OpenBB is an open-source investment research platform created by Didier Lopes (@didier_lopes). OpenBB grew out of a project called Gamestonk Terminal that Didier began working on shortly before the Gamestop short squeeze in January 2021. Today, OpenBB has evolved into an infrastructure platform that allows users to build extensions and access financial data with automation and customization.
Contributor is looking for a community manager! If you want to know more, shoot us an email at eric@scalevp.com.
Subscribe to Contributor on Substack for email notifications!
In this episode we discuss:
What Vice Media got wrong about OpenBB
Some major contributors to the project and the features or directions that they proposed
How a machine learning engineer from Bloomberg reached out about OpenBB
Different types of OpenBB users – students, retail investors, and other financial professionals
OpenBB’s exciting AI roadmap
Links:
People mentioned:
James Maslek (@jmaslek11
Artem Veremey (@artemvv)
Didier Lopes:
I didn't think of it as building a business. I thought of it like there's a lot of interest around these products. And so I was like, "This excitement means that there's probably more people that are willing to use their spare time to build this."
Eric Anderson:
This is Contributor, a podcast telling the stories behind the best open source projects and the communities that make them. I'm Eric Anderson. Today we have Didier Lopes, the creator of OpenBB on the show. Welcome, Didier.
Didier Lopes:
Hi Eric. Thanks for having me.
Eric Anderson:
You've created quite a stir in the open source community. Openbb is quite popular. Tell us what OpenBB is for the unacquainted.
Didier Lopes:
Yeah, so OpenBB is an open source investment research platform. So we started in Christmas 2020. I started working on this project called Gamestonk Terminal that went viral very early on. And today we are the number one open source project in the finance category with over 25,000 stars and 200 plus contributors and thousands of users.
Eric Anderson:
Yeah, it's amazing. Tell me about your background. Eventually we'll get into what OpenBB is and isn't. But I think your background's unique in that you spent some time in finance and software development, if I recall.
Didier Lopes:
So I actually don't have a background in finance.
Eric Anderson:
Oh, okay.
Didier Lopes:
So basically my background is engineering. I'm originally from Portugal. I did my master's in control systems at Imperial in London. And basically when I left university I was really excited about the field of AI, particularly AI self-driving cars. And I accepted the role as a software engineer for a GPS positioning company, working on the algorithms using C++ to develop code. And then I went into sensor fusion engineering. But I was really excited about AI, so I started learning Python in my spare time and I was sharing what I was learning on my LinkedIn. And then basically one day my previous maths teacher and statistics teacher from the Portuguese University reached out to me saying that he was working on his PhD thesis, which was around modeling and forecasting of financial time series using neural networks.
And he challenged me to help him because he knew the concepts but he wasn't into coding. And I was like, "This is perfect, because what I'm interested is in AI. I've been learning Python for a few months now and this is the perfect opportunity for me." So I started working with him on that and after a year I really enjoyed what I learned, but the thing is financial time series, just such a small thing within the all investment thesis when you invest. So after that, and because I had been working for a year, I had some savings at the time, so I was like, "Now I need to start learning more." So I started learning about fundamentals, learning about options, dark pool, technical analysis. And the thing is I started actually really enjoying what I was learning. I was using Reddit a lot to learn as well.
The thing is coming from an engineering perspective, everything I did at the time was optimized for efficiency. You do these programs, you use continuous integration, continuous development, and everything is fast. You learn the shortcuts and you do things fast. But then I went into research and everything is manual. And there's a lot of work that is wasted in a way that I cannot reutilize. So my typical workflow started at 30 minutes research, but the more I learned, the more big the workflow became. And there was no automation in that process. So I would've to open several dashboards, take a screenshot, drag it into a Word document, write some notes. And then if I wanted a different ticker, I would've to do all of it again. There was no automation on that perspective. And even the same ticker next week or two weeks after, if something happened in between, the analysis I had done didn't really matter anymore because it was already outdated.
So I was struggling because I had full-time job and I was like, "How can people stay on top of the field all the time when they need to do so much automated work?" And I looked for solutions and I didn't find anything. Even if I could afford a Bloomberg, it didn't really allow me to the level of customization that I wanted. And so I grew with that pain and I had my notebook. I started writing notes on what my process is and basically I was just repeating that process all the time. And this was during COVID, my flight got canceled, so I couldn't visit my parents, which was the plan, in Switzerland. And so I had a full week on my hands and I was like, "You know what? I'm going to start building this." And I went to GitHub to look for a project that I could build on top of, and I didn't find anything that suited my needs.
The closest was one was probably Yow Finance Repo, but it was Yow Finance Repo, so even though it's not an official source, it's unofficial, it didn't allow me the customization I needed, because it was focused on the data that Yow Finance gave. And so I was like, "Okay, I'm going to start this from scratch." And then the first problem I had is like, "Okay, now I get why companies don't really allow you to take control of the data and do whatever you want." It's because the data is expensive. And so most of these companies that I was using to take a screenshot and put on the Word, they only have display rights in terms of the data. And so that was the tricky part. I was like, "Okay, so how am I going to build anything because I don't want to pay for a lot of data that I'm not sure if I'm going to use in the future."
So I went to a few data vendors, website data providers for options, like financials, even scrap data at the time. And I understood what their business model was. And their business model is very simple. You sign up, they get your information, and in exchange they give you a free API key. And that free API key allows you access to a limited amount of data. And from my perspective, that's perfect, because this amount of data is enough for me to understand if there's enough value for me on it. And if I'm doing a lot of research and using these endpoints, I mean I don't mind paying for it or paying for a premium API key, because I'm taking the value out of this data. And then the interesting thing is that all the other data vendors have a similar model, but the data actually doesn't correlate with each other.
So if you're actually look into all of these different data vendors, they offer the free tier with different data sets. And I was like, "That's perfect. That's where I'm going to start. And from a technical perspective, it's just an API key. So if I get to a premium, then instead of having five years of financials, I may have 30. Instead of being able to do 10 requests per minute, I can do 100, because I pay for a premium." But it's very easy. And so that's the assumption of what I started. I called it Gamestonk Terminal at the time. I built it for two months on my spare time. I was super excited about it and then I open sourced it, because I thought that it was enough for people to understand where I was adding with it.
Which you could understand that I made this path convention that basically, okay, you have stocks here and then you go and you have a menu that is technical analysis, another one that is fundamentals, another one that is a discovery at the time, another one that was options. And then within all of these you'd have commands. And so basically you could use the command line interface because I was proficient in Python. So I did Python end-to-end because it was the fastest way for me to play with the data. And so yeah, I released it and then we went viral. That was the beginning.
Eric Anderson:
Going back to the data problem, you're saying that the incumbents, the existing alternatives charged you a high price in order to use the app, because you might consume lots of data and the data is expensive. But you're saying you didn't necessarily need all that data upfront.
Didier Lopes:
Kind of, so there's multiple data models. So for instance, let's say that you have financial data that you get and you want to display it. So it depends, first if you are just using for yourself, you probably don't need any license because you can argue that it's not for commercial use. But if you are showing it to a lot of users, then you need a commercial license and that's called a display rights license. And then if it's for instance, historical data, which is data that comes from the exchanges, you not only need to pay for the data vendor where you're getting the data, but you need to pay for the exchange, the feed.
And then basically, if you allow people to download it, so if on your app you have a button that allows you to download that data, then you need redistribution rights. And that's way more expensive. And so there's layers, but this adds up really fast and can get really, really expensive, because the redistribution, you could argue that someone could just download all the data and spin up a different app. And so those data costs are usually much, much higher. And Bloomberg for instance, allows this, but they rate limit you, so you are not allowed unlimited redistribution, if that makes sense?
Eric Anderson:
It does. And how do you address the data in OpenBB? Is it a bring your own API key?
Didier Lopes:
Correct. Yeah, that's correct. So you install the terminal locally on your machine and then you manage your API keys. And because it's meant for you to use for your own use case, you are fully liable for it. And OpenBB ends up just being the middleman architecture to have access to the financial data. So for instance, a common question that people would ask us, because after that we built like our SDK, the platform that you can build on top, and people would ask us, "Oh, so I can just set up my API keys and build an app on top and commercialize it." And I was like, "Well, from OpenBB side you can, but from the data side you can't, because when you get that free API key, you are signing up for the license agreement, you're signing up for it. And that license agreement doesn't allow you to commercialize. It allows you to use it for your own personal use case or evaluation purposes. But then if you want to actually commercialize, you need to pay for a premium service."
Eric Anderson:
Got it. Yeah, no, clever. So you allow people to sort that on their own, which means you can offer something more or less free without having to worry about a business model around the data. You do that, you launch it and it's just you and you push out this open source project during COVID, is that right?
Didier Lopes:
Correct. Yeah.
Eric Anderson:
I'm going to draw a little bit Didier from a talk you gave at Web Summit recently. In it you mentioned how important the name was Gamestonk. So I mean clearly it's related to GameStop, which was trending at the time probably.
Didier Lopes:
Yeah.
Eric Anderson:
I mean your initial users, what's the overlap with people who are excited about GameStop at the moment? A hundred percent?
Didier Lopes:
Probably like 90%, yeah. Because the thing is I've been pushing open source projects since I left university because I was learning on my own and they would get one, two stars and I was just on my LinkedIn. It was a very small reach. And I did the same for this one. And so when I actually pushed it, there was not a lot of traction. The thing is I learned so much from Reddit and because I was on the old GME saga, so I spent a lot of time there and understanding how the conspiracy theories that all these people had, I learned a lot on. And so I thought, "You know what? I'm going to put this on Reddit, because I learned a lot from these folks and they might find this platform useful to help them automate the way to do research." And so when I did that, that's when the terminal really exploded. We had like 4,000 stars in less than 24 hours, which is absolutely insane.
I mean, today probably that number is not this crazy because of all these AI projects gathering that amount of number, but back in December 2020, I think 4,000 stars at the time, within 24 hours was pretty impressive. And it was interesting because you mentioned about the GME correlation, because I put it on Reddit, we went viral there. Then someone commented, "Oh, you should put it on ACRE News." And I was like, "Oh, really? Okay, I'm going to try. I'll just copy paste."
Eric Anderson:
Yeah.
Didier Lopes:
And it went viral there too, and I was like, "This is wild." And after that Vice Magazine picked it up, they picked up the hype and they did a post on it. But on the post they called the terminal, if you actually read the article, it says that it's a terminal to do research on GME, which wasn't true. You could do research on GME as you could do on Apple or Tesla or any other tickers. But because of the name, they just assumed that and they wanted to come up with the news fast, so they just put it out there. And that also got a lot of visibility, but it was funny.
Eric Anderson:
Okay, so you will give some advice in your talk that you should do certain things before you go public, you should have this MVP, you should have maybe contribution things ready, a Discord ready. Did you have all those? I'm guessing, no.
Didier Lopes:
No, no.
Eric Anderson:
So these were learnings you developed over time?
Didier Lopes:
Yeah, I wish I had, because I didn't capture nowhere close to us as much of the people that wanted to sign up at the beginning as I could have. So for instance, the Discord came out of needs, because a lot of people started asking for it, and I was getting a lot of emails and pull requests and features. And I had an idea, but people didn't really know what my idea was, because I didn't really have a channel. GitHub is not really the place to be to have real-time conversation and discussion. And so that came a couple of weeks after because people were asking about the roadmap, and I didn't even have really a roadmap. I think I had the documentation was the markdown throughout the GitHub repo. So I had the terminal structure in terms of menus. It was organized into folders, and then in which folder on GitHub, you'd go in and you would have basically the documentation associated with what commands and menus you'd find within that architecture. But it wasn't prepared for scale, because I didn't expect to need that scale, if that makes sense?
Eric Anderson:
Yep, totally makes sense. So open source has historically been associated with infrastructure and people who are in the process of building things. Users of financial terminals haven't always been engineers necessarily. They've been investors and financiers. So did you tap into more of engineers who wanted to be investors or investors who were willing to code or one group or the other? I mean, what did the initial user group look like?
Didier Lopes:
Yeah, both to be honest with you. So one of the bets I've made as well at the time was that Python was going to be big. And so even when people pushed for building something on JavaScript as a front end, I actually wanted to keep it Python, because as I said, I think Python is going to be big in finance. Because at some point you're going to tap out in terms of what you can do on Excel in terms of automation, and so Python is the next best thing. And I think it was a good bet, because today you're seeing Python being integrated on the CFA program, and now most people from the financial world are either getting into Python or already have some Python experience to do some things. And so for me, that was important. And then that's one of the other learnings is that I didn't really have analytics at the beginning, so I only knew from people that were reaching out who they were. And there were a lot of retail investors from the old GME saga.
But also, one thing that I realized from the emails that I was getting is that there was also some people from JP Morgan and Goldman Sachs that were using the terminal in their spare time when they got home. Or students for instance, quite a few students were reaching out to me and saying that they had this coursework to work on and they just wanted to have access to these financials to basically finish their coursework. So there was a big breath, but it's interesting because in a way, so it was a common line interface, so the user experience is not as seamless as a web app. It is a bit more technical, which is the angle that you brought up. So when we talked more with these more senior enterprise financial investors, they didn't really like the interface. But then on the other side of the spectrum, you had people that absolutely loved it. And so we've always been between this love and hate. There was not many people that were in different to the terminal experience.
Eric Anderson:
You mentioned also creating content, responding to feedback. I think these are community development exercises. But you're a one man team at the time, and I'm curious how you go from writing code 100% of the time to... What's the split in the early days? It's half community work, half code, or is it just all community?
Didier Lopes:
No, yeah, it was half community half code would probably be a right answer. So I had a full-time job, so basically I would wake up a bit earlier before my job started. I would reply to the community. Then on my lunchtime I would do community replies. Then my work would finish at five, six for most of the days. Then I would reply to the community until dinner. And then all the time after dinner was spent working on codes, on features until, I don't know, two, 3:00 AM and then I would wake up, other day. The month after was really rough. I couldn't do anything personally. My weekends were in front of the laptop because there was just a lot of community and I didn't want to push people away, but yeah.
Eric Anderson:
Yeah, yeah. You've already expressed, you were like, "Oh, I wish I had a Discord up. I wish I had this up." So I can imagine the whole time you're like, "I need to get these things done while the people are still here."
Didier Lopes:
Yes, exactly.
Eric Anderson:
It doesn't sound like you went into this with I'm going to build a big business. And so I'm curious when you started to feel like, "I'm onto something, this could be a company," and what that looked like?
Didier Lopes:
So when we got the big traction for me, that was wild.
Eric Anderson:
Yeah.
Didier Lopes:
That was like, "Oh, okay, there's a lot of interest around this tool." Because starting a project, we think of it like, "Okay, it is just a star." But at the end of the day, someone that is a developer has to go onto your project and put the star, so that shows interest. It's not just seeing a website and then just closing it. Actually they need to star, because they want to show that, "Okay, I actually want to track and I'm interested in this." And so that first blow up at the beginning, I was like, "Okay, there might be something here." I didn't think of it as building a business. I thought of it like there's a lot of interest around these products. And for me, that was perfect because what I was interested is on building this for myself primarily. And so I was like, "This excitement means that there's probably more people that are willing to use their spare time to build this." And so that was the first reaction.
But then I think that when I actually thought that we could build a big business was when most of the PRs and feature requests and stuff, they were small fixes or an addition of a command. And I remember one day I woke up and I would check the GitHub first thing in the morning, and I saw there was a PR adding a full Forex menu. I had never traded Forex in my life and someone understood what I was aiming for, understood the architecture, and they added a full Forex menu with, I don't know, 10 commands with I think one or two different data providers.
And I was like, "Oh, well this is huge. Because this person very likely spent days working on these integrations and they had to read the documentation, understand I was working on the architecture." And I was like, "Okay, this could potentially be big because there's more people that are in the same boat as I am." And more proof than the star is actually someone spending this much amount of time or their family or friends time to work on this. And I think that's when I thought that, "Okay, we maybe have a shot at this."
Eric Anderson:
That would be super exciting. And Forex meaning foreign exchange, so trading in currencies around the world.
Didier Lopes:
Exactly.
Eric Anderson:
And I mean you were an investor then. You were building this in part for your own personal investments, taking a step back from software development. How has your investing hobby changed over the course of this? I imagine at some point you spend less time researching certain tickers and more time responding to the community and writing code.
Didier Lopes:
Yeah. So this was up and pretty fast. So I finished the thesis with the teacher and then I was only a few months spending time on research, but that months my pain was growing, growing. And then it hit Christmas, which was when I started developing this tool. So I stopped pretty much researching because I was so excited about building the tool first. And I was like, "Okay, I'm going to build these two points that I can now automate everything." And so I started building that. And then after two months, even my girlfriend at the time, my wife was a bit upset because we didn't have a lot of time to hang out. And I said, "Okay, I think the terminal now is in a good stage. I'm going to make it open source and now we can do more things together. And now I'm going to still grow the project, but I won't use all of my spare time to work on it. I'm going to balance it a bit more."
But then because of the explosion of people, it ended up not happening. I ended up not spending the amount of time that I wanted to do research, because there was so much inbound feature requests and pull requests to review. And people wanting to add new data sources, new feeds, because there was some scrapping data you'd need to maintain some that changed the interface in the front end. So yeah, I ended up not even using it straight away for research. I only used it after when we raised money and I had a team, then I went back to spending more time doing research.
Eric Anderson:
I came into this interview thinking that, because I'm late to discovering the project, it's got a user interface now it looks like an app. And there's a lot of open source apps these days. And I wonder if there's some debate about how important it is to be an open source if you're an application and your users aren't programming much. In your case, it sounds like your users are programming quite a bit. And your initial users were all using the CLI, so it doesn't quite apply. But as the project's grown, I'm curious what you think about the of open source for some of your less programming audiences.
Didier Lopes:
Yeah, so to be honest with you, what we soon realized is a lot of people that were using the CLI, what they were doing was exporting the data, because they either wanted to process it on Excel or they wanted to have access to it on a PNG that they could put on a report. And so even someone that used to work at BlackRock, what the person did before we did it, was they forked the project and they basically cut the CLI to basically just use the data, the model view controller architecture to ingest the data onto a notebook that they could then process the raw data and use Python to build. Because the command line interface is all Python, but you have an interface that you can use. And once you get the data, you can't really work on top of it, and this person wanted to work on top of it.
And so once you realize that we actually started working more towards becoming an infrastructure platform that people could build on top of. And so in a way, the terminal is just a wrapper around that platform where you can access financial data. So currently today there's the command line interface, which is we call the OpenBB terminal, but at the end of the day, it's a wrapper around the platform that gives access to these financial data connections. And last year we actually spent quite a few resources in making it modular enough so that people can build extensions to it. And it becomes like an ecosystem, where if you are interested on fixed income and you have a data source that you are integrated, you can just build that extension and now people that install the platform, they can add that extension to it and you can easily access that financial data.
And so we do have an application, but I don't consider the CLI as much as an application to pay for. I consider it more as an exploration tool to see what data we have accessible on the platform. Now, what we've been working on last year as well is on the Terminal Pro, and that is an application that basically leverages the platform, the data integrations that we already connected, and we showed those on the pro. And as I mentioned to you before, in these cases, the ones that we show, we actually pay for the right to show that data. And we pay for redistribution as well because we want the users to be able to download it, but also allow access to it through our Excel add-in. So in that way, the application is more closed source, because the Terminal Pro code and the Excel add-in code is not open source right now.
Eric Anderson:
Right. And part of that's related to the bundling of the data, that originally you bring your own key, but now you can bundle that into a commercial license?
Didier Lopes:
That is exactly correct, yep. So one of the things that we want to do is allow you to set your API keys in terms of a visualization, but in those instances, we'll tell you, "Look, this is your API key. It's up to you to have the rights with the company." And the reason why we do that is because some of the clients design partners that we are talking with, let's say that they're already paying for a data vendor, but the data is so niche it doesn't make sense for us to pay for all the users to have access to this data when these companies are the only one that is paying to this vendor already. So we can just bring the integration that is already built on the platform and we just say, "Okay, if you want access to it, you just add your API key and the data becomes part of a widget collection that you can bring to your dashboards, but you are responsible for that contract."
Eric Anderson:
So at what point did you change the name? And maybe as you talk about that, you can address the fact that the enthusiasm around investing has also changed some, the GameStop madness has waned. Have you viewed that from the perspective of your community? Has your community changed along with that?
Didier Lopes:
Yeah, that's a good question. So the name changed because we were raising money. So Gamestonk Terminal was until we started the conversation with our lead investor, OSS Capital, in specific Joseph Jack's partner. And so we wanted a name that was less Mimi, which made sense if we wanted to build a business. And when we raised money, it wasn't just me. So James, which was one of the first contributors of the project when I was having so much tickets being created and so many... I didn't have enough time. And James was one of the most active developers of the product in the first one, two weeks being out. And he saw it on Reddit and he was developing a lot, and he was reacting well to feedback. He was fast at coding. So I just messaged him, "Look, do you want to be a maintainer of this project with me? Because I actually don't have the time it takes for me to add features."
And so I didn't know anything about him. I'd never seen him, never spoke with him. So I was just like, "I just want help. Look, I'm desperate. I need help." And this guy replies to me and he says, "Yeah, sure." And I was like, "Oh, that's cool." And he started doing a lot. And then I got to know him and it turns out that he was finalizing a PhD in quantum physics, so a super, super smart dude. And it just happened that it was the perfect timing for him because the labs were shut, because of COVID, so he had a lot of free time. And so I got lucky.
But anyway, so after I made it open source, he was with me on this journey and we were talking even more than we did with our wives. And so when we raised money and we said, "Okay, we need to change name." And I mean, he was an investor on GameStop as well and so was I. So we said, "Oh, we should keep the tradition of naming the company with something related with where we are invested." And both of us at the time, were losing a lot of money on BlackBerry, and the BlackBerry ticker is BB. And then because of Open, because the code was open source, we said, "Oh, what about OpenBB?"
Eric Anderson:
I had other assumptions on where the BB came from. That's cool.
Didier Lopes:
Yeah, I can assume that. We actually have that on the website just to be clear about the story. We have some docs saying, "Okay, here's the origin story, and this is..." We've done that very early on to say, "Look, this is actually the main story." And actually to reply to your question around the community, there wasn't actually a very big backlash, I think, which you could think that could be, because Gamestonk Terminal feels like more retail investor, but I think that it wasn't because they understood that we raised money and that the project was going to be better over time. They understood that we're spending all of our spare time on it, we weren't making any money, not even thinking about monetizing that product anytime. So yeah, there wasn't a lot of backlash at all.
Eric Anderson:
Well, I guess more about how has your user base evolved? I wonder if some retail investors that were really active have maybe moved on to other things? And have you found more professional investors over time?
Didier Lopes:
100%, yeah. I think that on the terminal we have under 10,000 monthly users usually, but most of those are actually not retail. And I actually have this strong opinion that there's no big market for investment research on the retail side. So I think there's a big market when we talk about execution, like apps like Robinhood, they do really well because of this payment for the flow. So they are charging to you by the price that they give to a Citadel for clear [inaudible 00:27:49]. And so I think that model is incredible, because you don't see that you're being charged. It's Amazon when you think, "I don't pay for the delivery costs." But if the price is higher, does it really matter? Anyway, that model is different because you need it in order to do transactions. And basically you'd prefer to have zero fees on your broker than none. So that's a very successful model.
But in terms of research, people actually do not spend a lot of time doing research unless they are managing money from someone. And they need to explain, "Okay, why did you put this capital onto Palantir? Why did you put this capital onto... What's your thesis?" And you need to justify it because again, it's not your money. But what I see a lot of retail investors doing is like, "Oh, X told me to put money on Z." And so they do it. Or I saw a tweet from someone that has 100,000 followers and they put money. And so the research process is actually very limited.
Eric Anderson:
You mentioned James. Are there others? I imagine as you're building the company, you might've uniquely drawn from this quickly growing community.
Didier Lopes:
Yeah, there was someone else that started also really early with us, which was a director at Workday. So that person actually made the repo beam much more nicer to interact with. So I didn't have a lot of experience in terms of big code architectures. Because at the time, I had only I think two years of software engineer also. And so it brought the black linter, commit hooks, CI, CD, the repo became much better and even in terms of structure. And that improved a lot and invited more people to contribute, because the repo was more standard across. Because at some point you could see that the code that was written by me versus different people. And so we made it uniformalized and then we had to spend a bit more time on the reviews, but the code was more uniform.
So Artem was actually really important on that at the early days. And so when we raised money, he became one of our advisors of the company. And still today we talk with him pretty much on a weekly basis about progress. He also brought in one of our coolest features at the time very early on, which is the capability to create a template of a report to do research and then basically just run it using this paper mill solution from Netflix. And this was really cool, because he built it for himself as well to streamline his investment research process. So yeah, Artem was a big contributor. There were a few other big ones. Most of them, we ended up hiring them when we raised money, so now they're part of the team.
Eric Anderson:
I mostly think about contributions being for features and this idea that you could just bring process, contributing process and standardization is really interesting. And the fact that it could then unlock other contributions is also really valuable. Although I imagine you need a lot of buy-in, I guess from you. Those are big changes, I guess standardization is a big change.
Didier Lopes:
And there was backlash. There was one very active contributor that stopped contributing to the project because he didn't want that. He needed to spend time on the formatting. He saw it as being slowing him down early. But we are looking not just for the next week, but for the next years. It's okay now that it's 20, 30 people acting on the project, but when is 100, 200 then if someone adds, if there's no way to basically standardize that, it's going to be very hard to read the code. And because at the time we're thinking about, "Okay, if this ever becomes a business."
And so Artem actually did this the very nice way. He actually sent me a big email and he said, "Look, your project is literally what I've been looking for. I'm starting to use it for my own processes. But because I have 15 years of experience at Workday as a DevOps director, the project lacks the structure that I'm used to. And so I want to propose this, but this is a big change. And so I understand if you don't want to bring it up. But if you don't is okay, I'll just fork the project and I'll incorporate this to a different fork. But otherwise, if you want, I can propose these changes."
And I went through them and I was weighing the the pros and cons. The pros were so much bigger than the cons. And even when I looked at other projects, I understood that this is actually the right thing to do, it's just that I didn't have the experience to bring it up first. So I thank him for his changes and I said, "Let's do it. This is a really good idea and will put the project in a much better state for new people to contribute." And in hindsight, it was the perfect decision.
Eric Anderson:
The people you bring up, James, Artem yourself, they don't work on Wall Street. Maybe you could articulate for us what's the archetype of folks who are initially attracted to OpenBB? It sounds like grad students are a really popular group.
Didier Lopes:
So Artem actually had worked at WorldFront for four or five years. So he actually had financial experience. James was doing PhD in quantum physics, and a typical part of a quantum physics PhD student is finance. So he was actually looking for a job in finance, that's why he started contributing to the project. But yeah, so at the time I would say that we didn't really have a lot of analytics, so we had a lot of downloads and stuff, and we didn't even track that, because in hindsight, I should have done that much earlier. But we only introduced analytics way later on. And so I didn't really know a lot, only from people that I was speaking from the community. And if I were to do pools, there was students, there was a big, big, big amount of students. Then there was a lot of retail investors, but retail investors, they didn't spend a lot of time with it. And it was more of a toy project where they used it and then they shrunk pretty quick.
Because again, retail investors don't spend a lot of time doing research, and when they do, they pay $10 for a [inaudible 00:33:30] or a Fey App or something like that to just do something quick. Even most of the execution platforms like Robinhood, they allow enough of research that they need, just to basically see the stock price and then put an order in. They don't spend a lot of time. And then the other one, the other group that was actually very active was these analysts or consultants at these JP Morgan, Goldman Sachs. And even people at bigger companies like BlackRock or Blackstone were using it on their own on the weekends or after hours. And they were just trying to understand the project and how it was structured and how they could get access to the financial data to work on their own models in their spare time.
Eric Anderson:
That's so cool and makes sense. Professionals who had these tools at work and wanted them at home.
Didier Lopes:
So one person that actually surprised me a lot worked at Bloomberg. Because I was excited about AI, so one of the menus that I added early on was the forecasting menu. And so I basically brought in famous libraries like Nixtla to do machine learning and AI on of tops Financial time series. But any other data that you've had, so you could try to predict the stock price of Palantir using the S&P 500 Financial Times series using analyst estimates or any other data. And that was really cool, because it's all built in Python and Python and AI is the perfect language for that. And so this person reached out to me and she worked at Bloomberg, the senior machine learning engineer or something. And she was like, "Oh, do you mind if I work on OpenBB?" And I was like, "Oh yeah, sure. Well, let's have a conversation."
And I chatted with her and I was like, "Why don't you do it at Bloomberg?" They have this tool called BQuant, which for me is probably one of their best offerings. It's like this Python interface to have access to all of the data that Bloomberg offers to their BQL language. And the reason why she wasn't working on it is because the language that Bloomberg uses is very outdated. And so we're talking about I think Fortran or C++. And to do machine learning it's not ideal. Because today if you learn about AI machine learning, we're talking about Python is what people used to teach. And so she just wanted to use OpenBB because it was based in Python, and I was like, "That's fascinating."
Eric Anderson:
So that was where I wanted to go next was AI. You started OpenBB before the generative AI boom. What's the roadmap or your vision for AI in OpenBB? Sounds like there's already some forecasting in there.
Didier Lopes:
The forecast has actually been even before it was open source because I was already excited about the topic. But with the generative AI in specific, we actually give a really exciting roadmap, because we think that the fact that we have a platform that allows to bring AI as easily as it does, is a very big value added. And so I actually wrote about this on Twitter that I have this big prediction that I think that LLMs will become a commodity and all the firms will have either a fine-tune LLMs or even small language models to do very, very specific tasks. So they are trained on proprietary datasets that they may have internally or even connected with warehouse instances like Snowflake or Clickhouse or SuperBase and basically trained on their own internal data. And basically then, the analysts on the team or asset managers can interact with that proprietary version, which is just running locally. There's no outside requests because security is a big thing in the financial industry.
And so one thing that we are working on the pro is allowing you to bring your own co-pilot into that interface. So it's not just about the data that first we allow you to have access to, but we also allow you to bring any other data set that you want as a form of a widget. Whether you want to build the extension yourself using the open source architecture of the platform or whether you want to connect with the database, you can do that through the pro today. Even though the pro is early Alpha, you can already do that. But one thing that we are working on this quarter is actually being able to bring your own co-pilot. And being able to bring your own co-pilot, which you can interact from the pro interface, but also allowing access to all of the data that the pro is displaying. So basically you have a financial analyst right at your fingertips that you can have a discussion with when you are looking at doing due diligence on a ticker.
Eric Anderson:
Didier, we're approaching our time here. Anything you wanted to cover that we didn't cover today?
Didier Lopes:
We are now in early alpha on Terminal Pro. We have more than 5,000 people on the wait list. By the time this is live, we probably rolled out access to quite a few people. But yeah, I'd suggest everyone that listens to the podcast to give it a try, the Terminal Pro and Excel add-in and if they have any feedback, they can reach out to me on LinkedIn or on Twitter. I'm most active on Twitter under @didier_lopes. And yeah, anyone can message me. I'm always happy to have a chat.
Eric Anderson:
Any favorite stock tips you've got at the moment?
Didier Lopes:
No, because I kept doing research, but now we have so much things to do, I'm actually in the process of also moving to New York to be closer to customers. So I actually sold all my investments, and I just want to be fully focused on OpenBB because at the end of the day, that's my biggest investment. I spent all the time I have on it. And I felt like doing research and investing was distracting me slightly, even though I have a long view. So when I invest, I don't really look back, but just having that on the back of my mind that, "Okay, I need to check how this is performing," even with the Bitcoin, which is something that I was investing in. I just prefer to sell this for now. I'll do it after.
Eric Anderson:
Congrats on an exciting several years. And thanks so much for the contribution. This is a big thing that we all get access to now. Thank you.
Didier Lopes:
Thank you, Eric. Thanks for having me. It was a pleasure to be here.
Eric Anderson:
You can subscribe to the podcast and check out our community Slack and newsletter at contributor.fyi. If you liked the show, please leave a rating and review on Apple Podcasts, Spotify, or wherever you get your podcasts. Until next time, I'm Eric Anderson and this has been Contributor.
To the Moon: OpenBB with Didier Lopes