Written by: TECHUB NEWS Hong Kong Report｜Reporter Alma Li｜September 2026
Introduction:
From project discovery, due diligence research to asset listing in the exchange Listing Team, Jack has experienced a multidimensional asset evaluation process serving platform operations. Now, as an independent investment researcher, he tries to transfer this "networked thinking" to the U.S. stock market, digital assets, macro and cross-asset research, and builds a sustainable, cross-verifyable, and reviewable personal investment research workflow using AI. In his view, what AI truly brings is not "cognitive equality," but rather equality in acquiring knowledge; ultimately, it is humans who define problems, judge evidence, and take risks.
In the many narratives of "AI + investment research," the most common scenario is a user inputting a prompt, and the system quickly providing a "buy" or "sell" conclusion. Jack keeps his distance from this imagination.
Core Viewpoint:
"Real investment research is not a button, but a systematic research project."
The value of AI lies not in making decisions for people, but in connecting originally scattered, repetitive, and hard-to-trace research actions into a verifiable, updatable, and reviewable process.
Since entering the cryptocurrency industry in 2021 and participating in the complete processes of project discovery, due diligence, business communication, and asset listing, his experience at the exchange has led him to develop a multi-factor and dynamic research habit. After leaving the institutional perspective and beginning to face the market with personal funds, the questions transformed into another form: assets are more diverse, while information is more fragmented.
Therefore, he started to build an AI-native investment research infrastructure for himself. It does not serve an automated trading commitment but addresses a more fundamental question: when a researcher faces a market mainline, an asset, or an unexpected signal, how can they form a complete logic faster without handing over judgment to a black box?
From Project Listing to Personal Investment: Research Goals Have Changed, but Underlying Thinking Has Not
TECHUB NEWS: What has your experience with the exchange Listing Team left you for doing independent investment research today?
Jack: When doing listings, it wasn’t just about the quality of a project, how much funding it raised, or how fast the sector was growing. From the perspective of the exchange's business scenario, we need to look at the project itself, real users and market demand, whether it can bring in new users and trades, whether there is sustained liquidity and asset accumulation, and whether it aligns with the platform's strategy and user risks.
The final judgment is dynamic: is it suitable to be listed at this particular point in time? The object of research may seem to be the project, but it actually serves the business objectives of the platform.
When the research identity shifted from the exchange to individual investors, the goal changed accordingly: it is not about judging whether a project is suitable for listing on the platform, but whether an asset is worth allocating, why it is worth allocating, and under what conditions it may not be worth allocating.
But Jack believes the underlying thinking of both is essentially the same. "Value and price are two different things," he says. Researching a company or an industry cannot only prove that it is "good" but must also consider fundamentals, expectations, valuations, prices, catalysts, and market environments. Truly effective research should be a form of networked thinking, rather than deriving linear conclusions from isolated pieces of information.
Pain Points of Personal Investment Research: Information is Not Lacking, but the Decision-Making Process is Fragmented
TECHUB NEWS: What is the biggest challenge after transitioning from institutional research to personal research?
Jack: As assets become richer, research and decision-making have paradoxically become more complex. Today's trading platforms can offer a vast number of targets, including digital assets, U.S. stocks, and gold, but the process from "seeing an asset" to "forming an investment judgment" is often fragmented. You need to step away from the trading interface to look at market trends, news, social media, industry data, and company information before piecing these fragments back into your own logic.
In his view, what individual researchers truly consume is often not the lack of opinions, but rather the decision-making friction generated from collecting, comparing, and updating information across platforms. The starting point of the AI research system is precisely to reduce this friction, connecting "discovering opportunities – understanding assets – verifying logic – continuous tracking" into one workflow.
This workflow first needs to identify the mainline currently being traded in the market: whether funds, industry trends, fundamentals, prices, and events are pointing in a certain direction; once the mainline is established, it’s necessary to understand specifically why certain assets might benefit, what their business logic, project logic, and transmission paths are; thereafter, supplement the evidence chain with independent data; finally, place judgments into continuous tracking and review, instead of ending after a single output.
"It comes from a real research workflow, not a process designed just to showcase AI," Jack says.
AI is Not a Buy/Sell Button, but a Replicable Research Assistant
TECHUB NEWS: Can you provide a case study showing how the system plays a role in real research?
Jack: Taking AI and the semiconductor mainline as an example, in the long term, they may still be significant industry directions for many years ahead; however, this does not mean that there should be indiscriminate chasing at any time.
In the interview, he recalls that while the long-term industrial logic remains strong with AI and semiconductors, being optimistic does not mean every moment is suitable for allocation. During July and August, the semiconductor sector experienced a noticeable pullback and differentiation. The change identified by the system was not simply to conclude that "AI or the semiconductor mainline has failed," but that the market began to transition from a relatively broad rise to clearer internal rotations and shifts of the mainline. On one hand, the performance of different sub-sectors and companies within the semiconductor industry began to differentiate markedly; on the other hand, sectors like pharmaceuticals and energy gradually gained more confirmation in terms of price, funds, and relative strength. Whether a company or industry is inherently "good" is one question; what the market is truly trading now is another.
In Jack's view, the key is not to provide explanations for market movements after the fact, but whether these signals can be reliably identified at the time and whether they can offer valuable judgments for the next one to three months. This module must also be continuously corrected and iterated through practice.
This precisely defines the boundaries of AI's role: researchers are responsible for defining problems, designing frameworks, consolidating industry know-how, and deciding which signals are worth incorporating into the system; AI is responsible for executing information processing, logic verification, and continuous updates according to those definitions. AI can expand the range covered by humans, standardize the research process, and retain available data and judgment at the time, but cannot replace the human responsibility for investment decisions.
"I would now prefer to define it as an assistant," he says. "It enables the investment research process to be replicable: not only can you see what conclusions were drawn today, but you can also review why the system formed such inferences under the information conditions at that time."
Data Layering: Define Evidence Quality First, Then Let AI Reason
TECHUB NEWS: How does the AI system avoid being biased by incorrect information or single sources?
Jack: Controlling AI errors cannot start only from the data sources; upstream of that is problem definition. For example, asking "which company in the semiconductor industry is best" is itself an imprecise question—best could mean the fastest-growing, the highest technical barrier, the best capital return, or currently the most worthwhile investment? Different decision objectives will lead to completely different research paths. Thus, he breaks down the entire validation process into five steps: problem definition, reasoning chain, data sources, cross-verification, and reverse evidence. Then he categorizes data into primary, secondary, and tertiary.
He categorizes regulatory disclosures, company financial reports, and earnings calls, as well as direct data published by the Federal Reserve, treasury departments, and statistical agencies, as primary data: this type of information is traceable and has the highest credibility, and should serve as the primary foundation for reasoning. Bloomberg, Wind, or high-quality financial media can serve as secondary information for supplementation and cross-comparison; while rumors on social media and discussions in forums are tertiary data, which should at most serve as clues and should not cover or replace primary evidence.
This layering does not mean that AI automatically becomes reliable; on the contrary, it returns the responsibility of "data screening" back to researchers. The model can process, organize, and present sources more efficiently, but which data is usable and which conclusions need to be refuted still require human rule-setting.
Jack believes that a long-term trustworthy AI investment research system should meet at least four standards: the research process is modular rather than outputting black box conclusions directly; each module has clear data logic and evidence chain; conclusions can be cross-verified and actively seek conflicting evidence; and when new information arises, the system can continuously update and accept reviews.
AI Brings Knowledge Equality, but Not Cognitive Equality
As large models lower barriers to information acquisition and learning, the question of "Will AI give individual investors the same cognitive abilities as institutions?" has become a hot topic in the market. Jack's answer is clear: AI can bring knowledge equality, but it does not equate to cognitive equality.
"Today, we can use AI to access top courses, research papers, and professional concepts more quickly, which is indeed a significant form of knowledge equality," he says. "However, having access to the same data and the same tools does not mean we will arrive at the same investment judgments."
Differences arise from real-world experience, as well as the ability to define problems. Even with the same data and using the same model, different researchers will have different understandings of industries, risks, verification pathways, and position management, leading to differing structures being input into the system, and eventual conclusions may even head in opposite directions.
For Jack, the truly hard-to-replicate "edge" is not a piece of open code or a seemingly mature prompt, but the ability of the researcher to continually discover problems, correct them, and iterate frameworks in real scenarios. The stock-picking logic publicly displayed is often just a facet of the system; how to identify changes that have not been fully priced earlier and how to recognize reverse risks is much closer to the core of research capability.
Verify on Yourself First, Then Consider Productization
Currently, this system is still in continuous refinement. Jack refers to it as OPC (One Person Company)-style practice: first serving his own research and then validating whether it truly addresses the real issues in investment decision-making during its usage.
Considering that investment involves high decision-making costs and high risk costs, he remains cautious about commercialization. The subsequent plan is to conduct small-scale internal testing based on sufficient maturity of personal use; only after the system can reliably help researchers reduce real pain points will he consider further productization, commercialization, and opening it up to more researchers or investors.
Joining the AI investment research elite incubation program also provides an external verification space for this system. Jack mentions that the inquiries from mentors and participants from different backgrounds, especially the question of "where exactly is the edge of this investment research system," prompt him to reassess the product's differentiation and iterability.
Independent research is most susceptible to path dependence: a person can continue to optimize their process but may not realize the blind spots. For him, the value of the incubation program lies in bringing implicit experiences into external discussions, accepting challenges, and then bringing feedback back to the system.
As more and more people envision AI as a tool that provides answers faster, Jack's practice offers an alternative path: rather than pursuing a "universal buy/sell button," it is better to first establish a research infrastructure that can explain evidence, accommodate counter-evidence, continuously update, and be reviewed.
Ultimately, the investment judgment still rests with humans; however, a rigorously designed AI workflow could enable this judgment to rely less on fragmented information, immediate emotions, and difficult-to-articulate intuitions.
Editor’s Note: This article is based on the recording of the interview, meeting minutes, and interview outline with Jack by TECHUB NEWS. The descriptions about his personal system, research methods, market cases, and subsequent plans are based on the interviewee’s statements and provided materials. The discussions about AI, semiconductors, digital assets, U.S. stocks, and other assets are for illustrative purposes regarding the research framework and methods, and do not represent recommendations, forecasts, or guarantees for any specific targets, industries, or market trends.
Disclaimer: This article is for information exchange and discussion of research methods only and does not constitute any investment advice. Securities, futures contracts, and virtual asset prices can rise or fall, and past performance does not represent future results. Readers should not solely rely on the content of this article for investment decisions and should evaluate it carefully based on their investment objectives, financial condition, and risk tolerance, consulting independent professional advice if necessary.
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
