Interviewee: Jocy, Founder of IOSG

Written by: Joe Zhou, Foresight News

After nine years of investing, we've gradually come to realize one thing: the hardest question to answer isn't "what kind of founder can succeed," but rather — assuming the track is right and the direction is sound, why do some exceptionally talented founders still fail to survive in the end?

Half the answer lies with the founder themselves, and the other half with the direction they chose and the timing of their entry. Having lived through four cycles, certain patterns have begun to emerge repeatedly — even though every founder's story is different, and each market backdrop has its own nuances.

But one conclusion has become increasingly clear: Successful founders each have their own brilliance, while failed founders share a striking resemblance.

Over nine years and more than a hundred portfolio projects, we've witnessed the rise and fall of countless Web3 entrepreneurs. Behind every failed investment lies a real cost in hard cash — ranging from millions to tens of millions.

To avoid repeating the same mistakes, I built a "Failed Founder Database." The purpose is simple: to prevent myself from stepping into the same river twice.

The capital market never offers a "do-over" option, but we can choose to turn others' pitfalls into our own signposts. Laying these failures out in the open is not only to improve our own hit rate, but also to help more entrepreneurs avoid unnecessary detours.

Six Portraits of Failed Founders

I have a habit: every quarter, I review each deal individually with my colleagues; every six months, the entire team conducts a deep alignment session; and at year-end, I pull out a list and lay bare all the successful and failed projects we've invested in over the past year.

With the first half of 2026 just concluded, we took this review opportunity to supplement and summarize a set of "Failed Founder Portraits." With this database, we hope to turn the pitfalls we've encountered into muscle memory, allowing us to steer clear of those fatal hidden reefs in advance.

After repeated validation across four cycles, these failure patterns have gradually crystallized — even though each founder's story is different and each market context varies, the underlying logic is astonishingly similar.

However, before we dive in, one thing needs to be clarified: the patterns below fall into two categories. One concerns founder traits — aspects related to a person's emotions, resilience, judgment, and self-awareness; the other concerns project structure — structural choices around token design and capital strategy. The former is about the person, the latter about the matter.

Founder Trait Category

These problems are rooted in the founder's own personality, mindset, and inner drive. They have nothing to do with technology or the track, yet they are often the primary culprit that kills a project first.

Type 1: The Emotionally Unstable

This is the most lethal type. When a project suffers an 80% drawdown, its community is under concentrated attack, and there has been no progress for three consecutive months, how the founder reacts almost determines whether the project can survive.

Failed founders get caught up in their emotions at these times — repeatedly justifying themselves on Twitter, erupting in internal conflicts with co-founders, or arguing with users in community groups. Successful founders, under the same pressure, are already deconstructing the problem and working on Plan B within the first week.

You don't actually have to wait for a drawdown to see this. It can be tested before investing: politely push back on their core assumptions during due diligence and observe their reaction. Some founders will seriously debate with you, standing firm where they should and making corrections where needed, remaining composed throughout. Others, when respectfully challenged, immediately become defensive or even counterattack. What the latter reveals is exposed earlier and more reliably than how they behave during an 80% drawdown.

Type 2: The Lack of Hunger / Safety Net Type

This type is easily overlooked because it isn't "conspicuous."

If a founder has a sufficiently soft safety net behind them — whether it's family wealth, a high-paying fallback at a major tech company, or a "it's okay if it doesn't work out" mentality — their choices during the darkest moments often deviate from the optimal solution. Entrepreneurship is a matter of life and death; without the foundation of "total commitment," it's very difficult to survive the cycle.

We once discussed a project at an IC meeting where the internal vote was deeply divided. It was a team that both Paradigm and a16z were willing to back. The founder had an exceptionally good family background and was personally an LP in several US mega funds (top-tier VC firms with over $5 billion AUM), all of which were also willing to support him. Looking purely at the investor structure, this was one of the most beautiful deals we had ever seen.

But that day, our debate over this project lasted until 1 a.m. In the end, I cast a veto.

The reason was that I believed what this founder wanted to do was far too challenging to human nature. His idea was to build a crypto bank for the African market, which required establishing a local ground team and execution force of over a hundred people. The founder himself had grown up in the US and China. To make this work, he would have to genuinely relocate to Africa, live and build the business there long-term, and be on the front lines. He repeatedly emphasized his determination during the meeting, saying he would plunge into the local African market without hesitation.

But it was precisely this kind of team that "looked right in every way" — top-tier institutional backing, a perfect investor structure, and a founder whose verbal determination was impeccable — that made us hit the pause button. As it turned out, although the project later successfully completed its TGE, it fell far short of their original vision of becoming the "neobank of Africa."

This is the same principle as the "execution machine." A team can score full marks on every dimension that can be graded: institution, structure, track record, determination, plan. But the most critical thing in entrepreneurship is often precisely the one thing that cannot be graded — whether there is an irreplaceable, non-negotiable fit between this person and what they are trying to do. When all the boxes are ticked, it's easiest to overlook this question.

We've revisited this many times since: the problem was never that he wasn't good enough, but that he was so good at everything that we almost forgot to ask the only question that truly mattered — would a person who grew up in the US and China really want to, be able to, and be willing to spend the next five years of their life on the front lines of ground promotion in Africa?

Type 3: The Unchecked Ego Type

The external manifestations of this type are often the "polished execution machine" or the "professor-type founder."

Let's start with the execution machine. Founders with exquisitely refined OKR systems, decks that look like McKinsey reports, and who list "execution ability" as their top strength — in our data, these individuals raise large amounts of funding but subsequently attract fewer follow-on investors and have poor exit performance. This is because they excel at finding optimal solutions for known problems, but the most common occurrence in the crypto industry is that the foundation shifts. A finely renovated house is far more fragile than a bare shell.

But one caveat must be added: whether being an execution machine is a problem depends on the track.

In already validated, mainstream directions, distribution, recruitment, and repeated execution are the keys to victory, and a polished, execution-oriented founder might be the best choice. The problem only arises in emerging, non-consensus directions — there, you need someone with more imagination and the courage to venture into ambiguous territory. So this isn't an ironclad rule, but a judgment of founder-market fit.

Now, the professor-type founder. Their technical understanding is usually the deepest in the room, and they deserve respect. But we pay special attention to two questions: first, whether they truly understand business and are willing to make compromises for commercial implementation; second, whether they are coachable — willing to learn and willing to change.

When a professor sees themselves as the teacher and the VC as the student, the project usually gets stuck in place. Technical depth cannot be equated with product judgment, much less with business execution.

We have also invested in founders with extremely deep technical backgrounds and a simultaneously sharp business acumen. The key isn't their academic credentials, but whether they treat technology as a means and commercial implementation as the end, and whether they are coachable.

There's an even more subtle layer: the safety net.

People from major tech companies and academic backgrounds often have excellent fallback options. Once a project starts to sink, they find it easier to return to the comfort of a big company or academic path. This doesn't mean they are weak founders, but it might mean they lack that hunger — the drive of someone who has no way back and must prove themselves. We value more that grit of "if I lose, there's truly nowhere to retreat."

Finally, there's a type reliant on path dependency — those from major tech companies or winners of the last cycle who directly replicate their previous playbook. We call this "doing this cycle's things with last cycle's methods." Dai Yusen recently made a similar observation: "It's very hard to beat ByteDance by playing by ByteDance's rules." By the same logic, the winners of the last era are the most likely to lose to the next era.

Project Structure Category

These problems concern how the founder understands the project's underlying architecture — what a token really is, how the capital strategy should be designed, and whether they have personally experienced the brutality of a cycle.

Type 4: Token First, Not Product First

This is unique to crypto and also the most dangerous type.

It differs from the previous categories — the problem isn't with the founder's personality, but with their choice regarding the project's structure. However, this choice itself, in turn, exposes what they truly regard as the core.

The typical manifestation is: keeping revenue and equity in a separate corporate entity, using the token merely as a fundraising tool, with token holders having no claim whatsoever on the real business cash flow.

We believe that whether a token is a fundraising tool or the product's skeleton determines whether a founder can survive the cycle.

The judgment criterion is simple: if the token went to zero tomorrow, would this project still have any reason to exist? If the answer is no, then the token is everything, and the product is merely its packaging.

Type 5: No Day 1 Exit Thesis

This is a principle our team has consistently emphasized — "Exit before Entry."

If a founder cannot clearly articulate on Day 1 how they plan to exit three years down the line (acquisition, token liquidity exit, or an IPO of the company itself), their fundraising narrative in front of investors will constantly distort.

It's less about the founder having to figure out the exact exit path on Day 1, and more about them needing to understand the sequencing of capital strategy and milestones: what does this round of funding need to prove? What data will unlock the next round? How might the return path for future investors roughly materialize? Early-stage projects are often emergent, and whether the ultimate exit is via M&A, token liquidity, or IPO may be uncertain — but "what is this round for, and what will carry the next round" must be thought through clearly.

Failed founders typically say: "We're raising funds for a bigger vision." Successful founders say: "I'm raising this round today so that I can catch the next round in 18 months, and the metrics for the next round are XX."

The Final Dimension

The first five types share a common undertone — they are all red flags.

Specifically:

• Founder trait red flags: emotional instability, lack of hunger/safety net, unchecked ego
• Project structure red flags: token first, lack of a clear capital strategy

But the sixth type is different. It's not a red flag, but a pricing issue.

Type 6: The Incomplete Cycle Experience Type

Crypto operates on a 3 to 4-year full cycle.

A founder who hasn't personally experienced at least one complete bull and bear market will severely underestimate their own fragility during their first bear market. This isn't a capability issue; it's an experience issue — if you haven't seen it, you don't know what that pressure feels like.

This has become a hard sizing policy for us: for early-stage teams without full cycle experience, the initial investment amount is capped at $250,000.

The judgment criterion is also very simple: What were you doing in 2018 and 2022?

But this type is different from the first five.

The first five are red flags, serving to help us identify "who to avoid." The sixth type is not a red flag; it answers a different question: "On whom can we place a bet, and how large should that bet be?"

Strictly speaking, lacking full cycle experience does not in itself constitute a veto — it's more of a pricing factor. Those who have lived through a complete bull and bear cycle often better understand how to manage volatility, handle community pressure, and navigate the psychology of downturns; but there will always be exceptional geniuses who haven't experienced a cycle.

So our approach isn't to pass outright, but to hedge with sizing: for early-stage teams without full cycle experience, the initial investment is capped at $250,000, and we add more only after seeing stronger evidence of execution capability.

Reverse the Failed Founder Portraits, and You Get the People We Like

Listing these failure portraits isn't about labeling people, but about helping us see more clearly: conversely, what kind of person is worth betting on.

Type 1: Obsession with the Problem.

The best founders aren't just interested in a problem; they are consumed by it. They've thought through the edge cases, user behavior, how competitors will react, and what the second-order consequences are. They aren't pitching you a product; they live inside that problem. This is the hardest positive signal to fake in a reference call — you can feel whether a person is truly spending 24 hours a day with what they're building.

Type 2: Second-Time Entrepreneur + Non-Consensus Foresight

I place special value on second-time entrepreneurs who have experienced failure.

The failure I'm talking about here is a setback at the project level where they've figured out the reasons, not the kind of fatal character flaws mentioned earlier. The two are completely different.

Failure itself isn't a big deal; the key is whether, after failing, you can figure out where you stumbled.

More importantly, they must have their own non-consensus thesis — not someone who drifts with Twitter trends and second-hand information, but someone who truly thinks independently and dares to make anti-consensus judgments.

Type 3: Strong Communicator + Controlled Ego

Communication skills deserve to be singled out because they are so critical. A founder needs to articulate complex ideas clearly — to users, investors, partners, employees, and the community. We've seen too many technically brilliant founders who write beautiful code but can't express themselves clearly. In the end, the project reaches a state where, if the one person on the team who can communicate externally is absent, the entire project goes silent.

As for ego, it's more nuanced than one might imagine.

So the key phrase is not "low ego," but "controlled ego": ambitious, but not delusional.

Category 4: No avoidance, no self-imposed limits, resilient willpower

In the crypto industry, you are exposed to the public spotlight and intense pressure year-round. Without willpower as a foundation, you simply cannot withstand the cycles and will most likely be crushed halfway through. Internally, we have a core framework called the "Key Question": the essence of early-stage investing is not stubbornly clinging to a thesis, but continuously iterating on the prior and posterior of every key question.

Put plainly, it is Bayesian thinking — constantly updating your judgments and beliefs based on existing information (prior probability) and new evidence (newly observed data), rather than rigidly sticking to an unchanging conclusion. You can have strong opinions, but don't be held hostage by your own opinions — when the reasons change, your judgment must change with them.

Category 5: Three hard metrics in the AI era: global perspective, Agency, and Taste

Since its inception, crypto has been the most globalized tech ecosystem — capital, talent, and communities flow globally in real time. In an increasingly fragmented world, founders who do global business from day one are themselves a scarce commodity.

Now consider AI. It can solve problems within a distribution, but only humans can pose original problems outside the distribution. So we focus on two things: Agency (the ability to proactively break through a situation) and Taste (aesthetic sense and judgment). These two are becoming increasingly valuable in the AI era.

Only when a founder's creativity and imagination have been validated first can AI become his amplifier, not his lifebuoy.

Three survival tips for entrepreneurs: The cost of launching a token far exceeds imagination; the entry ticket is millions of dollars

We have an internal habit of being extremely candid in our reviews, even to the point of being somewhat brutal. We ask ourselves: Why did we make that decision back then? Where exactly did that fatal mistake occur? What would we change if we could do it over?

Recently, during portfolio management, we gathered all the founders of our invested projects for a meeting and gave them three harsh but life-saving pieces of advice:

First, cash flow is far more important than narrative. The projects that survive this cycle will not rely on TVL or MAU, but on real, hard-cash cash flow.

Second, don't launch a token just for the sake of launching one. A token is actually a heavy liability.

In the last cycle, the vast majority of newly issued tokens broke their issue price. According to our internal statistics, the break rate exceeded 80%.

So we advise our portfolio projects: if you can avoid launching, don't launch; if you can delay launching, delay it. Why? Because the hidden costs after launching a token are far heavier than most people imagine.

We've run the numbers: the hidden costs after a token launch far exceed what most people imagine, including market makers, liquidity, compliance, exchange relationship maintenance, etc. In this cycle, that is a liability measured in millions of dollars. If a project hasn't even raised that magnitude of capital in the past few years, it simply cannot afford to launch a token.

Third, respect liquidity.

Sell at the best times, buy at the worst times. The valuation at which a project raises money today determines what performance it needs to deliver over the next three years so that the next round can take over the position. If it can't handle that, it shouldn't raise that money. Furthermore, sell decisively when token liquidity is at its best, and buy back to support your own protocol when the price falls below the issue price.

In today's market, many people are pivoting to AI, and many are fleeing Web3. Founders need encouragement, and practitioners need support — everyone needs to gather together as a beam of light. So we will continue to output research and judgment in this industry, offering the most honest advice.

How do we evaluate founders? Borrowing three frameworks from Zhang Yiming

These judgment criteria didn't come out of thin air. We've borrowed heavily from external references, and personally, I've been deeply influenced by Zhang Yiming.

I've always remembered one of his metaphors: empathy is the foundation, logic and tools are the middle layer, and imagination is the sky. Corresponding to investing:

• Foundation — Empathy: Can you treat people as ends rather than tools? Can you get along well with the team, attract top-tier co-founders, and demonstrate genuine leadership? This is what we mean by "emotionally stable, low Neuroticism."
• Middle layer — Logic and tools: Can you use tools effectively and engage in structured thinking?
• Sky — Imagination: Can you see things that "could exist but don't yet"?

Zhang Yiming's favorite interview question was: "On what important matters do you hold a view different from most people?" This question is used to test whether someone has a habit of independent thinking, rather than just being a "parrot of mainstream media."

More than half of people cannot answer it. We often borrow this approach when doing reference calls — if a founder cannot name three things on which they disagree with the consensus, they most likely cannot produce a non-consensus thesis.

Beyond that, Zhang Yiming values two other traits that we are also adopting: one is intense curiosity and hunger — a willingness to spend time envisioning things that "could exist but don't yet," rather than staying in already-validated domains making small, marginal improvements. The second is the ability for long chains of thought — being able to independently reason a problem through to the end without external feedback. In the context of the crypto industry, this means being able to thoroughly think through a thesis alone during 18 months without user feedback.

Final words

What we've summarized over nine years is not how to find the best founders, but how not to misjudge them.

But these methodologies are ultimately just tools. We have an iron rule internally: even a founding partner cannot casually push a project through the investment committee.

If it's a maybe, it's a no.

This sounds simple, but this is the entire secret to how we've survived through cycles.

Crypto's foundation is overhauled every three years. What allows you to survive through cycles is not one or two brilliant calls, but whether you can repeatedly press that "pass" button.