FX: Centralized Liquidity and API Fragmentation

Ali Curi: Markets conversation is a new ION podcast where we discuss topics of importance to capital markets participants with product owners, subject matter experts, and industry leaders,

Eugene Markman: Nothing is static. Everything’s dynamic. Jobs are dynamic. Roles are dynamic, and I think you have to be yourself dynamic as well. You have to constantly learn new skill sets to, to stay within the market to stay relevant.

Ali Curi: Hi everyone. And welcome to Markets Conversation. I’m Ali Curi. On today’s episode, we’ll discuss a couple of modern challenges to foreign exchange traders, exchanges and their software, specifically centralized liquidity and fragmented APIs. We’ll explore why these challenges are happening and how firms can best manage them.

Our guest today is Eugene Markman from the Foreign Exchange Markets division at ION. We’ll discuss issues such as low latency that results from centralized liquidity pools, and why API fragmentation has proliferated in the market. Let’s get started.

Eugene Markman. Welcome to the podcast. Thank you for joining me.

Eugene Markman: Thanks Ali. Really excited to be here.

Ali Curi: Well, I’m really glad you’re here as well. Before we get to the conversation, let’s learn a little bit more about you. So have you been in FinTech, your whole career?

Eugene Markman: Actually no, out of college, I went to work for a bank.

So my, my career started off in banking. Um, and then, you know, focusing in on, on trading.

And one thing that I quickly learned was that technology was starting to drive how trading was happening at credit Swiss. Overall, it was becoming more and more difficult to make money trading’ because of market conditions, because of regulation, and what I saw as a good way to transition and take what I learned by working on a desk is to go to FinTech and apply it there and work in the same market, just from a different direction so that’s how I transitioned to FinTech about 10 years ago. I went from Credit Swiss, working in interest rate products to Market Factory, which was a startup at the time, and we were specifically focused on building feed handlers for FX.

Ali Curi: All right. So let’s get started. Let’s start with big shifts in the FinTech world. Let’s talk about centralized liquidity and for a little bit of context, a couple of years back, two of the primary markets in FX, as you know, were consolidated, one of them EBS getting its technology converted to the Chicago Mercantile Exchange. So what kind of challenges does that centralized liquidity create?

Eugene Markman: I think there’s two ways to look at this. Number one, we have consolidation right and centralization at a company level. So companies are acquiring other companies as we saw with EBS and the, and the CME, uh, London Stock Exchange and Reuters, but in reality, they’re still running multiple platforms. It doesn’t necessarily mean that liquidity is being centralized from a liquidity pool point of view, it’s just the ownership of that liquidity pool is being centralized. So, the impact to users and traders is slightly different as if it was truly, truly centralized.

The centralization. It really means that there’s a lot of changes that’s happening. Right. So, um, in every acquisition, everybody’s looking for efficiency or they believe that their model is the better way to run such business. Right? So what we’re seeing here is looking at EBS as our first use case is that EBS was migrated off their old platform and put on CME technology. Changing their model completely. So previously EBS would run three matching engines. One was in New York, one was in London, one was in Tokyo. They would match clients locally, but then also have the ability to match within each region as well. So if there was, if the match didn’t occur in the local center, it could communicate within the regions to see if there was some match on another exchange, but that has now changed.

The new trading model is that you have G -10 traded out of New York. For EBS, uh, and London will take care, will be emerging markets and NDF liquidity. If you’re a user, you have to start thinking about, okay, if I’m not based in New York, but all my G 10 trading on primary markets are New York based, how do I connect?

Right. And how does that impact me? Right. Of course. There’s. You know, there’s the API conversation of it, you know, actually coding to the new exchanges and all the work that goes associated with that. Connecting, testing, UAT, running in parallel, full migration. So that’s an impact, but really from a business point of view, right.

And if you are a London bank that wants to trade G 10 on EBS. Now you have a big latency to New York latency that you previously didn’t have to deal with before. You now have to account for it. Is the business model, you know, stayed the same for you, or do you need to adjust? Are you making enough money, uh, trading G 10 and providing G 10 to your clients where it’s worth, uh, Both paying EBS for the fees, um, as well as having the latency to trade in New York. Right? Or do you look locally, right? Do you change your liquidity providers? Right? Be because the liquidity landscape really has changed similar in New York. If you are an emerging markets or NDF trader, you have the same, same problem, right? You’re no longer trading locally, but you’re now trading out of a London exchange and you have to put the same thought into, either business model.

Ali Curi: So, can we go back a little bit and explain these latency disruptions and how do you manage them without having to go somewhere else?

Eugene Markman: To illustrate the point, you know, let’s say you are sitting in New York and you’re trading on exchange that is in New York. Uh, you received market data from that exchange, giving you a signal and you submit an order, right? Based on that market data, that order, the closer you are to an exchange will arrive at the exchange faster. And if the price is still available, based on the market data that you read, you will be matched, right. And you will execute it in a trade.

If it takes you longer to arrive at the exchange, the data that you have received is no longer relevant, right? It expired. Market data has changed. Those orders were filled. So your order might not be filled anymore, right. And therefore you’re not getting high fill ratios. That is the challenge, right. You don’t necessarily need to be the fastest, but there definitely is a time of reasonableness, uh, which you need to consume market data and electronic trading. Um, and have your orders arrive at, at the exchange.

Ali Curi: And is this affected, is this a software issue or is this a location issue?

Eugene Markman: It’s a location issue. Well, the true answer is all of the above, right? But we could take some things as given if everyone has good, strong technology that doesn’t introduce latency in code, then what we’re left with is latency that is based on distance, purely how close you are to the exchange and how short your line is your actual physical cable that connects you to that exchange through your ISP and then through or dedicated circuit, the faster than data will arrive. Now, if you’re routing something from New York to London, it’s a significantly longer line than it would be from somewhere in New York to the exchange in New York.

Ali Curi: So let’s talk about APIs for a minute. Let’s change gears, discuss the evolution of APIs in the market. As we know the APIs, they support or add functionality to software and platforms, but often too many of them and we start to see what’s called fragmentation.

So from your point of view, let’s discuss fragmentation a little bit and kind of give us an explanation around that. And then what are thoughts around this exponential API growth, like what is too much? And how does that really affect the landscape?

Eugene Markman: FX is not a security. FX has less regulation than, the equities market. Um, and that has led to environment where every company is able to create their own API and that API is slightly different. Now, why do we have so many exchanges in a, in FX and so many LPs, right, with not standard APIs? Well, that’s just the opportunity. That’s the size of the market as an asset class, FX is the largest global market in the world. Right. So we do have a lot of participants as FX was electronifying, it was at an interesting time. It learned a lot of what was previously done in equities because equities came first and a lot of people saw that opportunity to monetize. And therefore we saw over the last 15 years, a tremendous number of exchanges or ECNs, uh, electronic currency networks pop up.

Of course, every single bank is also an LP and a participant. And then after that, beyond banks, we saw non-bank market makers come into the market and also, uh, act as LPs, right. LPs that were actually able to take more risk. Internalize more flow, correlate it to, to other asset classes and have their own different view for the market away from everybody else.

Because of that, because of the competition, there was never really a need or a driver for standardization. Everybody wanted to be slightly different and it was in their best interest to be slightly different, right. Once you’re connected to somebody, right? And you’re their LP, you don’t necessarily want it easier for them to turn you off and turn somebody else on to, to migrate away, to find new ECNs, to connect to new LPs.

But also within FX, there’s also multiple products that we trade and that’s really where the growth became exponential. Not only do we have so many participants, but we also have the product that we’re trading and, and the APIs because they’re different post liquidity can look different. Banks will have, for example, banks’ LPs could have a specific API purely for spot FX, but they could have different API if you would like to trade other products like NDFs, swaps, well, they’ll have, instead of a streaming workflow, they’ll have an RFQ or RFS workflow, which is request for stream request for quote, same thing with the, the ECNs the pools of liquidity. Each had its own pool specific for that product. And for that product, it had a specific API. And there we really saw the growth go to a complete new level.

Ali Curi: So it’s not just that they’re added functionality, they’re adding functionality to the firm’s software. It’s also improving. It’s also, I would say it’s a differentiator, right? Between, you know, certain firms who would have a different API mine is faster, mine performs a certain way, but what can market participants do to manage this fragmention?

Eugene Markman: Well, it’s difficult. I mean, it’s a very, very difficult problem. Um, they have a couple of options, right? One is you decide what part of the market you’re going to participate in. And once you’re connected to those liquidity providers and those exchanges, you don’t have a need to connect to anybody else.

Now there’s a downside to that, right? You might be missing out right by not connecting to somebody else. You know, you don’t know what you don’t know. So it’s difficult to, to say this is what you’re missing and, and really the best way to try new liquidity is to connect and to give it a try. But that’s difficult. At this point, doing connectivity work is not difficult, right? It has become a commodity. I think the best thing clients could do and traders could do is use third parties as aggregators or purely as professional services to connect them to different markets. Um, and that gives them the ability to try different markets whenever they want.

Ali Curi: Excellent. Eugene, what is the next area of growth in FX trading technology? Where will innovation come from? And what do you think that will look like?

Eugene Markman: Very interesting question. Right? So what’s next, right? Personally, I think the next step in the evolution is so we’ve electronified now we trade through algorithms. We, no longer do things manually. We don’t pick up the phone.

We don’t have somebody sitting at a keyboard typing things in, right. But we still make the decisions. The next step is for the computer to make all the decisions for us. That’s really going to come from AI learning. And beyond just making decisions, but also learning and adapting as to what decisions to make as you go.

So for a machine to study the market, study the trades, study clients, study news, be able to take as much input as possible, right. And learn from it. Right. And make those decisions faster without having somebody to come in intervene. Cause currently what we have is. We have somebody able to write an algorithm, right, and that algorithm performs. Right. And after a while that algorithm can be updated, but for the algorithm to be updated, a developer or quant will have to go and start thinking about all the other use cases. Right? All of the other scenarios that aren’t accounted for and update that algorithm to function as they want it to.

What would be really interesting? What I would like to see is for algorithms to update themselves because they could learn significantly faster than we can, right. As humans, and they could consume significantly more data than we can. Right? And it will take out the human doing the analysis, but the machine will do analysis as it goes.

It should be able to update itself within minutes or within hours, you know, within the same day. Right. As, as opposed to, if we have to do it, you know, this will be a long process, long iterative process with a significant amount of testing. That’s where I think the next era of innovation will come from, that will be definitely driven by tech firms.

That’s where I see money being invested. Right. So you’ll have your PE and VC money. Uh, banks are a little conservative, right. And I’m sure they would like to be users and I’m sure they would like to participate, but because they’re risk averse, they will, never be the first movers on this. So, um, I, I think we have to watch and to see what fintechs pop up as, as startups and really make traction in this space or become disruptors because that’s, they’re gonna be the ones who are driving that change.

Right. And they’ll have that focus as opposed to somebody in the bank who’s mandate is significantly bigger. Right and their primary business is their biggest concern while innovation though interesting is not their primary objective. Right. It’s sustaining the current business. And the last question you asked me is, what will lead to this, right? What are the prerequisites for this innovation? You know, and, and that’s an interesting question. I, I think we’re nearly there, it’s the ability to have the computing power, which we do. It will have to become cheaper for tech firms to be able to buy. I, I think we have that computing power available to the likes of NASA or somebody at that scale, but it needs to be more accessible, you know, to, to smaller companies, right. That, that could purchase it and, and use it. Uh, and the second part is, is, is faster data because this is all gonna be data driven, right? So you need the input to have the output. So the ability to consume all that data, to have that data more normalized, uh, more standardized will be crucial in taking that next step.

Ali Curi: So I’m hearing, uh, AI, I’m hearing self-correcting code, I’m hearing minimal intervention from people all in exciting technology for the future of FX trading.

Eugene Markman: Yeah no, very excited for it. I can’t wait to see what comes next.

Advert: This episode is brought to you by ION. Whatever the size of your business, our FX solutions automate and simplify your trading, risk management, and operations in one easily scalable platform. To learn more, visit us at iongroup.com/markets, or email us at [email protected].

Ali Curi: Eugene Markman. Thank you for joining us today. It’s been such a pleasure having you on the podcast. I hope you visit us again.

Eugene Markman: Thank you so much, Ali. It’s been a pleasure

Ali Curi: And that’s our episode for today. You can follow ION Markets on Twitter and LinkedIn. Thank you for joining us until next time.