Here's the story of how we created one of the first tokenized hedge funds, shut it down after reaching an SEC settlement, and are now building an open source market making bot! This post is adapted from Daxia's recent podcast with Michael Feng, CEO and Co-founder of Hummingbot.
Table of Contents
- Feng’s road to Silicon Valley
- How did we start CoinAlpha?
- Smart contract meets hedge fund
- OMG, the SEC came after us
- Reflections on blockhain-based funds
- What is Hummingbot?
- What problem does Hummingbot solve?
- Why open source Hummingbot?
- How easy is it to use Hummingbot?
- General advice for startups
Feng’s road to Silicon Valley
After graduating from college in 2001, I started my career creating collateralized debt obligations (CDOs) at Citigroup and later at JPMorgan. At the time, most people hadn't heard of CDOs, but I was attracted to the field because unlike traditional investment banking, it was a chance to build something completely new.My job was to structure CDOs backed by subprime mortgages and other residential real estate risk, so for better for worse, I ended up working on some of the now-infamous transactions that contributed to the 2008 financial crisis.After the financial crisis, I was really disillusioned and depressed.
What originally drew me into finance was the use of math, statistics and financial engineering to create better products that were more optimal than what previously existed, but instead the opposite had happened. I had spent my entire career thus far creating products that cost millions of people their homes and jobs. It was a horrible feeling.I was fortunate to get into an engineering master's program at Stanford, so I quit my cushy finance job and went back to school. Meeting my Stanford classmates and taking classes in computer science and entrepreneurship really helped me make transition from finance to tech.My friend Max and I started a company called doxIQ, which used computer vision to extract tables and other structured data from PDF files. Eventually doxIQ was acquired by Nitro, a larger PDF company. After the acquisition, both Max and I joined Nitro, where I led product management and he led research engineering. In 2017, I caught the startup bug again and ended up leaving Nitro and started CoinAlpha.
How did Feng and his co-founders start CoinAlpha?
I remember Max telling me in 2013, “Hey you gotta look into this Bitcoin thing!” I remember initially thinking that it was all pointless and just a scam; clearly I was very bearish. In fact, I think I posted on my Facebook feed something like “How do I find a way to short this thing?”It wasn't until the Ethereum whitepaper came out in 2014 that I really started paying attention. I realized that smart contracts had the ability to put financial products like derivatives, loans and everything else on the blockchain, which could make a more efficient, transparent and open system.
Having experienced the 2008 financial crisis firsthand, I knew that one of the things that exacerbated the collapse was the fact that no one knew what their exposure to Lehman Brothers truly was.Every credit default swap is basically just a bilateral agreement between two parties, so when you have all of these private agreements between different banks, back-to-back agreements all chained together, there is no feasible way to trace anyone’s ultimate exposure to Lehman. Basically the whole financial world was frozen for a three-month period starting in about August of 2017.
What really got me into crypto was the belief that eventually smart contracts would replace paper contracts, creating a more efficient and transparent financial world.
That's what led us to start CoinAlpha in the middle of 2017. I was fortunate enough to link up with my friend Carlo who had taken a similar path with me. We had gone to college together, he also spent a lot of time in investment banking structuring derivatives and now he had done a tech startup in Asia. Our other friend Martin, who had previously founded another startup, was working at Apple as a machine learning engineer. But his passion was trading Bitcoin, so in his spare time, he applied his engineering and machine learning experience to building a bot that would try to predict where the price of Bitcoin and Ether would go and automatically execute trades based on that.We felt that investment funds were a good area to tackle because traditional funds are extremely opaque and inefficient. Small fund managers have a tough time getting started because the ongoing fixed operating costs of a running fund are so high that you need $10 million in capital before you're solvent as a fund manager. We felt that smart contracts would allow almost anyone, even if you only had a few thousand dollars, to create a vehicle that would allow you to earn management fees. So we decided to prove that it was possible.
Smart contract meets hedge fund
We created an open source protocol for blockchain-based funds called Fund Protocol. But we also felt that it would be hard to get people to use it if we didn't know exactly how it would work. So we decided to also create the first application of that protocol and started a hedge fund called CoinAlpha Falcon, LP. We went through all the hoops that normal hedge fund managers go through, like registering it with the SEC, issuing a PPM (Private Placement Memorandum), and establishing processes to vet and onboard investors.
But the biggest difference was that rather than having a traditional administrator and auditor, we used a smart contract.
I think we were one of the first tokenized funds, but because we couldn't accepting retail investors, we never marketed it as such. Since fund regulations don't allow people to transfer their shares freely, we felt that calling it a “tokenized fund” would attract the wrong types of people, so we just called it a blockchain-based fund and emphasized the improvements efficiency and transparency. By not having a human administrator, we were able to have a really low minimum investment of 20 Ether ($6,000 at the time) as well as daily liquidity for subscriptions and redemptions, which no other hedge fund offered.
OMG, the SEC came after us
One day, we received a letter from the US Securities and Exchange Commission (SEC) that said we may have violated securities laws by engaging in something called "general solicitation" of our hedge fund.As background, like most hedge funds and startups, we utilized the Rule 506-b exemption, which means that you can market to accredited investors or people you with which have pre-existing relationships, but you cannot engage in general solicitation in "any advertisement, article, notice or other communication published in any newspaper, magazine, or similar media or broadcast over television and radio." (source)
From our standpoint, we didn't fully grasp all the nuances around the general solicitation rules. We presented a lot of information in our website, blog posts, and media interviews because from our perspective, we were marketing the technology rather than the fund. We wanted our blockchain-based fund template to be something that the entire hedge fund industry would eventually adopt.While we flew to Washington DC and explained to the SEC what we were trying to do, they ultimately believed that given the information we had made available publicly, they had a general solicitation case. Given how broadly that rule is defined, we felt that a settlement was the best way to close the matter so that we could concentrate on our core business of being a technology company.
We learned a hard lesson: in finance, the “move fast and break things” approach doesn't work.
From this experience, I now advise other crypto startups to comply with the most conservative level of regulatory guidance possible rather than assume that you're going to fly under the radar because you're small.