JUNE 27-30

JOIN NOW
Champions
of Data + AI

Data leaders powering data-driven innovation

EPISODE 20

Cloud-Powered Innovation

More and more organizations are becoming technology-powered companies and are embracing data and AI to fuel innovation and introduce new products and services. In this episode, Mike shares his thoughts on Nasdaq’s transformation journey and highlights some of the coolest new AI-enabled use cases. We’ll also discuss why the cloud is an inevitable certainty for all parts of financial services inclusive of the exchanges themselves.

headshot
Michael O’Rourke
SVP, Head of Artificial Intelligence and Data Services Technology, Nasdaq
Mike O’Rourke serves as senior vice president and is head of artificial intelligence and investment intelligence technology. He oversees the development of Nasdaq’s data and index business technology, as well as the advancement of using machine learning throughout Nasdaq.

Mike’s team combines proprietary data with advanced analytics and machine learning to produce sophisticated solutions for Nasdaq’s global customers. These efforts have resulted in several technology breakthroughs and patents for the use of machine learning in areas of market surveillance, advisory, and new data products — including Trading Insights and the Nasdaq Analytics Hub.

Mike is also responsible for the data technology that powers Nasdaq’s U.S. and global market data feeds, global index offerings (including the Nasdaq 100 and Nasdaq Composite), and critical market infrastructure, such as the UTP SIP.

Mike joined Nasdaq in 1999 as Director, Global Access Services. Previously, he held a professorship at Fairfield University and management roles within the Global Technology Services division at IBM.

Mike attended Fairfield University, where he earned a master’s degree in software engineering and a bachelor’s degree in computer science.

Read Interview

Alex Mysak:
Welcome to the Champions of Data + AI series. I’m your host, Alex Mysak, and I’m joined today by Michael O’Rourke, SVP of machine intelligence and data services technology at NASDAQ. More and more organizations at becoming technology power companies and are embracing data and AI to fuel innovation and introduce new products and services. In this episode, Mike and I explore the transformation at NASDAQ and get his take on some of the coolest new and AI enabled use cases. We’ll also discuss why the cloud is an inevitable certainty for all parts of financial services, inclusive of the exchanges themselves, and hear from Mike on how he got started in AI and some pertinent career advice he has for data and technology leaders. Let’s get started. Michael, it’s great to see you again.

Michael O’Rourke:
Thanks, likewise. It’s great to be here.

Alex Mysak:
So in order to introduce you to our viewers here today, I would love to ask you just a very high level question, first of all, as a champion of data in AI, how did you get into AI?

Michael O’Rourke:
Well, I started in my technology or started getting familiar with technology pretty early in life. I think my first brush was when my uncle had come home from the Navy and he was stationed out in Korea and he came back with this old IBM 8088 personal computer, of which no one in my house had any interest in. It’s this big box with just two floppy drives, no hard drive, that’s it. And I was enamored with it because I saw the opportunity that this could be used for games, for sure. And I don’t know if anybody… I’m dating myself here, but back then, no internet, no this… There was very little games out there. They sold these paper books in which you would open up and it would have the code for your game in it.

Michael O’Rourke:
And so I would sit there and you’re just typing the code for the game into the computer. And then inevitably, it’s like a thousand lines of code, which you would compare that to games today, like Skyrim or something has five million lines of code. These are very small games, but I’m typing them in, and then I learned the code. Essentially, in middle school by debugging the stuff that I was just copying out of books, and it’s just by troubleshooting and then started making my own games. And then, later went on to college for computer science, software engineering. And it’s funny because we had a group of friends that did our capstone project, which was creating a massively multiplayer online role playing game. So I had the beginnings of my educational career and the end of my formal education in technology started and ended with the pursuit of games.

Alex Mysak:
That is such a wild coincidence because, similarly, my father had one of the first IBM mainframes in the UK and I also at the age of six to eight was learning to code on visual basic with him, these basic Arcadian games. So I don’t think there’s probably many people in the world that actually know that story, but we do have that in common, very strangely. So that’s pretty wild as a first kickoff and we know each other and we didn’t even know that about each other, that’s why I love this series. So we’re going to talk a little bit about NASDAQ, where you work and one of the Databricks partners, and many would say today that NASDAQ is a technology company first. Do you think that’s always been the case or do you think it was a transition that took place over some period of time to get this new identity?

Michael O’Rourke:
Yeah, I think NASDAQ always thought of itself as being a technology focused company. Back in the 1970s, when it digitized the stock market, they bet that technology was going to be the future of capital markets, but we were an exchange first and a technology company second. And in 2000, I think it was 2007 when we merged with a Swedish company called OMX, that dynamic started to change because, we saw they not only ran their own markets for Sweden, Denmark, Finland, Iceland, but they then sold that software to other exchanges throughout the world.

Michael O’Rourke:
That really started to change the way we were thinking about our business, that, geez, we could have a lot more impact than just the technology to run our own exchanges. And then, in 2017, when Adena had took over, she firmly believed that NASDAQ should be a technology company that happens to run 27 exchanges. And really every action that she’s taken since that time has really reinforced that. And you’re starting to see where are our revenues, how we generate revenue is really coming from the fact that we provide technology solutions rather than the fact that we’re a core exchange, which is still an important part of our business, but it’s really about where the focus is.

Alex Mysak:
Yeah. And also, thank you. Thank you. And picking up on the couple of things that you said, you mentioned clearly the exchange business, some of the technology services companies, could you also maybe help our viewers understand the full umbrella that is NASDAQ today or the different suites of businesses that you have?

Michael O’Rourke:
Sure. NASDAQ is so much more than an exchange, although I think that’s what most people associated with, they see the ticker on the news and they say, okay, that’s NASDAQ. We have four core business lines. Our investment intelligence business line, which focuses on data analytics. We have our corporate platform systems which provide systems for public companies to really make it in the public company space, things to manage their boards for them to, suites of tools for interacting with investors. We also have our core markets business. That’s going to, more of our traditional business where we run 27 exchanges. And then we have our market technology business where we create technology solutions for both exchanges, brokers, and more recently, we have a big push within the anti-financial crime space with our acquisition of Verafin and we’ve acquired SMART some years ago for trade surveillance. So it’s a whole suite of technology solutions for the capital markets.

Alex Mysak:
And so clearly I knew that, but I think it’s always amazing to showcase because as you said, it’s really diversified out of some of the core suite into more of these services solutions. And we’ll get to a little bit more of that for your external facing customers a little bit later in this discussion. If you think about your role then at NASDAQ, as one of the thought leaders for financial data and financial intelligence at bio machine learning, what is it that the role that you and your team are playing throughout NASDAQ to drive projects, thought leadership and outcomes?

Michael O’Rourke:
Sure. In my area, we focus on data and AI and for many of our customers, we’re more than just a technology provider. They look to us as being their research and development in many cases. While we focus on data and AI, we provide solutions for really all of our different business lines. This year, we launched a product we called NASDAQ Data Link and the NASDAQ Data Fabric, which we are partnering with Databricks on. And these technologies allow firms to improve their whole ecosystems. From being able to have centralized the discovery of their data, doing data quality management, access, titlements, really providing APIs and interfaces them to really become data driven and create more intelligent products for their businesses.

Michael O’Rourke:
And then NASDAQ, we run the AI team there where we leverage all of these data technologies then to help our businesses create smarter products, enhance existing platform, and then find efficiencies within our business processes to create scale. For instance, in our corporate platforms business, we provide advisory service for public companies, so they can better understand their investors and how they stack up with competitors. And one aspect to that business is just understanding earnings report. Each year companies put out their earnings reports every quarter and historically an analyst would sit down, read the report and then they could pick out-

Alex Mysak:
[inaudible 00:09:51] do that for my job. Yes. Thank you.

Michael O’Rourke:
[inaudible 00:09:54] themes, right? And qualitatively assess the prevalence of those themes and the sentiment of those themes and then, they could express that to the client. But in this scenario, it’d be difficult for an analyst to do more than a handful of companies. They have to physically read these documents. We saw an opportunity here where we can make this better for both the clients and the analysts by using natural language processing. And here we use a large model called Burt. It’s a probably available model to read hundreds of thousands or hundreds to thousands of earning transcripts documents.

Michael O’Rourke:
And then we access a topic prevalence and sentiment at the sentence level. And then we aggregate that up to paragraphs and documents. And not only are we able to provide the traditional analysis, but then we’re also able to take a look and say, “Well, what are the overriding trends? What’s happening in these sectors? How do you stack up against all of your competitors?” So you end up with a much better solution. And then in the end you also have this new data set that you’ve created. So you’re creating this opportunity that you have compounding innovation later where now that I’ve digitized this information of created data, I can now come back and create new solutions that stand on top of what’s already been done.

Alex Mysak:
Yeah. This is an area that I, again, given I just mentioned part of my old world as a financial analyst equity salesperson and what we’re also seeing as well, and I’m sure you probably have some initiatives with this internally, it’s not just the… let’s call them formal public filings, but also the informal filings. So your Twitter sites and scraping these feeds because you actually saw, particularly if you look at areas where ahead of a political election and the candidates are making comments and the market moves a certain way based on Twitter ahead of anything in the formal channels.

Alex Mysak:
We’ve actually seen a lot of strategies around the NLP extraction of all of this text as well, to be able to supplement some of the more formal strategies, so it’s an area I’m quite passionate about. We see it replicate quite a lot. If you think about, you mentioned that Adena, your president wants to ensure that NASDAQ is a technology first, I feel that you’re very much there today, what do you think is the progress on that strategy? Do you think you’re fully there? Do you think there are any gaps and if so, what do you think those gaps or areas future growth might be?

Michael O’Rourke:
Yeah. I think that we’ve made quite a bit of progress in that area, but I do think this is an emerging area. I don’t think we’re in the final states like, yep, this is what AI has to offer us in the markets. I really feel that we’re just beginning the road where we see more and more AI, intelligent applications, not just in financial markets, but certainly within the capital market space, but I think more broadly than that. I think if you look, what have we done so far? What we tried to do is look at each of our businesses and what are some of the core offerings that they have and how can we enhance and make those core offerings better? And in which cases, in some cases, how could we create new business lines?

Michael O’Rourke:
Not too many years ago, we acquired a business called Quandl that does alternative data, just like you were just talking about bringing in Twitter data and things like that. We consider that alternative data since it’s not traditional financial data. And in that case, just to be able to participate with alternative data, you really have to have expertise within the machine learning space to be able to process those data sets and turn them into something that would be usable to a customer. Likewise, in the anti-financial crime space, it’s just expected now that in order to flush out fraud or look for illicit activities that you’re going to be using machine learning in order to help and assist that. A person can’t check every single cheque that comes through, every transaction that’s written. So it makes sense to be able to bring machine learning models that can look for nefarious activities.

Alex Mysak:
Yeah. And as you know, in the public domain, we’ve had even some of the regulators using us for exactly that kind of use case. So it completely makes sense for your own due diligence internally that you’ve got the equivalent, couldn’t agree more. Okay. So we’ve talked about a range of different projects, everybody has their pet project. What would be one of the use cases that you have been most proud of and the impact that it had?

Michael O’Rourke:
It’s always hard to pick one, but I think one that stands out in our options markets, for those familiar with options or not familiar with options markets, on the options markets we list strikes and strikes are basically a contract and buy or sell an equity at a particular price. And on any given day, these options markets generate between 40 and 50 billion messages across the different markets. So there’s just a huge amount of traffic and there’s about one to 1.2 million strikes that are listed at any point in time and every day or every week we see about a hundred thousand new strikes come into the market. So there’s just an enormous amount of information and traffic going on in the option space. And the issue is that that just creates a lot of complexity for exchanges, market makers and traders, when they’re trying to select what’s the right strike for my strategy.

Michael O’Rourke:
And also it creates issues where we’re just generating so much messaging that it’s difficult for systems to even keep up with all of this. So this is an area where we looked at and said, “This is a good problem for AI.” And here, we kicked off a project about a year ago, maybe just over that, that we call Strike Management and specifically for our Boston options market. And here we use something we call predictive control, a predictive control model, which really combines predictive models to understand the near term future, and then control model for making an effective decision. A good way to think of a predictive control model, if you were to look at the weather report and there’s, let’s say, a 90% chance of rain, well, you’re going to weigh, what’s the cost of being that being wrong versus the reward of you bringing your umbrella and make a decision based on that.

Michael O’Rourke:
But if there was a 5% or a 30% chance of rain, right, okay, I’m going to look at that and make a different decision. It’s very dissimilar here, what we do is, first we predict the likelihood that one of these strikes, one of these million strikes are actually going to trade. And then we also make a prediction about, well, how much traffic is going to be generated because this thing is listed. And then we try to create, make a portfolio level decision to say, okay, which strikes really should list that’s going to provide the most benefit to the market. That’s going to create the most amount of capture and actually have a chance of being traded. And so we just started deploying that this past year to start to reduce the amount of messages within the markets and reduce the number of overall strikes in the markets. So it’s a niche complex use case, but you can start to see how you can take some rather large problems in the capital market space and then use that to create efficiency that benefits the whole market.

Alex Mysak:
So, Michael, then looking ahead, how do you see your role evolving as AI becomes even more pervasive in our day to day lives? And what do you think it’ll mean to develop AI solutions?

Michael O’Rourke:
Yeah. Right now, we run a center of excellence at NASDAQ where we bring in the best data science talent. And then we look at where we can make the most impact within our different business lines. I think that if we’re ever successful, I laugh, I’m like, “Yeah, if I’m successful, I’ll work myself out of the job in a way.” Because I do believe that AI, if it becomes a core capability for every business, then we want to have dedicated teams that are embedded into the organizations to really drive all of these type of initiatives. I also think though that as data science starts to become… data science talent becomes more prevalent and easier to come by, that will enable this as well, because that’s, another reason why we have a center of excellence is that, data science talent right now is very hard to acquire and very hard to keep.

Michael O’Rourke:
So you want to make sure that you’ve got just the best talent working on the most effective use cases. But I do think we should see somewhat of a trend like what happened with mobile. At first mobile came out, this was the new emergent technology and everybody had a mobile group. And then as that just became part of how we do business, you don’t need a mobile group anymore. Every group has their team or their players that develop mobile. And I hope that’s what we’ll see here as well, is that every team will really have their data scientists and they will have data science, savvy, software engineers as well that can help implement the models.

Alex Mysak:
Yeah. And I was on a webinar this morning and actually stated, I think that the scarcity of resource, a human resource, so the skilled, no, skilled, sophisticated quants and data scientists and engineers is probably the largest bottleneck over anything. Even work from home, that our part of the industry and technology is faced in the last two years so we’re very much aligned in that thought. I’ve got a juicy question for you. There’s a body of people out there that are naysayers regarding cloud, and so what would you say around the success of your strategy being dependent on embracing a modern cloud data stack? So NASDAQ as an exchange and some of these banks out there, people say we’ll never get fully to the cloud so what would you say back to them?

Michael O’Rourke:
I would say that it’s an inevitability, at least that’s my prediction. We have almost all of our non latency sensitive systems, so think of our core trading systems right now are still running in the traditional data center space. But any of the non latency sensitive applications are generally running in the cloud already because that’s where the innovation’s happening. If you think about how do I stay competitive? How do I compete? You look at your rate of innovation and how fast can I innovate versus my competitors. And if you’re going to stay relevant, then you really, you have almost no choice, but to embrace cloud so that you can increase the rate in which you’re able to innovate. And recently we just made an announcement that not only do we have all of these other systems in cloud, but we are moving our core trading systems to the cloud as well. And so we expect over the next five, 10 years, whatever it might be, that we’ll be, almost all of our systems will be cloud based. And certainly that’s where we’re going to see the innovation is within the cloud space.

Alex Mysak:
Yeah. And I love you sharing that. Thank you. Because I think that will make some in the industries draws drop that, Amen to that I look forward to being on the journey with you to see that evolution. I guess also very similar strategic space is that a couple of the big major financial services companies we’ve seen Goldman Sachs and even NASDAQ with Data Link have become more than even just a tech company, you’re now looking to be more than just a data company, it’s actually a data services company. It sounds like this as well, given the emergence of data as an asset is going to become more of an emerging trend and theme and required set of products. What do you think about the strategy going forward for NASDAQ and the market more generally?

Michael O’Rourke:
Yeah. I think you can see the trend, you mentioned it. We made an announcement just a few weeks back about our new NASDAQ Data Fabric that’s being launched. Goldman made a similar announcement, which we think is fairly complimentary to what we’re doing. And really it’s because as firms are evolving in the cloud, they all need these data services. And the more of these data services that can be commoditized, then the more firms can focus on their differentiators. If they have to build data bricks before they can begin building the rest of their application, they’re wasting a lot of time and money and opportunity. And likewise, where we come in creating the NASDAQ Data Fabric, we have a whole set of tools that are specific for the capital market space. Let’s say, here are all the things that players in the capital markets are going to need, and let us provide those as core capabilities in which you can then now take those and grow your business on top of them so you don’t have to spend the time catching up on things that have already been commoditized.

Alex Mysak:
So Michael, there’s an increasing focus from governments on protecting citizens from the misuse of AI. How should organizations in your view become proactive with self-governing the use of AI and what does that mean from building competitive advantage?

Michael O’Rourke:
Yeah, I think this is an area where firms really should be looking at, when they deploy models, how they govern those models they go out there. There was a paper that was published by the University of Washington called Why Should I Trust You? And here they talk about a project where they were creating models to predict whether or not something was a husky or a wolf. And they created this model that was actually highly predictive. They could pick out huskies and wolves, huskies and wolves, but then when they went in to look at it and discover why did it make the predictions that it did? They found out what they had really built was an effective snow, no snow predictor. Where wolf pictures had snow and husky pictures did not. And so they really hadn’t built something that could recognize huskies versus wolves at all.

Michael O’Rourke:
And I think this is a good example of why firms need to focus, not just on whether or not the models work or not, but why do they work? And then, ask themselves the questions, what type of information is going into this and what type of information leakage could come out of that? And then what are the type of biases that this model could possibly introduce? And is that something that we want to take a chance on? So a lot of people are now starting to introduce governance committees for model deployment or at least have their businesses go through it and actually understand the models before they put them into use.

Alex Mysak:
So to close out, usually we ask for a couple of pieces of advice, if you will, please, Michael. The first would be, what piece of advice would you give to somebody starting out in their data career today? And then speaking more across to your peers, what piece of advice would you give to those that are looking to transform organizations towards data and AI?

Michael O’Rourke:
Sure. All right. For the first one, I would say focus on projects and initiatives, they’re going to bring about positive change in your organization. You really want to be focused on those things that matter. And not only does it give the chance, it’s very rewarding to actually do something and make a difference, but it’s also going to put you at the center of attention as well, get you noticed. And that can be very helpful for your career. So I would think about it. It’s like not all work is equal and you really want to focus your time and efforts on those things to [inaudible 00:27:23] change.

Alex Mysak:
Yeah. Bring a problem with a solution, they say.

Michael O’Rourke:
Yeah. The second part of it where your question was more about if you’re trying to transform your organization to be AI or data driven. And I think it really comes down to great AI comes from great data systems. And so if your data is disparate and all over the place, and it’s hard to access and it takes too much time, then it’s going to take too much time and be hard to create intelligence on top of that. So there’s no absolutely great AI group that has bad data systems. So you start with having great data systems, easy to access, and then you put really talented data scientists to work after meaningful problems. And that’s the secret sauce for having an AI driven group and AI driven company.

Alex Mysak:
Michael, thanks so much for sharing all of your insights on Champions of Data + AI. And thank you so much for being such a great partner.

Michael O’Rourke:
Thanks for having me. And it’s been great partnering with Databricks.