|
Christina Qi Matthias Knab, Opalesque for New Managers: When Christina Qi co-founded Domeyard LP from her MIT dorm room with $1,000 in savings, the high frequency trading (HFT) fund grew to trade up to $7 billion a day. Today she runs Databento, a market data platform that has become one of the fastest-growing infrastructure companies in finance - profitable, with revenue growing 20-30% per month and near-100% enterprise customer retention.
The company has just raised a $97 million Series B led by NEA, with participation from strategic investors including DRW. In this conversation with Opalesque's Matthias Knab, Qi explains why some of the world's largest quantitative trading firms are tearing up their old playbooks and outsourcing data to a startup, why the SEC's competing consolidator rule shaped Databento's product from day one, how AI labs quietly became a major new customer segment, and what allocators should really be asking quant managers about their data infrastructure.
Matthias Knab: Christina, we connected in 2017 when you were still running your HFT hedge fund Domeyard. I heard you speak at a panel and I remember thinking, heck, that girl is clever. So I'm really happy to talk to you today, not only about the hedge fund days but foremost about Databento, the company you started in 2020 which now connects I believe over 24,000 customers offering market data from more than 60 venues.
Databento has grown rapidly since its public launch, you won the UBS Female Founder Award in December 2025, and recently spoke at NVIDIA GTC 2026 arguing that data access - not compute - now determines competitive advantage in algorithmic trading.
So, Christina, you ran Domeyard for nearly a decade. What did you see from inside an HFT fund that convinced you the bigger opportunity was selling picks and shovels - data - rather than mining for gold?
Christina Qi: The honest answer is that I sucked at mining gold! So I decided to take a jab at selling one step upstream instead. I knew how to build a company - I had done that before - so why not build a product and sell it to the industry? We were very familiar with data. We had seen the problem in real time, over a decade. It bothered us and kept us awake at night, and we decided: we should probably work on this.
Matthias Knab: So what was the opportunity set? What did you set out to do differently?
Christina Qi: Data is a really old idea - you could argue it's the oldest idea in the world. There were a thousand data companies before Databento. Even Bloomberg doesn't consider themselves the incumbent; they see themselves as a challenger. Before them you had Thomson Reuters, and before that, carrier pigeons and the Pony Express carrying stock price information. So the real question isn't "why data?" but "why now?"
We noticed that a lot of the data startups before us struggled to scale because they were very retail focused - they sold to individuals but couldn't sell to sophisticated financial institutions. At Databento, we said from day one: let's sell to institutions. Let's make this a truly sophisticated product. Of course, we welcome everyone as a user - you can be a student, you can be an individual - but we built the product to be suitable for financial institutions from the start, and we targeted them.
And how we built it: we took no shortcuts. A lot of data providers, a lot of startups, are a reseller of a reseller. They'll take LSEG data, rebrand it, and resell it. That's not how you build a game-changing business. We reinvented the entire data playbook from scratch - everything from colocation to building our own virtual private cloud. We own 5,000 CPUs, we run our own ASN. Our tech stack behind the scenes, how we process this data, is almost unheard of. We are not reselling someone else's data - we capture everything as close to the source as possible, directly from the source, and clean it up in the most sophisticated, correct way.
To the point where - I can't name our customers - but some of the largest high frequency and quantitative trading firms that used to colocate at the exchanges themselves and capture their own data are changing their playbook and saying: we want Databento instead. We asked them, "You used to colocate and clean your own data in-house - why would you use us?" And they said: "We're paying our quants over a million dollars a year. Not to be data plumbers - but to generate alpha. And you guys are the best data plumbers out there. There was simply no data plumber that was good before Databento." It's been nice to see firms that used to do all of this in-house now outsourcing it to us.
Matthias Knab: When you moved from Domeyard into the data business, did you already have the playbook scripted out based on your past experience, or was there a phase of trial and error?
Christina Qi: We actually had a pretty strong playbook, because of personal experience. For example, the time it took us to get data from one of the incumbent providers was 11 months and about 111 emails. That was deeply frustrating. We asked: why can't I get data self-service, off the shelf, instantly?
You have to understand the persona in our industry - the quants, the traders, the systems people. It's no secret that they can't stand the traditional sales process. Think about what buying market data has typically looked like: you go to a vendor's website, and there is no product to try and no price to see - just a landing page with some videos, telling you to fill out a form and "contact sales." Then you wait for a salesperson to call you, sit through a scheduled demo, and only at the end of that process do they quote you a price - a price that is completely intransparent, because you have no idea what anyone else is paying or whether you could have negotiated a better deal. For people who know exactly what they want to buy, that whole ritual is painful and unnecessary. In our industry, data is bought, not sold.
So we knew exactly what we wanted to build early on: make data easy to access, affordable and scalable - so students can access us, but we also scale up to the largest financial institutions. That's how you not only survive but thrive as an organization.
One more thing we had in mind on day one: we knew the SEC was about to pass the competing consolidator rule - we had a hunch it would pass. So instead of building around SIP data, we built out the proprietary feeds instead. We essentially built our whole product around a regulation before it came out. Playing 3D chess a little bit - but we were pretty solid on that.
Matthias Knab: You also pioneered usage-based pricing, which is pretty unique in this market.
Christina Qi: Yes, I think we were the first to bring true usage-based pricing, as well as what's called PLG - product-led growth - into the market data industry. Product-led growth means the product is essentially designed to sell itself. Again, the people we sell to in a way hate salespeople and the traditional sales process which is lengthy, cumbersome, and nontransparent. You have no idea what anyone else is paying. Maybe you negotiate a discount, maybe not.
For us, everything is online. Here's the data, here's the price, here's what you get. And rest assured, everyone is getting the same price. When people know it's fair, everyone feels better - nobody is getting special sweetheart deals. Of course, if you buy a giant package and pay two years in advance, you take bigger risk and you'll get a discount for that. But nobody gets free data.
The result: you can sign up self-service and just use our product. We've had customers pay us hundreds of thousands of dollars and I've never met them once. They don't know who I am - and that's exactly as it should be. They should never need to know who anyone on our team is. They should just say: "We love your product."
Matthias Knab: That's an interesting point - because at the same time you are very active and post a lot on LinkedIn. Do you have to do that, does it help the business, or do you just love it personally?
Christina Qi: Part of it is naturally my personality - one of my weaknesses is that I seek social validation. I'm 35 years old, should be too old for this! [laughs] But I have a sense of community and I naturally love posting - sometimes I'm quiet, sometimes I post a lot. And it is true that we gain customers who say, "I found you on LinkedIn, I found you on Twitter." I'm always surprised - and when I ask which article, it's always some drama that attracted them to the product, which in a way embarrasses me, but as a former hedge fund manager, I know exactly what they are talking about. But it does lead to unexpected customers. So social media has those two sides. If I could be offline, I would - but our product depends on it sometimes.
Matthias Knab: Let's talk about level 3 data. Delivering full level 3 feeds over the internet was considered impossible, or reckless, not long ago. What technical bet did you make that others wouldn't?
Christina Qi: The answer is, again, that we took no shortcuts. Databento is the only data provider that offers level 3 order book data over regular internet. Other providers talk about level 3 data, but you have to establish dedicated connectivity - which is tough, and they may not even offer that connectivity. We offer connectivity too, by the way, in case people need it - we do everything.
Our level 3 data is actually very small in size - we developed a format that compresses the data to be quite compact, which makes it easy to access. We rethought everything from the ground up to make the whole process easy from a user perspective. And our team - people call us a haven for ex-quants. We have people from firms like Citadel, Tower and similar shops working with us, and that has really helped us build a tier-one product.
Matthias Knab: So it's not just data - it comes with a significant level of IP you've developed. Now let's talk about AI. AI labs are now buying market data - do you see that happening?
Christina Qi: Yes, we do. Pretty much all the major AI labs are using Databento, starting from about a year ago. We were surprised at first - we asked, why are you using financial data, what's going on? And I realized there's an arms race going on in AI to build applications in financial services. Some are trying to build AI hedge funds, some are building charting and analytics tools, some are building risk management systems, some are building trading apps - the use cases are all over the place.
It's been an interesting new customer base. And for them it was a no-brainer. We always ask: why Databento, why not the incumbents? And they say: "The incumbents sell to Goldman Sachs. They're not selling to AI labs. We need a data provider built for the machine era, built for the modern day, for companies like us." What we're seeing is everyone seems to be moving in opposite directions at once: AI is moving toward finance, and the financial industry - the funds and asset managers - is trying to adopt AI.
Matthias Knab: As a former hedge fund manager, you're still observing what's happening with alpha. When every fund has access to the same clean, cheap, nanosecond-timestamped data - what's left as edge?
Christina Qi: It's almost like handing everyone the same starting point, the same tools. Now it comes down to: what can you build with those tools? And I think that's actually more fair.
I've thought a lot about fairness since starting my own fund - especially when "Flash Boys" came out and people accused high frequency traders of front running. What's fair and what's not? Surprisingly, data was never something I felt was unfair. I paid the same vendor the same cost every month and got the same speeds as everyone else. What felt unfair were other things - for example, there were lawyers who refused to work with us even though we were well capitalized and well funded, vendors who rejected us because we were young and, I get it, looked sketchy.
So if everyone has access to the same clean, cheap data, that's a great thing. It means markets are fair, everyone starts on the same playing field, and now you can just build - focus on trading and building good models rather than fighting to reach the same start line.
Matthias Knab: Data as the great equalizer - and from there, it's your intelligence and your models that make the difference.
Christina Qi: Exactly. It's not "I have insider connections, so I have something you don't" - that, to me, is unfair. I want a world where everyone gets the same things, and now let's go and use our intelligence from there.
Matthias Knab: Emerging managers tell us data costs are one of the biggest barriers to launching. What does a realistic minimum data budget look like today versus when you launched Domeyard over ten years ago?
Christina Qi: We've made data a lot more affordable, because we designed it to scale. If you're an emerging manager, you can literally start with usage-based data - pay as you go. Maybe you don't even know yet which data set you need or which exchange you'll be trading on. Then don't commit to a two-year vendor contract - just pay as you go and test out a few symbols on that venue. Once you're ready to commit, there are plans that get a lot cheaper as you consume more data, and cheaper annual plans as you add more people to your team. It's designed to scale the way a product should scale.
And it's not just the cost of the data itself. We're seeing PhD students doing really interesting quant research that simply wasn't possible ten years ago. Back then you needed a team at a sophisticated trading firm to access this kind of data. Now a grad student in a dorm room can write a thesis or do a homework assignment with it. I really like that - and credit to a lot of people who came before us who helped democratize this industry. Back then, we had to colocate, pay expensive providers, sign multi-year contracts that were hard to get out of, and commit to one data format. Very inflexible, very frustrating.
Matthias Knab: If you were allocating to quant managers today, what would you ask them about their data infrastructure that most allocators never ask?
Christina Qi: First, I would simply ask: who is your data provider? And then: where does that provider source their data from? Because if their data provider is a reseller of a reseller of a reseller, you're in big trouble - I can guarantee you they're going to go down for days at a time and not know what's going on. And sometimes the reason managers buy from those providers is to save just $100 a month - and then they go through the headache of days of downtime. It's a nightmare.
Data is not something you want to skimp on. You don't have to go too expensive, but if you go too cheap to save $100, you end up buying a lot of trouble. So ask their data vendor directly: where do you get your data from? The answer should always be "directly from the source, in colocation." So my recommendation is to take this a step further and due diligence the quant manager's data providers. You can even say: show me your colocation architecture, your colocation diagram - what's your latency profile? If they can't answer those questions, why are they a data provider?
Matthias Knab: You run all of this from Utah - far from New York, Chicago and London. Handicap or hidden advantage?
Christina Qi: First of all, I'm actually from Utah originally - I just call it home. I've never lived in a financial hub like New York, Chicago or London, and I love it. I was in London last week and I miss it already. But honestly, in this day and age I don't view it as either. Sometimes it's a handicap - there are events all the time in New York, and of course I miss those, and that's okay. But sometimes it's also nice to be in the quiet.
Matthias Knab: How do you run a hyper-growth company differently than you ran a hedge fund? What are some key lessons you can share?
Christina Qi: In many ways they're run relatively the same: you want to manage a good team, you want a good culture. The key difference is what you chase. At a hedge fund, you're chasing P&L. At a company with a product, it's about selling that product. I struggled at first to transition from thinking in P&L numbers to thinking in revenue and ARR numbers. Sales and marketing were a new territory for me, so that transition was difficult at first.
But it worked: we raised the $97 million round, the company is profitable today, revenue typically grows 20-30% per month, and we have 97% - almost 100% - enterprise customer retention.
In terms of lessons: being candid as a founder - sometimes too candid, sometimes too annoying - is actually good. Being vulnerable, being candid, is unusual in this industry. Quant is a cutthroat, cold, brutal industry - and we try to create a safe, welcoming, even loving space instead.
Databento does free community events - we did one in London last week. Anybody can come: students, professionals, everyone. We're not selling memberships, we're not even pitching our data sets - no strings attached. We watched the World Cup together, we just had fun. We create welcoming spaces for the community in an industry known to be brutal. And I find that really awesome. Beyond the job, beyond the salary - I just find it fun.
Matthias Knab: How often do you host those community events, and where?
Christina Qi: Every few months, usually in a different city. We've done Chicago, Shanghai, Tokyo, and now London. And if any city wants us to come, we're happy to host an event there - or co-host with a local partner.
Matthias Knab: Following up on the $97 million Series B led by NEA, how are you planning to deploy this fresh capital to scale Databento?
Christina Qi: The capital is primarily focused on scaling our infrastructure, expanding our asset coverage, and building out our global footprint. Over the next six months, we are expanding our exchange co-location presence to over 20 data centers worldwide, while significantly strengthening our direct feed coverage and technical operations across Europe and APAC.
To support the massive volume of high-frequency historical and real-time raw data captures, we are doubling our infrastructure capacity to over 100 petabytes of usable storage. We are also deploying capital to support a major new driver of growth: demand from frontier AI labs and tech leaders like Nvidia and OpenAI, who rely on clean, programmatic market data feeds for model training alongside traditional quantitative funds.
This interview is part of Opalesque's ongoing series of conversations with leading alternative investment managers and industry innovators. For more, visit www.opalesque.com.
|
RSS
