Markets, speculation, and risk. This is the Chat with Traders podcast hosted by Aaron Fifield. Boys and girls, welcome. What's up? So I trust by now you've had a chance to hear the two previous episodes with Aaron Brown. In my opinion, he was a real standout guest, someone we're worth listening to. If you haven't yet heard those two episodes, please add that to your to-do list. I think that's very important. Now on this particular episode, I have another brilliant guest for you, and that is the CEO of Trading Technologies, Rick Lane. You may recall Trading Technologies have sponsored the podcast and my events in the past.
But bringing Rick on the podcast was 100% my idea. This is not like a sponsored interview or anything funny like that, just to be clear. The real reason why I did ask Rick to come on the podcast was because of this. You know, if we're being real with each other, not everyone is cut out to be a trader, okay? And that's totally fine. Trust me, it's not a bad thing. And saying this, that also doesn't mean you cannot play an important role within trading. And for some of you, that means leveraging skills which you've already attained. And this is something I want to bring attention to.
Rick has never been a trader and prior to teaming up with his cousin, an interest rate trader, he hadn't even heard of a bid and offer. And if he had, he certainly didn't know the difference or the meaning of the two. He used to model combat scenarios for the military. He was also a product manager at Google at one point. And like I mentioned, Nowadays, he's the CEO of Trading Technologies, which is one of the leading futures trading platforms widely used by banks, hedge funds, prop traders, etc. So I won't say too much more on this because this is what we discuss in the first part of our conversation. It's really interesting. And then later on, we discuss some of the unique challenges to companies developing trading software. Are the latency wars driving to a close? Plus, Rick brings up something which will probably blow your mind. It certainly blew my mind. And then later on, Rick also shares some advice for anyone who may wish to play the role of a technologist in the world of trading.
All right, so let's get into it. This is episode 151. Here is my guest, Rick Lane. How was the FIA Expo the other week? Yeah, it was good. Busy as always. We had a lot to talk about. Norensic, our acquisition of Norensic was recent news. And so that drew a lot of attention, a lot of excitement. There seems to be a real interest in that space in general, but also the application of machine learning to it, but also to capital markets more broadly. So there are a lot of people stopping by the booth and wanting to dig into the Norensic platform.
Okay. And were you on any panels or give a talk or anything like that? You know, I managed to escape any speaking duties this year. So it was a relatively laid back conference for me. Oh, nice. Nice. Well, tell us a little bit about Nurensic because we were talking about that last time before the power cut out. Last time we were speaking, I think you'd only just acquired them. It was like a week or so ago. So what is Nurensic? Tell us a little bit about that and what got you interested in it? Sure. So Nurensic was a company founded about two and a half years ago in 2015 here in Chicago.
Their product and their innovation was applying machine learning, a subset of artificial intelligence, to the trade surveillance problem. I call it a problem because trade surveillance software has existed for many, many years. By that, I mean software that aims to identify when a trader or a trading firm is participating in manipulative trading practices or bad behavior in general, as defined by whatever governing body is defining that asset class or that trading area. It's been a problem historically because it's... As volumes of trading data continue to grow with the advent of automation, and now that just about every trader these days has some level of automation, finding bad behavior can be very difficult to do as somewhat of a needle in a haystack problem. But the real problem has always been false positives. And so if you're a compliance officer at a trading firm or at an FCM or at a bank, and your job is to monitor your firm or your customer's trading, generally...
Historically, your options have been rule-based systems. By that, I mean you encode if-then checks that basically say if you see this behavior, this action followed by this action, followed immediately by this action, then flag this as what's called spoofing or name your bad behavior. The problem is that it's very difficult to identify accurately traitor intent. Applying these relatively rigid rules really results in lots of false positives that The aim is to not let anything slip through the cracks and fly under the radar. And so you end up designing these rules to basically catch everything. And so it becomes the job of the compliance officer or the poor soul whose job it is to sift through these reams and reams of data of these alerts to basically determine were any of these actually bad behavior. The vast majority of them likely weren't. And so long story short, it becomes a very difficult problem to solve well. And what you end up having is with all this noise and all these false positives, it becomes hard for the compliance officer or whomever to really do at the end of the day what they're trying to do, and that's identify bad behavior.
So with the advent of machine learning and the application of it to this problem, the folks over at Norensic really caught on to something that allowed them to apply machine learning to really identify with far fewer false positives, essentially traitor intent. And so they're able to train this model on known. data sets that regulators have given them, that clients have given them, obviously anonymized data sets, but known instances of bad behavior. And they're able to feed this through this machine learning platform to essentially get smarter and be able to better identify similar patterns with varying levels of confidence on real-time data. And so it has resulted in what's called Norensic Score, which does just that. It's a platform whose reason for being is making it incredibly easy to take hundreds of millions of records of trade data within a trading firm and very quickly identify, sift the noise from the meaningful bits of insight and allow a firm or a compliance officer or even an individual trader to understand very easily their trading behavior to make sure that they're staying above board. Where we find this really compelling is it's still to this day the Norensic Score product.
is an installed on-site application. And there is an integration process and an installation process when a firm wants to say, hey, I'd love to have my trading activity monitored by this. Where we see this as being a really compelling bolt-on to what we're doing at TT is in making this really software-as-a-service the same way our TT platform is. So if you're an FCM, a large bank trading firm, or even an individual trader, and you want to have the protection of something like SCORE, it's literally just a click away and all of your trade activity without doing any integration, without doing any installation, we'll run through this process and you'll get your reports, daily reports, hope soon to be real-time reports. So really bolting what's still kind of a relatively high installation and integration cost platform like Norensic that does these really innovative, really game-changing things in the surveillance space and making it as easy as ticking a box for the trader is something that we think is really going to bring this type of technology to the masses. And I think it'll be good for everyone.
Okay. So by the sounds of things, it was certainly up until this point, it's been widely used by people in compliance roles. Is that going to continue to be the case moving forward now that you guys have acquired it? Do you feel as though it's going to be used or be of benefit to maybe even individual traders themselves who are using the TT platform, for example? So we do. So what I would say is actually the Norensic Score product has penetration in trading firms as well. And that's trading firms who may be doing that on the request of their FCM, of their broker. But it may just be a trading firm who...
just simply wants to make sure that they're taking every step they can. Trading firms can get very large and being able to police all the traders within a firm can be difficult. So we're actually seeing adoption kind of across the board. But to answer your question, we absolutely do think that even the individual trader would benefit from something like this. And now that it's as accessible, and we haven't really figured out individual trader level pricing yet, but suffice it to say, the commercial impact to an individual trader would be so small that it would be a no-brainer for them to turn this on. And by that, I mean not just monitoring.
as an individual trader, my own trading activity. But imagine now with MIFID, this thing we all love that's coming down the pike here a couple months, MIFID has certain algo testing and stress testing requirements for automated traders. Imagine now not only running your algo against known data sets to make sure that from a MIFID compliance perspective, your algo doesn't go haywire or doesn't violate risk limits, but also now imagine being able to see your compliance score for that algo. Because you might be building something that you're worried, you think it's above board and you think it's not participating in any of these known manipulative trading practices, but it's... Programming is difficult and building automated trading strategies is very difficult as you know, and as I'm sure a lot of your listeners know So making sure that maybe some emergent behaviors that become that come out of that algo aren't doing something that could that could cause a red flag So it's almost like preemptive compliance analysis before I even turn this out go on I just want to I want to rest assured that it's not doing something That's you know, that could be found upon so we see this as really having application even down to the individual trader and where we want to take this Beyond compliance and that's where I think long term This is really exciting to us because machine learning is you know a couple years ago It was that that interesting thing that very few people knew about just that it was
doing some cool stuff at places like Google and Facebook and Amazon. Now it's really becoming a household name. And I think in just a few more years, you're going to see machine learning and sort of deep learning that science of artificial intelligence really start to have applications throughout trading much more than just compliance. And I think this actually gives us a shot in the arm at TT to take this type of technology, which still Even though it's starting to be a household name, it still takes a fairly unique skill set and experience set. And this gives us a shot in the arm to really take this type of platform and break it out of just the compliance space and start to give traders who have access to all their data on the TT platform really valuable insights that would be very difficult to obtain in the traditional ways of analyzing trade data and market data. So we're excited both for the short term but also where we think machine learning can be applied down the road.
Just briefly, can you explain to us what is, was it MIFID or METHID? What's that about? So MIFID, and I'm not going to try to explain what that acronym stands for, but it's the new European regulatory, MIFID 2 actually, it's the sequel. It's new European regulations for trading. And a big component of that is ALGO, automated trading, registering and approving of ALGOs in futures markets, European futures markets. And part of that requirement is not only at a trading firm having an approval workflow. So if you're an ALGO developer,
and you've built a new version of an algo, you can't just go turn that on under MIFID. And in production, you need to have someone kind of green stamp it, say that this algo is approved to be run in production. And part of that is requirements around algo stress testing. So making sure that your algo, when run against, essentially backtesting, when run against known date ranges and for the given contracts that it's trading, that the algo doesn't violate certain constraints, that it doesn't exceed risk limits, that it doesn't over message. So that stress testing piece of this is the biggest amount of work for companies like TT, or at least for TT it was. It's actually a good thing, though, because backtesting is something that we've never really developed a backtesting solution at TT that is commensurate with the rest of our products. And so this is kind of forcing our hands. So we're actually really excited. It's a lot of work, but we're really excited about it. And this requirement actually becomes live January 2nd. So a lot of firms are scrambling, a lot of our customers as well.
to make sure that they're method two compliant. But this ALGO backtesting is a big component of that. So who does this apply to? Surely this doesn't apply to independent traders who are just trading their own account from home. I believe it does. I believe any automated trading, regardless of how large or small the firm is, even an individual trader, is subject to the ALGO constraints and the ALGO regulations around stress testing and what have you. Right. So this is only in futures and only in European markets? That's correct. The Markets and Financial Instruments Directive is what it stands for.
That's correct. It is only European and it is only derivatives, I believe. Right. Well, if anyone's interested, we'll dig up a link and post it in the show notes for more information. I suspect your listeners who are active automated traders trading European markets are well aware of it. But if they're not, the clock is ticking. All signs are pointing to that date holding. So the clock is ticking. Okay. Okay. Right. Well, let's hear a little bit about your backstory, Rick. You have a very interesting background. You actually, prior to trading, you were working with the military. Can you tell us a little bit about some of the things you were doing there?
Sure. I was actually working for a U.S. consulting company whose primary client was the U.S. Department of Defense. So the various branches of the military, the Defense Department, the CIA, the NSA, the FBI, so on and so forth. I worked in a modeling and simulation group there. So we essentially built war games, everything from campaign level.
War game simulations. So you know if we went to war with country XYZ what would happen? And if it was in the year 2025 and and we had these capabilities and they had these capabilities sort of what would happen? And so we you know we would we'd build a computer simulation that would Obviously not spit out the right answer because, you know, these were all very wild theories and guesses. It was more about facilitating the exercise itself. And so, like most militaries, I suspect, the U.S. military partakes in war games just to keep their mind sharp, just to think through some of these thought exercises. And so, our platform, the technology we built helped facilitate that. So, everything from that type of war game all the way down to very granular analysis around combat modeling, around how terrorist networks grow and are manipulated, things like that.
In a lot of ways, those skills that I learned and the types of clients that I had, and I did that for about three years, in many ways similar to my experience coming into trading. I originally came into the trading world because my cousin, who was actually one of the largest interest rate traders at the time, this was around 2005, he saw an impending change in the interest rate markets and in the futures markets in general. He saw the inflection point where automated trading... was really going to change the dynamics of trading in general. And so he was a floor trader. So he basically convinced me to leave this world of wargaming to come out and help him evolve into an automated trading shop and sort of evolve with the industry. And so I did that. I took him up on this offer and came out and learned the industry for the first six to 12 months, which was one of the more humbling experiences of my life. But then started to build our own internal trading system that let us try to stay ahead of the curve with respect to automated trading. And he was right, obviously. I remember when I came out to visit him before I accepted the offer.
In the fall of 2005, the floor was still a madhouse. It was just still craziness. He brought me down. It was one of the most vivid memories of my life. When I came back and started in April of 2006, so about five or six months later, you could hear a pin drop down there. It changed incredibly rapidly, as everyone knows. The rest is history. The reason I say it's similar in a lot of ways to what I was doing in the defense world, in the wargaming world, was because...
In the wargaming world, I was building really complex technology to do really complex things. So applying graph theory, for instance, to terrorist network modeling. And there's lots of really interesting science and math around graph theory, around how to look at a network and identify key players or identify key channels of communication, that sort of thing. Well, this technology is being used by two, three, four-star generals in the various armed forces. They're not technology experts. They're not graph theory experts. The challenge was really taking very complex technology and presenting it in a way that people who didn't necessarily understand the underlying technology could still gain insight from the analysis that it was doing for them.
back in the trading world now, fast forward to a year or two years on the job, I was the only developer there. It was me and a bunch of other traders, guys, that were still down on the floor. And it became sort of an untenable situation for me because I was the only one writing code. Everyone else was down on the floor, and I was up in an office, and we had this intercom system. So anytime some market imbalance occurred, a big order came in, or someone saw an opportunity, they'd yell up through this intercom system.
And as quickly as I could, I would write up a program that did exactly what they needed to do. They'd say, you know, we want to lean, we want to work this leg. When we get filled, turn around and do this, this, and this. If we don't get filled, then do this. And I would frantically try to string this together as quickly as I could, hating every minute of it because they wanted to turn this thing on in the production markets, you know, immediately. And everyone in my body said, well, you know, software always has bugs and we need to test it. And so there's always this kind of conflict and tension between me, the main person producing the technology, and the traders. And so I basically said, screw it. I'm going to build.
platform that lets these traders who know nothing about programming essentially control their own destiny, allow them to build their own trading algos on the floor. They all had these kind of handheld Windows computers. So I built a visual interface that let them essentially put these blocks together. And these blocks all did something that these guys understood really well. We had a block that would make markets in a certain order. We had a block that would pull prices in from a certain instrument. And they could string these things together in interesting ways that let them essentially build their own algos on the fly without me having to be in the loop.
and do it very quickly and also do it very safely. And so in a lot of ways, the same kind of lesson learned that I had in the wargaming space was paying dividends in this space too because I was having these traders who would profess that they knew nothing about programming, essentially programming, although they didn't know it. And so it was just a really interesting thing. Anyway, just to catch you up to sort of where we are today, we realized we had something with this visual algor development, something unique. Let me jump in here. So I have a few questions just before we get too far ahead of ourselves. Around the time we came onto the floor,
or started working with your cousin, I should say. But just one question prior to that, these types of things you were doing with the working for the military or contracting to the military, this sounds like very complex, advanced type of modeling and work. How were you able to test the accuracy of your models? Is that something that you had to do? In some cases, we did. So in things that were testable, we did. And I probably need to be a little careful about what I talk about in terms of what we're actually doing. But put it this way, some of the modeling we were doing, we knew what the right answers were or should be. And we knew the constraints. And we knew, based on a certain set of inputs, what the output should look. And so we were able to test that type of model very well. And we had enough data and enough know-how and subject matter expertise in-house to test that.
things so that the more uh the more uh high level models where we were really testing you know if this capability is even possible 30 years from now um what will it mean for our naval forces well first of all we don't even know if that technology it will exist 30 years from now so we certainly don't know if what the computer tells us is going to happen is is realistic or is correct so there's really this this spectrum and a sliding scale between these types of modeling exercises that we had to do and obviously it was depending on the client uh and depending on the problem they're trying to solve but in some cases we could in some cases testing that was just completely impossible sure okay I'd be interested to hear a little bit more about your cousin's motivations for wanting you to come and work with him. So how come he had the foresight to want to automate some parts of his trading operation? Why did he feel as though it was necessary to do so in order to stay relevant? What if he didn't do that? What did he think would happen? What if he continued to just keep doing what he'd been doing? Great question. I mean, what gave him that foresight? I don't know. I think he probably saw enough.
He probably saw enough instances where clearly something that could move faster than a human was getting an advantage over him. He probably saw the inflection point of technology in that space just in general becoming more mature. These were guys and girls that were down in the pit trading all day every day. But nevertheless, technology was becoming more and more prevalent, whether it was the handheld computers they were using in the pit, whether it was advancements and clear investments that the CME and the Board of Trade were making in their electronic trading platforms. I think he just really, at the end of the day, realized that technology had kind of disrupted just about every industry save the capital markets and trading. And it had already done so to some extent in the equities market. And I think he just was...
thoughtful enough to realize that it would happen to him as well. And he didn't want to become a dodo bird, essentially. So he didn't want to become irrelevant. And so it started out slow for us. It wasn't fully automating his trade, because he was a really, really good trader. And in a way that even today, even with today's incredible machine learning technology, I think it would be really tough to replicate with a computer. So this was really more about augmenting his trading initially, just giving him that extra little bit of edge in the firm, that extra little bit of edge, even not necessarily in terms of trading, but even looking at post-trade analysis.
overhauling the tools that he had to look at his trading from yesterday and figure out what went wrong or what went well. So he didn't realize he had to fully automate on day one, but he knew that's where things were going. And we took baby steps to get there. Right. And when you came to work with him at the time, did you have any interest in markets or did you just kind of see it as something that might be a cool technology project to work on?
That's a really great question. So I went to Georgia Tech undergrad for college, and there was zero exposure to capital markets or to trading or really to finance in general. It's an engineering school through and through. That's changed, actually, I'm happy to say. But back 15 years ago...
I came out of school with zero experience. I didn't know what a bid an offer was, believe it or not. And so, no, I came in about as green as someone can. It was really the latter. It was me seeing that there was a really challenging industry, that technology hadn't quite taken a foothold. And I thought it would be an interesting challenge for me personally. So I was obviously right. But no, I came in knowing about as little as you can know. And so the first six months or so was essentially boot camp for me, just getting up to speed and knowing the mechanics of the markets themselves that I could even hope to start to contribute in some way.
on the computer side. Now, before you started telling us a little bit about some of the things you began to build and develop once you started working with your cousin and the other people who were part of his operation, other traders, your goal from the sounds of things was really to try and make things simple and easy to use, right? Yes. Okay. Do you think there's a downside? This might be a bit of a strange question, but do you think there's a downside to making things too easy for users? Like, are they maybe doing something which they don't really understand?
I don't know if that was a good question. It's a great question. It's one that we've asked ourselves. Let me just say, the reason I ask this question is because sometimes I get people emailing me who are interested in getting into automated trading or they're like, I've got the strategy, I want to be able to backtest it. And I know there are some software platforms around which make it very easy for you to be able to do that, especially more simple strategies. But I think that can also be dangerous as well. Someone doesn't actually take the time to understand some of the pitfalls of backtesting and how the results can be misleading, etc.
So by making software really easy for the end user, is there also a downside to that? Does that make sense? It makes total sense. And the answer is, in certain cases, absolutely. And I think what really matters, if you're going to give someone something that really makes what used to be very difficult very easy, particularly in a space where there's so much to lose if what they do is broken or wrong or has unintended consequences.
In order for that to work, I think you need to also offer more than just whatever that tool is. In other words, one of the things that we realized early on, because trust me, the initial users of what we call the Algo Design Lab or ADL of this were people who had no business, and my cousin will laugh if he hears this, who had no business building ultra-low latency trading strategies without applying any testing or any quality assurance processes to. But the reality is we knew that going into this, and we made sure that all of the surrounding aspects of the technology. So whether it's a scripting language or whether it's a visual tool like ours was, we put really in the user's face what is going to happen when they run this algo, coupled with really sufficient and granular risk controls around making sure that they can control their automated trading in a way that even if...
they do something dumb even if they're you know there's a flaw in their logic or or you know they back test something and and think that the back testing is indicative of what's going to happen when they run it going forward that they can still mitigate problems when they arise and so it's a great question it really is and it's something that we've always taken seriously because we are giving you know this is a really bad analogy but it's the first thing that pops into my head um you're giving someone a loaded weapon and it's a very powerful weapon and you want to make sure that they that they use it well and that They follow all the safety procedures and that it's incredibly hard to do something dumb with. And so we made sure and really have continued over the last now eight or nine years as this product has been on the market to make sure that we are guiding users through it so that, for instance, you can build a bit of logic that we will visually call out to the user and say, hey, we need to get your attention. We think that what you just did here could lead to.
endlessly lifting the offer or hitting the bid. Or we think that you could get into a loop of placing orders and violate your risk limits. So we're actually able to highlight these common pitfalls. And frankly, we do this because these are things that I did when I was writing code when I first came out to Chicago to build these trading strategies. These are all the problems that I had and mistakes that I made.
we encoded these common pitfalls so that the trader or the user or the developer who's using the platform can really rest assured that when this thing says it's ready to go, it's going to do what they intended it to do. But nevertheless, it's still obviously possible to make mistakes. But it's a great question. And I think you just can't be cavalier about it, whether you're the user or whether you're the technology provider. You need to understand that you're giving something that is very powerful to the end user, and you need to make sure that they are confident and feel confident that they can control the experience. Yeah. I did like your analogy of...
So it's like handing someone a loaded weapon. I think that's great. Now, initially you developed a software for internal purposes. At some point you decided to start offering the software out to other people as well. And I find that kind of interesting because obviously, like I said, you developed it for internal purposes and it was working well for your guys. Why did you want to share it with others?
Great question. So this was probably middle of 2008. So the world was changing. And we realized that with the Algo Design Lab, we had something that, while it was an advantage to us, it allowed our traders to control more of what they used to depend on me or others for. We also knew that in the hands of others, we didn't view that as cannibalizing our existing business and our existing trading. We also knew, frankly, that...
Trading Technologies, the 800-pound gorilla in the room at the time, didn't have something like this. And so we thought, well, TT is a really successful company. They are in the business of building tools for traders, and lots of professional traders have it. Why can't we supplement our existing trading operation by selling this software commercially? And unlike TT at the time, this was built by a trading firm, built by traders that's in active use today. And in fact, when we launched, we went to FIA Expo, the big annual trade show that actually just wrapped up, as you mentioned, a couple weeks ago.
We went and at our booth in our demo, we had live trading on my cousin's account. We had algos running all throughout the day. We were so proud of that that we brought these giant speakers and these subwoofers that every time we got filmed, it would shake the room that we were in. And it worked. People came over and they couldn't believe that we were letting these things run as we were engaging in conversation. And we're like, oh yeah, by the way, the people who built these algos, they've never written a line of code in their lives. And so we thought we had something unique compared to TT. So we said, let's give it a shot. Let's try to sell this thing commercially because we didn't think it would jeopardize our existing trading. And so that's why we initially set out to –
to build the commercial trading arm of that. That's very cool. And it actually reminds me of something that, are you, do you know of Minaj Naran? It sounds vaguely familiar, but no, I don't believe so. Yeah, he started TradeWorks. He's since left that, started Manor Partners now, well known in the high frequency space. I interviewed him, must have been earlier in the year. And that's actually something he said. He said that sometimes a lot of people don't realize the actual technology and a software that's developed like inside some of these trading firms actually has more value.
than just the P&L that it can extract from the markets. You know what I mean? Yeah, so that's a really interesting point. Now, obviously... this software was acquired by TT and you've come on at TT now. So I'd just like to ask a few questions, I guess, more around fintech and general kind of things about the technology space. One of those being, well, I think this might be a good question to kick things off. Having your background, you know, you come from a developer's background. Does that have significant benefits for someone who's in the role of CEO at a technology company? Like, are there great benefits to the company to have a CEO who comes from a development background? Well, I'm biased. So I think it does.
I think it does. And look, I think it also brings its own set of challenges as well. You know, I've always long thought very analytically about every, you know, most of the problems that I've faced in my career. And in a lot of cases in this role, you have to kind of take a step back. But I really, I think it does. And I think it does for a couple reasons. Number one, I mean, our job, our reason for being is delivering and building and deploying technology. And so I think understanding the constraints in which we're able to do that well. And by that, I mean, you know, the rate and the pace at which we can innovate.
helps me forecast. It helps me keep priorities in line where they should be when you're constantly trying to balance a global sales force who is hearing feedback from the trading community and wanting us to build, build, build, and build new features and build new access to new markets and build new access to new asset classes with the realities of how long it takes to actually build these things and do this work. I actually still contribute code to this day. It both helps me balance out my sanity to some extent, but it also keeps my finger on the pulse of how the machine that we have that builds and that continues to innovate, how that machine is running. And so I think having that insight is really valuable for us. And I've obviously surrounded myself with a room full of really great leaders and great people who can help me with things outside of technology as well. But no, I think it helps a lot. And I think it just keeps me closer to the people and the processes that we have and helps me keep track and really, at the end of the day, plan better. And I think that's half the job.
is planning and prioritizing and trying to figure out the best way to apply the limited and scarce resources that we have. Sure. Yeah. And I realized as soon as I asked that question that, of course, you might be a little bit biased there. But no, awesome answer. Why do you choose to stay involved in fintech rather than technology in another industry? Like, is there anything which excites you specifically about fintech? Absolutely. I mean, and it's why I came back. So we glossed over a little bit. So as you mentioned, DT acquired the company that we co-founded back in 2010. I spent about two years integrating.
the Algo Design Lab into TT's platform. But then I left. I actually went to Google outside of this industry, about as outside of this industry as you can get. And pretty much immediately, I felt the draw to come back. It turns out that less than a year later, the former CEO called me up and said, hey, I want you to come back in the CTO role. And really, I want you to rebuild this technology from the ground up. And at the time, it was a no-brainer for me to come back. And the reason for that is because...
at least the trading world, and fintech is an incredibly broad term, everything from banking to payments to trading and what we're doing in our little niche corner of the world, but at least in our space, there are problems that are being solved that are truly not being solved anywhere else and in ways that aren't being applied anywhere else. Working at a place like Google, you're working with a lot of really smart people who are solving a lot of really challenging scale problems and data problems and latency problems, but those pale in comparison to a lot of the scale. and latency problems that we have to solve in our world. And so I think just coming into work every day, knowing that you're doing something that a lot of people have failed at doing and trying to solve problems that a lot of people have failed at solving is exciting in and of itself. But I think our space in general, and in particular, I should say, really poses some problems that are unique in the technology world. And that's just fun to do. Are you able to maybe give us a couple examples of some of those problems and just to bounce another question off that? And you don't even have to be specific to T here, but what are some of the greatest challenges for companies who are developing trading software in current times?
I'll start with the first part of that question. You've probably had two dozen people talk about this on your show, but it bears repeating again. Our technology, in lots of cases, likely isn't as fast as the fastest elite prop trading firms in the world. Nevertheless, we measure success in terms of latency by anything less than 30 microseconds from tick to trade, which is just a ridiculously small number, and trying to shave even five microseconds from that. is such an incredibly challenging problem that not only requires an incredible mastery of software development and really eking every last bit out of traditional CPU and computer, but hardware expertise and networking expertise and that race to always be incredibly fast so that our clients and our end users have that great advantage, but do so at global scale.
No one else in the world is solving these problems. Google obviously has global scale, but when you do a Google search, they're happy if they deliver that answer to you in 30 milliseconds. Well, 30 milliseconds is a lifetime. You may as well be down on the floor trading if it's going to take you 30 milliseconds to make a decision. So it's those types of problems that really, generally, in the technology world, you can either have really, really fast.
or lots of throughput at global scale. Trying to do both is a problem that, you know, still we invest a lot of money and time in solving and, you know, keeps a lot of people up at night because it is an incredibly difficult problem to solve and to stay, continue to innovate in. But then the second part of your question, you know, what are challenges generally for people entering this market or even in this space? Everything I just said, you know, if performance is, and there are lots of platforms out there that don't have to shave microseconds off, that do really great things. And so that's not always a problem for everyone in this space. But nevertheless.
It's an industry that changes, I suspect, rapidly than most and as unforgiving than most in terms of You know for us, and I think for anyone who's trading multiple markets or building technology that connects their users and their clients to multiple markets, you've got different regulatory regimes. You've got different exchanges who trade at different hours of the day, who use different technology that lets you connect to them, who have different rules around how orders need to be formatted. And they're always changing. And exchanges have changed their platforms, upgraded their platforms, have unplanned outages, have different holiday schedules because one's in Asia-Pac and the other one's in the US.
And that problem in general, if I were trying to be an upstart technology company in this space, that stuff would keep me up at night. Because that's, I think, why there is such a barrier to new entrants in this space. This industry and this world in general is growing and changing at such a rapid clip. And it's not slowing down. And it's only becoming more complex. And layering on making sure that your technology is there when your clients need it most. Because this is an industry that forget building technology if you're a participant, if you're a trader.
You need to know that the market is accessible and at your fingertips when you need it. These are all just really challenging problems to solve. And even if you're not trying to shave microseconds off, in some capacity, you have to solve them. And so I think that really presents the toughest kind of set of challenges for other providers. Yeah. And just as you bring it up there, what is your take on the whole latency race? Obviously, it's been a big thing over recent years, but are we almost at a point where it's literally, and I don't like to really use this word, so I'm going to use it loosely, but impossible to get much quicker? Are we almost at the end of that trend?
So what I'd say is some people are, and in some instances. In other words, just pick the most sophisticated prop trading firm or hedge fund that you can think of, and whoever that is probably has – in fact, I know they have some really, really smart people doing some really, really interesting things in terms of performance for very specific trades. They can embed that algo on silicon. They can put it in an FPGA, basically a programmable hardware device. They can even embed algos in network switches.
But those are few and far between in terms of the firms that have the resources and the capability to do that, but also few and far between in terms of the applications where that type of trading can be applied. Then you take it a step back and look at a company like us. So to answer your question, I do think a select few are about as close to zero as they're going to get. Obviously, they measure their latency in nanoseconds, and you can get smaller than nanoseconds, but they're really butting up against the speed of light constraints more than anything else until we can go faster than speed of light. we'll have a problem getting faster. But what we're also seeing is technologies that were five years ago only available to a select few become commoditized. And so where TT stands is we're not embedding algos on network switches, but we are leveraging some commodity, but even some what we would probably not classify as commodity because it's slightly higher end and slightly more specific, but still lets us and our client base.
remain incredibly flexible. In other words, if you've built an algo and you embed it on a network switch and you realize, oh crap, I did this one thing wrong and I need to now go change it. Well, good luck getting that fix embedded in there. Obviously, it's possible, but it's going to take forever versus you're using a platform like the Algo Design Lab. You can, within a minute, make a change, test something, see it doesn't work.
make another tweak and deploy to production and do so where your algo is quoting in 20 to 40 microseconds, that's really, really compelling. And that's possible because a lot of the technologies that we now leverage, which were incredibly expensive and probably out of our and most people's reach five years ago now are. And so I think what you'll see is there's almost two curves. There's the one that the Uber top 0.01% of participants are able to do in terms of microwave towers and all the types of things I mentioned earlier. And then there's kind of the commodity curve. And they will eventually converge. 10 years from now, TT will certainly have microwave towers.
is embedded on silicon for certain applications and for certain clients. But I think that's the way this is going to continue to play out. It's still something that we have to invest in and focus on and make sure we're 100% acutely aware of our performance and of our client's performance, because it is still very important. But that race to zero, I think, is pretty close to over for some. OK.
Can you just explain to us what it means to embed an algo on a network switch? Sounds sophisticated. What actually is that? Believe me, I know only enough to be dangerous here. In general, building hardware-based solutions for automated trading is ideal because software is, in relative terms, incredibly slow when it comes to hardware. The moment a message comes from an exchange on a network device, and that network device passes that message via an Ethernet cable.
to your server and that server receives that message and now gives it that now the server passes that message up to your trading application that that process passing up to the trading application and having a trading application react to that price update by saying oh you know what now i want to go submit a new order that's incredibly incredibly slow software is is relatively incredibly slow and so what a lot of firms do in terms of embedding algos on chips is never never reaching that last point in other words that if you've got a switch that's on your network a price update comes in the switch itself can can decode that price update can encode a new wire to go, whether you're quoting options or whether you're doing something like spreading two contracts against each other. That switch itself at hardware speed, so at line rate speed, meaning very, very fast, most switches, all their job is to do is to take a packet on the wire and send it somewhere else. These switches can take a packet on the wire.
actually gain some meaning from it, whether it's a price update or whether it's an order fill or something, and actually have some embedded code that says, ah, based on that price update, I'm going to change my price of this order to this price. And it can do that without ever, not only ever having to go to software, but we never even need to send that packet to a server to process. So the switch is able to, as fast as anything else in a network can do, it can actually respond and participate in anonymity trading. So that's where that comes into play. Again, it's where we've seen it being applied. It's only for very specific types of strategies that are even possible to embed on a chip like that.
Some people have trading strategies that simply you couldn't do that for. But for those that are very specialized and highly tuned, that type of performance is pretty tough to beat. That's incredible. It really is. Do these sorts of things have a computer screen where they can be monitored? That's a great question. That's a great observation. And the answer is generally not, or at least it's not the type of computer screen that you're used to.
It's a great observation because as you start to go further and further down this rabbit hole of incredibly precise, incredibly – tailor-made solutions for gaining as much, reducing as much latency as possible. All the other things we take for granted in terms of software become harder and harder. So having a monitor that you can see everything that's happening, where you can have a stop button, that all becomes much, much more difficult. Making changes, as I alluded to earlier, becomes much more difficult. So the people that are doing this, they've got systems in place that they can shut things down if they need to, and certainly they've got...
monitoring software that can look at something like the equivalent of a drop copy or that can keep track of orders that have been placed. But it's not a screen like you and I are both used to looking at because it is something sitting on a network device, essentially. That's incredible. Yeah, it really is. Man, wow, my mind's blown. Yeah. Okay, so are there any other trends which you anticipate may evolve in the near future? Like, are there any new features you're thinking about adding or like any up and coming?
things that we should be mindful of? You know, I mentioned at the start of this call, I do think that everything we just talked about, actually, in light of that, how can a trader differentiate him or herself in the face of such rapid technological innovation? And I think it really, it's about trading smarter. And so we've talked a lot about this internally, TT, over the last few years, you know, as much as we're investing in speed, we also need to and want to invest in giving tools that allow... the trading community to trade smarter and trade quicker and quicker to market. If you're not going to be the fastest one in a trade, if you could be the first one doing it or the safest one doing it, there's real value in that. And so I think because of that trend, I think with something like machine learning, where one of its primary benefits to our space is taking a problem that most traders are facing today, which is being inundated with data, both their own trade data, market data.
trying to glean insight from that so that they can trade smarter, so that they can analyze their trading activity, potentially glean insights from that data. We think that technologies like machine learning, as they become more accessible and more approachable, will start to empower even an individual trader to And that really, we think, is the new frontier in trading is enabling that trader to make more sense of their data and turn it really from something which is a burden today into something that could be a goldmine for new insights. And so I think we'll see that trend continue. It's exciting for us. I mean, we're now just getting our feet wet with machine learning by virtue of the forensic acquisition. But I think you'll start to see a lot of data-related technologies that are more about scale and less about performance continue to emerge and evolve. Right. OK. Awesome, man. Well, Rick, let me ask you one last question.
One of the reasons why I wanted to bring you onto the podcast is just to speak with you because you're not a trader, right? But you're very involved in the industry. Obviously, you're the CEO of Trading Technologies. And obviously, not everyone is cut out for trading, okay? that doesn't mean they can't be involved in the industry in some capacity. For anyone who likes the path that you've taken or would like to get involved in a technology company that works in the trading space, is there any advice? I know it's a very broad question, but is there anything you might suggest for someone to give them a leg up and help them get into this kind of space? We have conversations like this. We have a big intern program, internship program at TT, and obviously that's somewhat of a predetermined selected pool of people that have an interest in our space. But they ask those types of questions all the time. And this is an industry that...
is changing so much from a technology perspective that even the definition of trader has fundamentally changed. Not I think, I know. And I think when we look at our user base, more and more the trader, even what you would probably classify as the traditional trader, in more times than not, they're coming into this industry with a highly technical background, computer science, engineering, mathematics. And I think that's... really just a sign of the times, but also indicative of the types of skill sets and talents that someone needs to really just to make sense of this world. You know, you can't, unless you're a very select few, even just being a quote-unquote, you know, old school trader still requires you to be able to manage and analyze and process an incredible amount of data and do so quickly. And I think that's more and more becoming, you know, the skill set of the typical type of, you know, new college graduates coming into this space. But if someone's a technologist and wants to have a leg up,
I think in a lot of cases, at least from TT's perspective, having an interest and a track record of wanting to and doing and solving interesting technology problems will certainly give you a leg up. I would say that roughly 50% of the people we interview for technical roles at TT have some experience in trading. I think it's an industry that if you don't have some passion for the industry itself and not just the technology, it's going to be hard to... be willing to face the incredible challenges that you face every day. And so I think the folks who come in to interview for a technical role who have a have a clear passion for wanting to solve some of these really interesting problems. And sometimes it's even market structure problems and not necessarily pure technology problems. I think inherently they have a leg up because we know how grueling and brutal this industry and unforgiving this industry can be. And so we, I think, think that people that come to the table with a clear predisposition to be interested in the space are probably the ones who are going to thrive the most. So that's what I'd say. It's not for the faint of heart, as I'm sure most of your listeners know. And so we want people to come in who think the space in general is an exciting one for them.
Very good. That's awesome, Rick. I think that's really helpful. So if someone wants to find out more about Trading Technologies, obviously they can go to tradingtechnologies.com. And what about you, Rick? I know you're on Twitter. What's your Twitter handle? It's the worst for things like this. So the company, our TT account is at trading underscore tech, T-E-C-H. Mine is at Rick underscore Lane, except it's R1CK, not R-I-C-K. It's R1CK underscore L4N3. Don't ask, but that's how I spell Rick underscore Lane for Twitter. But you can...
The company is always tweeting my handle out as well. So at trading underscore tech is a good one to follow as well. Great. And as listeners know, I'll be sure to include all those links in the show notes. Rick, I really appreciate you doing this. Thank you very much. Thank you, Ryan. This was great. You've reached the end of this episode of Chat with Traders. But rest assured, there are more episodes loaded with real market insight and zero hype on the way soon. So to stay updated with each great new release, subscribe to the podcast on iTunes. And we'd love it if you leave a rating and review. We'll catch you next time on Chat with Traders.