How to Replicate 4502% Returns with Domestic AI: Serenity's 3-Step Stock Selection Method

Recently, I've been seeing @aleabitoreddit in my feed, reporting her 4502% investment return this year—multiplying her capital by over 45 times.

The number is staggering, but what intrigued me more is that her logic isn't complex. Ordinary people can even try to replicate it using domestic AI tools instead of relying on GPT.

I spent some time deconstructing her approach and organized the core logic and operational steps to share with everyone.

Core Logic: Don't Chase Trends, Find the "Choke Points"

Most people buy stocks by looking at what's rising or rushing into hot sectors when influencers mention opportunities. The result is often entering when prices are already inflated by institutions.

Serenity's method is the opposite.

She doesn't chase the wind; she looks for the "throat" of the wind—the link in an industry that is most easily bottlenecked, hardest to replace, has the most lagging supply, yet is absolutely indispensable.

For example:

If a sector explodes, the first thing to run out isn't usually the downstream assembly plants, but an upstream material, a key component, or an obscure intermediate process. This segment might have a small market cap and little news coverage, but if it gaps, the entire industry chain is held back. Finding this step is what she calls "where the value lies."

Three Practical Steps: My Deconstruction Method

I followed this logic myself and broke it down into three steps to avoid emotional stock picking.

Step 1: Determine if the macro trend is solid.

Don't look at today's top gainers. Find directions likely to move forward over the next few years, like AI robotics, domestic substitution, low-altitude economy, advanced packaging, or key energy storage materials. The requirement is that demand has certainty for 2-3 years, not just a short-term policy or one quarter of orders.

Step 2: Break the industry into a detailed chain.

Once a sector is chosen, don't look at stocks immediately. Map the industry chain: upstream raw materials, key equipment types, intermediate components, modules, downstream system integration, and final applications. You can use AI for this, which I'll explain later.

Step 3: Find the "Choke Point," not the leader.

This is the most critical step. I repeatedly ask:

• If industry demand spikes, what will run out first?
• Which link has the highest technical barrier and is hardest to scale?
• Which link is hardest for customers to replace once integrated?
• Which link isn't the largest company but is most likely the "bottleneck"?

Going through these usually filters out a few "choke point" candidates. You're no longer seeing "this sector is rising," but "if this spot is cut off, everyone else has to wait." This value is often what institutions realize late and isn't fully reflected in the stock price.

How to Use Domestic AI Tools for These Steps

Many think you need GPT or Gemini for industry analysis, but domestic tools like Doubao, Qianwen, or DeepSeek are perfectly adequate if you ask the right questions. My experience is to avoid vague questions and instead make them specific and penetrating.

For example, you can ask step-by-step:

1. "Please help me map the complete [Robotics] industry chain, from upstream materials to downstream assembly, listed by hierarchy in detail."
2. "In this chain, which links have the highest technical barriers? Which have the slowest capacity expansion?"
3. "If robot orders grow 10x next year, which link will hit a capacity bottleneck first? Why?"
4. "In these bottleneck links, which are the main domestic companies? How is their customer stickiness and replacement difficulty?"

Ask about each link separately and dig deeper with follow-up questions.

I ask Doubao, DeepSeek, and Qianwen separately and only extract information mentioned by all three, then cross-verify with prospectuses or industry reports. This provides high reliability even without overseas models. While domestic AI may lag slightly in global chain penetration, this cross-verification method effectively closes the gap.

Personal Reflections

The essence of this method is replacing a "follow-the-crowd" mindset with "industry chain thinking," using AI to quickly structure information. It might not make you rich overnight, but it returns stock selection to the fundamental point of "what is truly scarce."

A small suggestion:

If interested, don't rush to buy. Take an industry you know well and run these questions through DeepSeek to see if you can find "choke points" you previously ignored. Truly valuable information is often hidden in these overlooked links.

Finally

The above content is a personal summary of investment ideas and does not constitute investment advice. Investment requires your own judgment. Feel free to discuss in the comments and follow me for more.