news. Use step-by-step reasoning, cite sources where possible, and structure responses clearly with bullet points or sections for readability. Never give buy/sell advice. Ready to start?

Business Model Reality Check

Before you worry about valuation or charts, you need to understand what the company actually does.

The "Golden Retriever Test":

Explain how [COMPANY] makes money in simple terms, as if to a non-expert: 
- What do they actually do and who pays them? Break it down into core products/services and customer segments. 
- Is this asset-light or capital-intensive? Provide examples from their operations. 
- What are the main revenue streams and profit drivers? Quantify where possible (e.g., percentages from recent filings). 
- What makes their revenue go up or down? Identify key variables like economic cycles, pricing power, or input costs. 
- How do they compete and what's their edge? Reference real-world evidence, not just company claims. 

Keep it clear, concise, and jargon-free but don't skip important business details. Base your explanation on the latest available data for accuracy.

If you can't explain the business to your dog, Warren Buffett wouldn't invest in it either.

Next, you need to know whether they're any good at it compared to everyone else trying to do the same thing.

Analyze [COMPANY]'s competitive position using a structured framework: 
- What makes them different from competitors? Highlight unique aspects like technology, brand, or cost advantages with supporting evidence. 
- Who are their main competitors? List 3-5 key players and compare market shares or performance metrics. 
- What barriers prevent others from copying their success? Evaluate moats like patents, network effects, or regulatory protections. 
- Who are their main threats and how serious are they? Assess emerging risks like new entrants or substitutes, rating severity (low/medium/high). 
- What happens if their primary advantage disappears? Scenario-plan potential impacts on revenue/profits. 
- Are they gaining or losing market share? Use data from recent filings, industry reports, or analyst estimates to support. 

Show evidence from recent filings or industry reports, not marketing speak. Structure your response with bullet points and cite sources for each claim.

The China/HK Special Requirements

If you're looking at Chinese or Hong Kong stocks, these markets have their own rules and ways to separate foreign investors from their money.

In China, the government isn't just a regulator—it's like the ultimate board member who can change the rules whenever they feel like it.

The Policy Risk Scanner:

For [CHINESE/HK COMPANY], assess regulatory and policy exposure in a comprehensive, evidence-based manner: 
- How does this company align with Beijing's current Five-Year Plan priorities? Reference specific plan sections and company activities. 
- Any exposure to recent regulatory themes (data security, anti-corruption, debt reduction)? Detail impacts with examples. 
- Cite specific policies, enforcement actions, or state media commentary from past 24 months, including dates and sources. 
- Is this sector facing headwinds or tailwinds from government policy? Quantify where possible (e.g., subsidy changes or fines). 
- Check for any mentions in provincial government reports or industry association guidance, and evaluate local vs. national implications. 

Look beyond English-language sources if possible, prioritizing official Chinese sources for accuracy, such as the CSRC official site, http://www.csrc.gov.cn/csrc_en/.

Rate overall policy risk (low/medium/high) and explain step-by-step reasoning.

Next: figuring out who actually owns what. In Chinese markets, the ownership structure can be more complex than a Russian nesting doll designed by a tax lawyer.

The Ownership Structure Detective:

Map [CHINESE/HK COMPANY]'s ownership hierarchy in detail: 
- Identify ultimate controllers (state/family/private equity) and their ownership percentages, using a simple diagram or list if helpful. 
- Are there substantial related party transactions? What's the scale vs. total revenue? Flag if >10% and assess fairness. 
- What is the salary/compensation of the top people? How does it compare to company performance? Use metrics like ROE or EPS growth. 
- Any funny business going on? Look for asset stripping, unusual acquisitions, or circular transactions, with specific examples. 
- Is company interest aligned with shareholder interests? Evaluate via dividend policies, voting rights, and incentives. 
- Trace offshore entities and cross-holdings - any circular ownership structures? Diagram if complex. 
- Flag if >30% revenue comes from government contracts or single customer, and discuss dependency risks. 
- For SOEs: Check for restructuring or merger mandates, citing announcements. 
- Recent insider selling or major shareholder changes? 

List transactions from past 12 months. What's the real power structure behind this company? Provide a balanced view with evidence from filings and news, rating alignment (1-10).

Balance Sheet Analysis

The balance sheet is where companies hide both their treasures and their disasters.

The Forensic Accounting Analysis:

Conduct enhanced forensic analysis of [COMPANY]'s financials, focusing on red flags: 
- Cash position: currency, location, accessibility, and any restriction—break down by jurisdiction and usability. 
- Hidden assets (real estate, land use rights, investments)—value them conservatively and note impairments. 
- Compare cash flow from operations to net income over 5 years—calculate ratios and flag if CFO < 60% of net income in 2+ years. 
- Check if pledged collateral exceeds 20% of assets—detail types and potential triggers. 
- Revenue authenticity: unusual receivables growth vs peers—compare DSO trends and industry averages. 
- Hidden debts: trust loans, wealth management products—estimate total exposure. 
- Off-balance-sheet obligations and guarantees—list and quantify contingent liabilities. 

For Chinese companies, pay extra attention to opacity red flags like inconsistent reporting. Use step-by-step calculations, cite filing pages, and rate balance sheet quality (strong/moderate/weak).

For Hong Kong and Chinese stocks, you also need to worry about whether you can actually trade the stock when you want to.

The Liquidity & Market Structure Check:

For [HK/CHINESE COMPANY], analyze market dynamics quantitatively: 
- Average daily turnover vs free float (flag if <0.5% for 30 days)—calculate based on recent data. 
- Short interest levels and recent changes—compare to peers and note spikes. 
- Mainland investor flows (Southbound Connect data)—track trends over past 6 months. 
- Whether shares are used as loan collateral—check for pledges and risks. 
- Any history of trading suspensions or unusual volume spikes—list events from past 24 months. 

What are the liquidity risks here? Rate overall liquidity (high/medium/low) with supporting metrics and sources.

Recent Filings Review

Companies love to bury bad news in regulatory filings. Your job is to find what they're trying to hide.

The Recent Drama Screen:

Review [COMPANY]'s recent regulatory filings and communications systematically: 
- Any profit warnings, guidance revisions, or missed targets—list dates and impacts. 
- Management or board changes and stated reasons—assess if voluntary or forced. - Material agreements, acquisitions, or disposals—evaluate strategic fit and pricing. 
- Regulatory issues, investigations, or compliance problems—detail status and potential fines. 
- Changes in accounting methods or auditors—flag if frequent or unexplained. 
- Unusual insider trading activity—quantify buys/sells and timing relative to news. 

Focus on items from the past 24 months that could impact the investment thesis. Structure chronologically, cite document references, and highlight materiality.

Check out my X.com

The Multi-Model Cross-Check

Here's where different AI models show their strengths. When researching Chinese or Hong Kong-listed companies, I run analysis through both DeepSeek (or Qwen) and a Western model like Claude or Perplexity. DeepSeek tends to better understand Chinese business practices and regulatory nuances, while Western models provide more skeptical outside perspective.

The AI Fact-Check & Cultural Bias Detection:

Another AI model analyzed [COMPANY] and provided this assessment: [paste analysis]. 

Please review their findings and: 

- Identify significant differences from our analysis, quoting specific sections. 
- Point out areas where they might be missing context, such as local market nuances or data sources. 
- Check for "face culture bias" (downplaying management issues)—flag subtle language that avoids criticism. 
- Highlight insights that strengthen or challenge our conclusions, with evidence. 
- Flag any potential errors or oversights in either analysis, suggesting verification steps. 

For Chinese companies, have the Western model critique potential cultural blind spots in the Chinese model's analysis. Provide a reconciled summary integrating both views.

You maybe realized that I omit valuation at this stage entirely because anchoring on price can cloud qualitative judgment, and basic multiples are a two-second lookup on any finance site.

The Final Decision

After all this analysis, you need a clear framework to decide whether this company deserves more of your time.

The Screening Decision:

Based on our comprehensive analysis of [COMPANY], provide a structured assessment:

Top 3 Investment Strengths: Most compelling positive factors
Top 3 Key Concerns: Most significant risks or red flags  
Policy/Regulatory Risks: Specific exposure to government policy changes
Governance Score: Rate transparency and shareholder alignment (1-10)
Biggest Research Gap: What we still need to understand

Screening Verdict: 
- Does this pass basic quality and governance thresholds?
- Are policy risks manageable for foreign investors?
- Are there enough positives to justify deeper research time?

Rate research priority (High/Medium/Low) and explain reasoning.

The Reality Check

Limitations:

AI research has significant blind spots, especially with Chinese stocks where local context is everything. I'm researching a company right now that looks horrible according to AI—it flagged supposed government risks that only took me a few minutes to realize were completely irrelevant to this particular business. AI can identify regulatory concerns but often misses the nuances of how policies actually get implemented. Western models frequently overlook "face culture" dynamics where problems get systematically downplayed in public communications.

But here's why this doesn't matter: when AI highlights just one or two negative points, it's incredibly quick to verify their validity if you have any local knowledge or industry context. A few minutes of targeted research can confirm or dismiss these flags entirely.

Why This Setup Is Still Incredibly Valuable:

Despite these limitations, I find this approach invaluable because it immediately eliminates obvious duds and structures my thinking before I begin the real analysis. If there's only one red flag, it's easy to quickly validate or disprove that specific concern and decide whether to invest more research time.

More importantly—and this is an underestimated benefit—reading through the AI analysis primes my brain for the real work. When I subsequently read through actual filings and reports, I'm already focused on the points that matter most. I know exactly what to look for. This cognitive priming dramatically speeds up my personal research because instead of reading documents blindly, I'm hunting for specific issues the AI surfaced.

It's like having a research assistant highlight the potentially problematic sections before you read the full document—your brain processes the information much more efficiently when it knows what patterns to watch for.

The Bottom Line

This isn't about letting AI pick stocks. It's about using AI to quickly eliminate the obvious disasters and surface the companies worth your limited time. Your golden retriever would approve of the clearer thinking. Though they'd probably still prefer you just threw the tennis ball instead of explaining VIE structures.