YourHandSucks
@yourhandsucks
@smsunarto @hosseeb Bankroll management & variance are the *real* write-up needed. Dreams are free; survival isn’t. Prepare for months of zero profit.
399K views

I have this bad habit that whenever I accomplish something, I'm compelled to write about how I did it.
We just launched Dragonfly Fund IV, a $650M crypto VC fund (at a time that half of the media seem to believe that crypto is dead--again). We have ~$4B under management, we have about 45 people across NYC, SF, and Singapore, and we're now one of the largest VC platforms in an industry that few have managed to survive in. So when a few people asked me to write about how we built Dragonfly, I thought: sure, why not.
Because, to tell you the truth, it would've been really valuable to me at the time we built Dragonfly to have a blueprint of how to build a VC firm. Nobody really tells you.
In reality, this will be useful to something like 0.01% of my readers, so I don't know that there's much point to writing this in all this detail. But fuck it. If you're thinking of building a VC firm--or if you're me from 10 years ago--this post is for you.
I first came into crypto VC when, in most people's eyes, it was already dead. In 2018, the ICO bubble had just popped, and the industry was in free fall. Most people I had entered the industry with had already left. But I believed that crypto was fundamentally here to stay--that it was one of those ideas that, once you'd really understood it, you couldn't un-understand. So when people ask why I'm so positive and optimistic on crypto, the answer is simple: if I wasn't, I also would've left a long time ago. It's too late for me. The optimism has metastasized to my hindbrain.
So when I met Bo and we decided to partner up together on Dragonfly, we were not expecting a lot of enthusiasm from the market. But every VC firm has to start from somewhere.
Lesson #0: Your first fund, you have to put your life on the line.
The lifeblood of a VC firm is money. To have a fund, you need to first and foremost raise money. If you have no access to money (or partners who can help you raise money), then you are not yet ready to build a fund.
To raise your first fund, you must first raise from your friends. Your boss. Your boss' boss. Anyone rich or prestigious you've ever met or have a passing connection to. If your reputation is not riding on your fund, you have not risked enough to make it work. Many first-time managers I meet make the mistake that they want to somehow spare their reputation if their fund doesn't work out.
This is a fantasy. If you are not all-in, you have no chance of succeeding. If you fail, yes, you will embarrass yourself and you will have lost other people's money--people who matter. But to have any chance at succeeding, you must use every resource available to you and make the first fund work. If you're not willing to do that, you shouldn't try to build a VC firm.
Once you get some initial capital from the people who have good reason to bet on you, you have to go upstream to the larger pools of capital: family offices (uber rich families), funds of funds (funds that just invest in other funds), and "institutions" (university endowments, foundations, sovereign wealth funds). These are roughly in increasing order of difficulty and prestige.
So OK, you're now pitching your fund to all these thick-pocketed investors. But as a first-time manager, what gives you the right to manage money? The answer: you need a clear, articulable edge.
Lesson #1: Find some niche or angle that you're better at than anyone else, no matter how small it is.
At the time we started Dragonfly, crypto VC was a small space. But even then, there were already several dominant crypto VCs: Polychain, Pantera, a16z. To us, they were juggernauts. It's hard to describe how dominant they were at the time.
So when we started Dragonfly, we had no shot at leading a deal. Nobody wanted our money. We needed an angle to get into rounds.
Like startups, a new fund must specialize.
The original idea was this: Bo was in Asia, I was in the US, and so we'd do this east-meets-west thing. Crypto is global, so we'd be the bridge between Asia and the US, and we'd help founders on both sides of the Pacific break into each others' markets.
This angle wasn't enough for us to lead deals. Nobody wants the "east-meets-west fund" to be their lead investor. But it was strategic enough to earn us a check in the round, and that was enough for us to start working our way in.
Lesson #2: Do the dirty work.
As it happened, this east-west arbitrage was ours for the taking. At first I wondered why no one else was doing this; it seemed so obvious.
I soon learned the answer: because it fucking sucked.
It meant we had to work punishing hours, every day, to run a firm across Asia and the US simultaneously. It meant more coordination, more late night Zoom calls, more language barriers, and much less of a normal life.
If you didn't need to do that to succeed, why would you choose it?
But we had no other path to success. So we put ourselves through it. We outworked and out-jetlagged other people.
Many people idolize VC as a genteel job, where you take the summers off and go skiing at quarterly offsites. We didn't do any of that shit, because we couldn't. We didn't have the money, the time, or the breathing room. The closest we had to wintersports were the crypto winters.
Lesson #3: Optimize like a startup.
Once you've established an angle and you've started getting into rounds, you now need to start building feedback loops. Investing is all about feedback loops, and the tighter they are, the better.
Most investors expect their startups to be rigorously numerical and data-driven. And yet they don't do this themselves.
You should be tracking everything. Your discussions, your misses, use AI to record and analyze your pitch meetings. You should be reviewing what the biggest deals in the industry were, why they were successful, crystallizing theories about what succeeded and why. You should be studying the great investors before you and the profiles that made them successful. Having AI makes this a lot easier today.
Most investors don't bother. They basically "vibe invest." Their success often boils down to how lucky their network happens to make them. That might work for a while, but luck is not a strategy, and it doesn't compound the way that ruthless optimization does.
Lesson #4: Talent is everything.
VC funds are terribly managed. I mean in the sense of corporate management. Really simple bread and butter stuff like 1:1s, mentorship, KPIs, clear lines of responsibility, communication, transparency, all-hands. For a while I wondered why so many VCs are managed so poorly. I eventually understood: VC does not select for management in the same way that companies do.
A badly managed company will eventually collapse or be outcompeted. But venture is a power law business, and usually only a few people are making that power law happen. So long as those few people are able to do their work, a VC can continue to survive even while being poorly managed. But being managed well is an edge in the long run. It allows you to retain your best talent and grow them beyond the original partners. VC firms famously suck at generational transfer and elevating existing talent, and partners are often terrified to hire juniors who are smarter than they are.
At Dragonfly, we were able to attract people and retain them who really should've been at better, bigger platforms than ours. But we took care of them. We gave them stability, voice, and independence. We showed them through our actions that we valued them, and now they are a big part of why we have outperformed.
Lesson #5: Be stupidly ambitious.
It still amazes me that most new VC firms, when you ask them what they want to become, they can't really tell you. "We want to invest in great companies and be our founders' best partner."
Barf.
That's like a founder saying "my goal is to grow the business in a way that maximizes shareholder value."
Have a real fucking ambition. Tell people that ambition.
When we started Dragonfly, our ambition was simple: beat Polychain.
That was it. We wanted to beat Polychain. At the time, Polychain was the preeminent crypto VC--they were the OG everyone measured themselves against. Many years later, after we started beating Polychain, I realized I had to upgrade that ambition. So it became: become a top 3 crypto fund. That animated us for a long time. We are now, by my lights, a top 3 crypto fund, so the most recent goal became to become the #2 crypto fund, and then the #1 crypto fund. Will leave where we currently are on that journey as an exercise to the reader.
Lesson #6: Fake it, but then make sure you make it.
Your first fund, you have no brand. So it's essential that you leverage what little social proof you have to immediately fake a brand.
Get into every hot deal if you can. Even if small. Just collect logos, and use those logos to collect more logos. We wrote tiny, meaningless checks into as many hot companies as we could in Fund I: dYdX, Anchorage, Starkware. The checks couldn't possibly move the needle, but being associated with those names, that little bit of legitimacy gave us a wedge to get into the next deal.
We called ourselves a research-driven fund. The "research" was just me writing “wouldn’t it be crazy if” posts and explainers. We called it "Dragonfly Research" and I just posted whatever I used to post on my personal blog onto there. At the time, that passed for research.
We told people we had the best connectivity in Asia from the West. And that was theoretically true. But in the beginning, we had no idea what people actually wanted from Asia (and to be fair, many of them were also not sure what they wanted), so we figured that all out as we went. Eventually it became much more systematic. But in the beginning, we just pushed the story as hard as we could, and figured it out in real time. It worked.
Lesson #7: The trend is not your friend.
Resist the siren song of trend following. Crypto, as all hot sectors, is full of stupid trends. NFT issuers, TCRs, P2E, chatbot tokens, VC-backed memecoins, and so on. Our best wins and moments as a VC came from avoiding the crazy stuff--Terra, Axie Infinity, Yuga Labs, etc., as well as doing big bets when others left the sector for dead. We did the Ethena seed soon after the collapse of Terra, and Polymarket before the 2024 election run-up.
Every cycle has a narrative that feels irresistible. A category that's so hot that every other pitch deck is about it, every conference panel is about it, and every LP is asking you about it. You will feel pressure--from your team, from your LPs, from Twitter--to pile into whatever the theme of the moment is. And every cycle, most of those themes turn out to be a waste of money.
The discipline required here is mostly psychological. When you pass on the hot deal that everyone is fighting over, and the next week that token is up 5x, you will feel like an idiot. Your team will second-guess you. You will feel like your competitors are pulling ahead. But the flipside of trend following is that you inevitably end up with a portfolio of "what was popular 18 months ago," which is the single worst portfolio construction strategy imaginable. Your job is to invest in what will matter in 3 to 5 years, and hot markets are reliably incapable of thinking that far ahead. Don’t let that be your fund.
Lesson #8: Own your distribution.
People used to say that a16z was a media business with a venture arm. That used to be a joke, but now it’s just reality
VC requires you to cultivate an ear for telling stories that resonate with people. It means you must build an audience and let everyone within your firm become a beacon.
Push your team to build their own brands, reward them for putting themselves out there. Personalize yourself and your partners. People want to work with people: unless you're Sequoia, a VC's brand has no resonance outside of individuals. VC is not like a company. It’s a people business.
Some firms literally discourage their employees from tweeting, which blows my mind. You must build an audience so that your name rings out. If you expect your founders to be able to master social media, why should they not expect the same from you?
Lesson #9: Cultivate power.
This is one of the last steps in transforming your fund from an upstart to a power player.
As Dragonfly grew in prominence, doors began to swing open for us without us even trying. Big exchanges, banks, market makers, projects we hadn't backed, they were suddenly interested in building a relationship with us and being helpful. At first, I foolishly ignored this. What a distraction, I thought. What is talking to market makers going to get us rather than talking to new deals? We should be talking to new companies, not old ones.
But it later clicked for me. You see, VC is the business of branding money. You win a deal by convincing a founder that your money is better than someone else’s. In reality, everyone's money is green.
Marc Andreessen once described it as: VCs are in the business of lending their brand and power to companies that don't yet have either. So as a fund, you must not only build a big brand, you must also cultivate power. Founders want to know your voice has sway. That you can get them in the room. That you know the right people. At the highest level of deals, this is where the competition happens.
As your fund grows, this is where you have to evolve from a pure investment shop into a platform. The best founders don't just want your capital; they want your ability to actually move things for them. At Dragonfly, we built out a platform team that helps with everything from token design to exchange listings to executive recruiting. None of this is glamorous, and it doesn't directly generate returns the way picking winners does. But it compounds. Every founder you go to bat for becomes an evangelist for the next one. And once that flywheel is spinning, your competitors can't just copy it.
Lesson #10: All of the money gets made in a few deals.
There is a simple matrix that describes VC investing.
Many hot deals are consensus right. Meaning most people believe the company is a winner, and it is in fact, a winner. These deals are usually fine deals, but you’re not going to make that much money from them, because they’ll be aggressively bid up.
Almost all of the money gets made in the non-consensus right deals. This is because these deals will be idiosyncratically underpriced, and that's where you're overwhelmingly likely to get the 100x+ outcomes.
Venture returns follow a power law, and the math is merciless. In a typical fund, the top three investments will generate more returns than everything else combined. This means the vast majority of deals you do won't individually matter very much. What matters is that you are in the one or two deals that define a vintage.
The implication is counterintuitive: your hit rate barely matters. What matters is how many big swings you take. Thus, you should be asking yourself on every deal: does this have a chance of being a fund-returner? If the answer is no, why are you doing it?
And the painful corollary: consensus deals almost never produce these outcomes. If everyone agrees something is great, the price already reflects it and your upside is capped. The truly generational investments are the ones where other smart people think you’re an idiot for doing it.
Lesson #11: None of this matters if you don't win the deal.
There are four stages to the VC value chain:
Sourcing => Selection => Winning => Supporting
Sourcing is the first step of what you need as a new VC firm. You must build an engine that actually finds deals.
Selection is what most people think is the most important skill ("picking"), but it's actually a pretty small part of the overall game.
Winning is the most important step. You can have the best deal flow in the world and the sharpest judgment, and none of it matters if the founder picks someone else. At the highest level of venture, the scarce resource is access. The best founders are oversubscribed and can choose whoever they want. So you have to give them a reason to choose you. This goes back to brand, to platform, to the relationships you've built and the reputation you've earned. All of those earlier lessons converge here.
Supporting is the last step, and it feeds back into why you win on sourcing and winning. Support is where your NPS score comes from, and why the whole loop continues. If you do right by your founders, they become your best salespeople. They refer the next great founder your way. They vouch for you in the group chat. This is a small and incestuous industry, and reputation travels fast. One pissed-off founder can poison a dozen future deals, while one happy one can open doors for a decade into the future.
Lesson #12: Venture is a "get rich slow" business.
You will see many, many people in this industry rise fast and become meteoric successes.
You must outlast them. Some will make too much money, some will get lazy, become convinced they deserve it. Crypto selects brutally for this. Every cycle mints a new class of overnight millionaires, and every cycle most of them vanish. Traders who 50x'd on some shitcoin retire to Lisbon. The founders who raised at absurd valuations quietly shut down. Eventually the tourists leave.
You are not a tourist.
It takes many years to measure progress. There are no overnight successes in venture. Most of the value in your funds is still tied up many years later. That means you are the embodiment of that famous NYT article:
That's OK.
Your job is to keep the ship steady. Flotsam, jetsam, high tide, low tide. You must be there for your team. For your founders. For the ecosystem. You are paid to be the long-term capital.
So be long-term.
Lesson #13: Raise when the raising's good.
As much as founders hate to fundraise, VCs have to do it too, and it's no better.
Raising as a VC is very different from raising as a founder, because the cultural dynamics are completely different.
I grew up middle class. I thought I knew rich people from when I was a professional poker player. Little did I realize--nope, totally different scale.
Fundraising is an art of its own, and it depends wildly on who you’re raising from. Fundraising from family offices is all about relationships. These are multigenerational wealth dynasties with their own idiosyncrasies, and it takes time to build trust with them. They are very social proof driven. Institutions and funds of funds are a different beast: process-driven, diligence-heavy, they are more won over by spreadsheets more than dinners. They want a track record, they want process, and they want to see a durable edge. You must learn to speak both languages to be a great fundraiser.
But in general, to successfully fundraise, you must be on your shit, have returns, and if you don't have returns, tell a really fucking good story of where the returns are going to come from.
And lastly, timing is everything. LPs buy high and sell low, almost always. So you should do the opposite. It sounds simple, but in practice it's agonizingly hard. Your best fundraising window is when the market is hot and LPs are excited--which is precisely when you should be cautious about deploying. And when the market is in the gutter, when everyone else is despondent, that's when your LPs least want you to be investing, which is precisely wrong. The best VCs learn to raise when the raising is good and deploy when the prices are good, and those two things almost never coincide.
These are some of the lessons I've learned building Dragonfly. I'm sure I'm forgetting some, and no doubt there are lessons I haven't learned yet. Building a VC firm is one of those things where the rules keep changing, every cycle brings a new cast of characters, and there's always unforced errors lurking around the corner waiting for you.
But the fundamentals are the same as they probably have ever been. Put your reputation on the line. Find your edge. Do the work nobody else wants to do. Hire people better than you and actually take care of them. And be patient. Venture rewards the people who stick around long enough to see what's on the other side of the cycle.
This is by no means the last word on how to build a VC firm. But this is the kind of thing I wish someone had written for me. I hope you find it helpful. And if you’re building something cool in crypto, I’d love to hear from you.
Disclosure: This is not financial advice, building a VC fund is hard, you will probably fail, but who knows, maybe you should do it anyway. Godspeed.
Reactions and replies to this article.
YourHandSucks
@yourhandsucks
@smsunarto @hosseeb Bankroll management & variance are the *real* write-up needed. Dreams are free; survival isn’t. Prepare for months of zero profit.
Stavros Michailidis
@stavrosmic4321
@hosseeb What's the relationship between lessons 10 & 11? If the non consensus bets are where the most money is made, isn't it easier to win those?
James Kramer 🔥
@hardtotelll
@realpennybags Might that one unbluffable player at the table represent a common challenge when former partners clash publicly?
Czar102
@czar102
@hosseeb Good article. No need to worry that very few people will create funds based on this. Most writing is for the 3 ultra-valuable people to join your network and propel the company. This writing also builds Dragonfly’s image as “the ones who made it”, which works on most readers anyway. But now I’m telling you stuff you alrrady know.
CryptoStoic
@cryptostoic404
@hosseeb Venture investing is a get-rich slow game. The money indeed is in the long - term. Guess I am part of the 0.01% then. @hosseeb What advice do you have for retail investors to adopt a long term mindset and not get cycled out when sentiments die down?
The BitWhale
@thebitwhale
@hosseeb Just tell me who pumped Enso token, multicoin or polychain ? Any other such projects ? Hardly seem delphi making any giga pumps on krw
pollo
@pollowinworld
@hosseeb Sir, everything you said applies equally to on-chain traders. This gave me a lot of inspiration. Thank you.
志成
@trillion_i_are
@hosseeb Use 𝕏 chat and buy $chat the 𝕏 chat meme coin 5umKhqeUvXWhs1vusXkf6CbZLvHbDt3bHZGQez7vpump Make it 1 billion Market Cap and everyone will know the @chat app @elonmusk Please make this happen it would be so amazing 🤣🤣 https://t.co/d8tSgYOmAe