Episode Transcript
Jamie: Hello everyone, and welcome to another episode of Hedge Fund Huddle, where we go behind the scenes of trading to uncover the latest trends and talking points in the world of investment. As usual, I am your host, Jamie MacDonald. Today we are going to be looking at Foreign Exchange, or Forex to the shorthand lovers out there, but in particular we are going to be looking at best execution within Forex. As you know, different asset classes trade differently and currencies are no exception. I'm sure I'm telling you things that you already know, but equities typically trade on exchanges and they need to conform to certain rules and regs to stay on that exchange. Bonds are mostly OTC, but currencies are different still. They trade across a global network of banks and financial centres across the world. And because of that decentralisation, it can be a rather fragmented market, which is an issue that's now starting to get solved by companies such as Tradefeedr. And we have Tim from Tradefeedr to talk us through this today. And so I think without further ado, what I'll do is introduce today's guests and I'll go through them one by one, their luckily from all corners of the market. So representing the sell side, we have Tan Phull who's Head of FX Execution for Bank of America in London. From the buy side, we have Alan Martin Lucero, who's an FX trader at Norges Bank Investment Management and from Tradefeedr we have Tim Cartledge who's the Chief Data Officer there. For those of you who don't know Tradefeedr yet, it's a relatively new platform launched in 2021 but already won Best Execution Product of the year in 2023 at the Risk Markets Tech Awards. So gentlemen, welcome to the pod.
All: Thank you, thank you, thank you very much.
Jamie: So I do want to start with a little bit of context and background. Maybe Alan, you can start here to talk a little bit about how the execution of FX has worked historically up until today. Talk about the innovation of algorithms, how that journey has happened. RFQs do they still exist, how they exist? And if you can just give a little bit of a background about yourself as well.
Alan: So initially all the trading was done on the phones. We actually still have turrets. Sometimes we still pick up the phone and we have to talk to our brokers. But that was the traditional way of trading FX. Later on, and probably Tim has quite a bit to share about this, the electronic networks came in and suddenly we had access to a lot of liquidity providers that before they were not available to you. You can see the prices and you can actually trade those prices. Bilateral connectivity, like connecting direct to the brokers. And if you think about what's happening right now, I would say most of the buyside are actively using algos. There's pros and cons on those. RFQ is still a thing. So request for price to trade. And you know with the opening of algos then you start to think, okay, much of our problem has been solved in terms of access and liquidity. We still need to figure out exactly how to optimise that execution. And that's how we end up with very fancy models. And when I think about my current role, I'm not sure if I'm going to be doing the same thing in three years time, five years time, things are changing very, very fast with the new technologies.
Jamie: So Tan, maybe I can turn it over to you. And I again fall back on my equities background and reserve the right to ask fairly stupid questions. But Forex is the largest market in the world. It seems like it's the most liquid market in the world. Why has liquidity ever been an issue?
Tan: I also, by background spent many years, most of my career, in equities. And so I've seen the electronification of that marketplace. And we're seeing parallels to FX. So to kind of touch upon some of the things that Alan mentioned, the provision of capital, which is effectively how an RFQ works, what does that mean? That means that risk price, which is immediacy of an execution, i.e. where can I buy and sell right this moment now, make me a price that still is very prevalent in FX and will continue to be so for many years. It will not go away. And in fact most of FX as we know it today does transact that way. There is an increasing proportion of the market that is starting to use algos. What algos do not do for you is provide you an instant price for your entire size. So there is a delineation to be made between the two, which is there are going to be parts of transactions that need immediacy for which an algo will not be sufficient. And there are other parts which can be more passive in nature or liquidity seeking, or over the day, and so on and so forth. That proliferation of the algo trading is becoming bigger in FX, although it's still less than half of the marketplace at least, in aggregate. Now, how does that compare to something like equities? Well, in equities a much bigger proportion of the marketplaces is traded by algos. In FX, algos are a commission based product, whereas the provision of capital there is an all in price, which means there is no commission but the cost of the capital is baked into the net price at which a customer would trade. So this is certainly a conversation that's becoming more and more technical with more data. But one is certainly not better than the other. And both will co-exist for some years to come.
Jamie: So Tam you actually remind me of days of trading equities when I would pick up the phone and ask for a price in a million Barclays in those days and you get quoted a price a couple of bips off the bidder ask. The onus was then on the bank to go and fill that order and if it incurred a loss ratio then that was unfortunate for the bank. So flipping that around with these algos, how do you know that if you do want to trade in size, that you're not moving the market or affecting the market? Or put another way, how do you see what influence your trade is having on the market? Is it possible to even know that?
Tan: It's a very good question. And in the case of FX, it's quite nuanced. So if we talk about equities, where there is a consolidated account of volume, when you know what you're buying a million of Barclays, you'll know roughly how much is trading through the market at that moment in time. Whereas in FX there's an understanding or there's a gauge for it. But because the marketplace is so fragmented and so bilateral, the idea that you would know within a short period of time how much liquidity is transacting is very, very difficult. And thereby the overall impact of your own trading is much, much harder to discern. And that is one of the reasons why it is just easier to get a price from a liquidity provider, because it's very hard to know how much impact am I going to have. Whereas the liquidity provider may have a better idea, slightly better, and be able to warehouse or offset that risk in a more thoughtful or longer-term horizon. So what is complicated about trading FX, therefore, is the absence of that data that shows real time how much is trading. There are proxies for that information. And perhaps, Tim, it might be a good segue into some of the things that you're working on.
Jamie: Well, actually, I just want to go back to Alan for one more question and then come back to Tim and we'll talk through the problem that Tradefeedr is solving. But so for Alan, could you just walk us through your kind of investment process? You're going to put on a trade of a certain size, you're deciding whether an RFQ or an algorithm is the best approach. Presumably your style or your strategy works better with some algorithms versus others. Is that part of the process, you need to spend the time to work out which algorithms work best for you and your trading style? And then a third part to that question is what data is helpful for you, on the buy side as a client, to make sure that you do get best execution?
Alan: To mention again what Tan was talking about, the choice between algos and RFQ. If you look at the trading desk now, you have access to bilateral feeds, RFQs and algos, and you actually end up using all of them, even the phone as we spoke before. So the main driver of the decision tends to be volatility. So if I'm in a very uncertain period where I don't really know what's happening in the market, I have no certainty at all of what's the volumes being traded out there, I might be willing to do RFQ and what it means RFQ to me is actually I'm paying someone to take the risk for me. And that's probably the concept where you have to kind of buy sides, right? Those that have the capacity to warehouse the risk, those probably will choose algo execution and those that don't have that risk budget and then they will prefer RFQ. So the choice is always looking at volatility in the market and what immediacy I need. If I'm very uncertain probably I will prefer to pay up and give the risk to a broker dealer like Tan here.
Jamie: So Tim, let's turn the conversation to you. And maybe as a point to start with, could you give us a little bit about your background and how you came to be at Tradefeedr?
Tim: So I’ve been in foreign exchange for 25 years. So I've seen the whole evolution from it being a voice trading business, in the 90s, it was voice brokers and voice traders all dealing on the phones, to electronic trading in the 2000s. So I used to be an options trader in the 90s, a currency options trader. Then from 2000 onwards, it was when our market started becoming electronified. I initially started as a systematic prop trader as we got access to electronic venues, but then realised that there was a lot more money to be made in market making than there was in prop at the time, and built out some of the first fully all electronic FX market making platforms. And at Barclays, I think we can say it was the first that was fully automated and based on statistics and maths and engineering rather than kind of the feeling in your gut on which way the market was going. From early 2000s onwards through to 2015, I was running electronic trading businesses. 2015 onwards, I ran EBS, which is one of the main interbank electronic trading platforms. So I've seen the market from the sell side from Barclays and Dresdner amongst others, and from that platform side of EBS. We sold EBS to the CME, that's well known. And I retired. Then heard about the Tradefeedr guys and what they were doing, and it seemed very interesting in a kind of next project to be involved in, because you get to see all the rest of the data that I haven't seen so far, and you get to build a new layer of products that you couldn't build in the past because you just didn't have access to enough bits. And it's become a kind of fascinating data science project. So I invested in Tradefeedr as well as working there.
Jamie: So if you could explain what precisely is the problem that Tradefeedr is trying to solve? I mean, when I said the introduction, I alluded to the fact that Forex trading is decentralised, it's fragmented. There's no sort of centralised marketplace for these currencies to change hands. So what is the exact problem it's trying to solve and where is it along the evolution of that?
Tan: Jamie, if I may interject, I guess it would be helpful perhaps to just explain, because you talked about best execution, if I provide a framework on what that actually is for anybody who's listening. So execution is effectively what we're terming as trading, so the buying and selling of an instrument, in this case FX. And on each desk there's going to be a number of traders. And they may all have slightly different ways on which they transact their business, or they might have slightly esoteric things that they're trading between them, but overall they represent a trading desk, let's say. Well, best execution is a, in certain jurisdictions and certain asset classes, a regulatory framework, but in the case of FX, let's call it a best practice. And what it requires is for the trading desk to be able to codify their process of trading and then be able to evidence why and how they're doing what they're doing. Because I think different regulatory reforms have come along MiFID for example again in equities this has been mandated. And it requires a lot of rigour and discipline. FX isn't necessarily subject to the same framework. But a lot of trading desks have adopted this and said okay we are going to have a process and we will have a mechanism by which we evidence that we are doing to that process. So that means, therefore, that they require data in order to be able to do that. It is now no longer about I trade with my friends or somebody took me for a nice dinner last night, so I'm going to give them a trade. Now it has to be evidenced. And therein lies the data. So perhaps that was a kind of an introduction into the problem that maybe Tim is providing the data to in order to help with. No, no.
Jamie: No, no, I'm pleased you jumped in because I understand that the evolution of trading, there's got to be a sort of justification to the price of which you give each trade. And it's previously been very difficult to do that because it's been hard to find the data. But that's one of the things that Tradefeedr is trying to solve is, get more data out there. So, Tim, is it as simple as greater transparency is always a good thing? I can argue.
Tim: I can argue this from both sides. Solving the problem of fragmentation of data in foreign exchange is a very good thing. And let's talk through that one. And then I'll throw something else in at the end. For your audience that's not so familiar with FX structure, I'll just highlight what the FX trading structure looks like. You've got three central, three primary exchanges: CME, EBS and Reuters Matching. You know, so that's where you know BofA will meet Deutsche. They’re the venues, the interbank venues where liquidity providers, the buy side, actually meet each other. And around that you'll have maybe 15 venues where clients meet liquidity providers in an aggregated environment. Then outside of that, you've got the liquidity providers themselves. There's about 20 major liquidity providers, and you can deal with them directly if you want to as well, rather than going through a platform. So about, I would say about 90% of the volume traded in foreign exchange is not in the lit venue, so it's not in one of the primary markets. 90% of it is in a bilateral environment, in a hidden environment. You can't see where everyone's trading and you can't see what volume people have transacted and at what rate, you can just see your own data. So the data is fragmented across individual clients and the 20 main LPs and the 15 or so client ECM platforms that clients can deal with LPs through is fragmented across the whole thing. What our mission is, is to take all of that private data from wherever it's held, either from the client or from the liquidity provider, or from the venue that's in the middle, bring it to one point for the client where it's normalised, cleaned, available for them to access, either through reports that we produce or through our own data science environment. And it allows, for the first time, foreign exchange to empower clients to just be able to get at their own bloody data from wherever it is in the world.
Jamie: And just to be clear, when we talk about data, are we talking about who's trading, what, where and what size, that sort of thing?
Tim: We talk about the client's own data, but wherever i it was traded. So you say is more transparency a good thing? Well, now we get to the point where I say the no part, which is clients don't want their trades to be visible to other clients.
Jamie: How do you get clients comfortable with that?
Time: We have dedicated data environments per client. You've got ?? data areas between a client and an LP. And you've got client segregated areas where it's just the client seeing their data or the LP just sees their data. So it's all ring fenced. Because if you've got a buy side client that's executing a very large amount of a relatively illiquid currency, they do not want anyone else to know about that. That would be to their detriment if that information leaked. So it's not about just getting all the information in the world and then giving it to everyone. It's actually curating and corralling that information so that the right people see it. That's our mission. And then we build products on top of it and best execution and all the rest of it is part of that.
Jamie: And Alan, perhaps as a trade theatre user, you could talk a little bit about your experience with it and how it's how it's helped.
Alan: These sort of data has been around for quite a bit of time, but it was very painful to work with. So before we were just taking the fixed messages, making sure that every fixed tag will be mapped to the right feature in the data set. Then we had to iterate and enrich that data set in order to get benchmark for the execution. And then you had to have the conversation with your liquidity provider. And then they had different numbers, and then they had different benchmarks. And it was very difficult to collaborate, try to land a better product that will suit your needs, a better algo, a better execution overall. And the advent of these standardised data sets with APIs you can just pull the data into your computer and start to play around trying to understand your execution and your costs and your liquidity providers a bit better have been quite a game changer. And instead of spending 90% of the time cleaning data, we are using all that time trying to optimise our execution and just get better at what we do.
Jamie: And Tan moving on to the topic of scale, obviously you have very different clients keen to operate at different levels of scale. How do you handle that, that spectrum?
Tan: I guess we've got, by virtue of the firm that we are, we face sovereign wealth funds like Alan, corporates, real money hedge funds. The scale that we're able to access is actually the advantage that we have. Because of how Tim's described the liquidity overall, being as fragmented as it is, being able to tap into that scale is what gives us a competitive advantage, because we've got such a broad breadth of customers, and that allows us to internalise or match off various different interests. So it's a very heavy technology investment predicated with just the breadth of the franchise that we have. And we're incredibly lucky to have access to that.
Jamie: And I guess, in this day and age, AI seems to be a topic that can be sort of thrown into almost any conversation. But I'm keen to get opinions from you guys. How is AI influencing your each individual business perhaps? Tim, you want to start first.
Tim: It depends what we mean by AI. It’s taken over from blockchain as the word that everyone has to say. Yes, we do AI, we do machine learning.
Jamie: It’s to help with data crunching really?
Tim: It’s just the statistical technique, statistical modelling techniques that we use, I don't want to say basic, but basic machine learning models compared to, you know, GPT4 or whatever. The problem with the new large language stuff is it lies, it gets things wrong. And we don't want to be wrong. It tends to get signs wrong if you ask it, you know maths problems. So yes we do AI, the question is where does the new level of AI come into things? I think it will be in the sort of interface world where you can ask general open ended questions. What's different about my trading this week to last week? And you can get something that sits on top of the hardcore stats and then just gives a presentation layer of, well, this is the sort of thing that you should be interested in, but you can't just throw the data at AI and then hope it's going to get it right. It's not at that point yet.
Jamie: Not trustworthy enough yet. Alan, similar question to yourself.
Alan: Finance is a highly regulated environment. So the idea of just letting a machine do a trade, make an investment decision and then it goes wrong and then my chief calls me into the room and say, I don't know, the machine did it for me, sounds like probably isn't going to be the right approach. So the way we look at this is either we will generate very good machines that will give us insights on execution and investment decisions. And then, you know, it's presented to the human being. And we actually do the trade ourselves or we end up in these what we call explainable AI, which is a fancy way to call mathematics and smart computer science applications into what we do. But I'm a bit sceptical towards just letting a machine run a business that is highly regulated and you need to explain why you did what.
Tim: Yeah, there's quite a lot of regulatory situations where the traders intent is important. Did you intend to move the market or did the market just happen to move? That's a big distinction. And saying, well, a machine did it, I don't know what it was trying to do, it just makes money is probably not going to be a good answer as far.
Alan: As far as it makes money, probably everyone would be pretty happy about that. So no one will ask you questions.
Jamie: It sounds like the relationship between a portfolio manager and an analyst. If the trade went well, the PM will take the credit and if it went badly, will blame the analyst. I remember those days very well. So Tan just briefly on the topic of AI. On the sell side of execution trading, how does it factor in?
Tan: As Tim mentioned, AI has very different meanings. If we're talking about large language models, the GPTs of the world, there are proofs of concepts happening within our firm. There is an investment. We are taking it very seriously, but it's some way off being available externally for the reasons that have been outlined. But I'm excited about the space. I think it will be here to stay and it will make us more efficient. If it's simple things that summarise such and such analyst’s research or interest rate calls over the last number of quarters or whatever, there's going to be lots of different ways at which you'll make us much more efficient excited about it. But it has to be done responsibly and I think that's something that we're working on.
Jamie: And on the topic of cryptocurrencies itself. I'll judge by the expression on your face whether you enjoy questions like this or not, but are they, in your opinion, set to become more of an asset class? Are they a big part of Forex currency trading right now? How do you see that story playing out?
Tan: I think the headlines that we've had over the recent years have made the space less reliable, shall we say. We do not trade cryptocurrencies, in certain cases we trade the futures on the CME for certain clients. Do I think that Bitcoin will be around in the future? This is just my view, yes it will and I think it will become more regulated. And hopefully it will trade more FX like. I think we're some way off that yet just by virtue of the way it custodies and so on. So yeah I think it will be here to stay, there will be some trading of it in the future, but I don't see that happening anytime in the near future, if I'm honest.
Jamie: Tim just a personal view or otherwise?.
Tim: I’ll give you a personal view. We don’t see any crypto at Tradefeedr, it's not like clients are clamouring for it at all. I mean, it's not on our radar in the data sets that we see. The market is client led. So if there's a lot of buy side that want to trade crypto and it was speculated, eight years ago or so, it would be treated as an asset class and there would be major asset allocations to it. But that hasn't come to pass. It really isn't treated as an asset class by the professional community in the way that it was speculated to be. And the big issue is the lack of regulation. So it's like every single thing that's been banned in regulated markets seems like it's the only way to trade in crypto. So there's no-one that's going to work in a regulated industry that's going to touch it with a ten foot pole and use the techniques that people trade in. So if you if you're not allowed to spoof or paint the tape or trade on clients positions or whatever, you can't trade crypto. So I think it's going to have to be properly regulated, properly cleaned up, and then the professional community can have a go at it. Personal view.
Jamie: Yeah I see it goes both ways. I it feels like there's still a few more blow ups to come. But on the flip side, you know, every time one of these ETFs gets more formally validated, it seems to give it another leg up. So yeah, I can kind of see it go both ways. Gentlemen, this has been a very interesting discussion, and thank you. I just want to talk about current state of play, current markets. We're living in a world where there's more people going to vote in democratic elections than ever before in the history of the world. Do we see more volatility coming in currencies? Is it going to be a more interesting space to play? With companies like Tradefeedr is it going to become more centralised, more transparency, is going to be more attractive? I'm really trying to give a plug to this world for people listening who may have been scared off by Forex are now interested to coming into it. I'm just interested on your views on where we are today and looking out over the next five years or so. Tan, we can start with you and move to the others.
Tan: Macro backdrop is probably as exciting as it has been as long as I can remember. Given the elections, given the rate at which central banks will start to diverge in their policy, this creates a really interesting environment and probably a lot of volatility within foreign exchange. Against that backdrop, you've got companies like Tradefeedr who are providing a lot more data, a lot more transparency and more tools for which firms to be able to analyse and optimise their trading decisions. So I think overall, the look ahead for the rest of certainly 2024 and beyond is pretty exciting, as exciting as it has been in some years in effects, actually, because all central banks came into Covid, the inflation policy was pretty much consistent. And now the exit from that is looking interesting I would say.
Jamie: Alan?
Alan: FX in general is a low volatility asset. So we welcome always unexpected election results and central banks changing monetary policy any geopolitical risk and so on. So it gives us the chance to actually make money in this market. In terms of how these data help us to actually take advantage of them from our perspective, what we try to do is, well, we try to route an order, given an investment strategy to who we know, that will perform the best given our execution style and without having all our history and how we've been trading over these highly volatile events, it gets very, very difficult. And then it becomes a bit of how it was during the old days where you were just routeing to who took you for lunch last week. So it really gives you an indication of what has to be your broker panel, what algos you need to have there, what are the best strategies out there for you to implement your execution.
Jamie: So, Tim, over to you.
Tim: The foreign exchange is driven by macroeconomics, geopolitics and interest rate differentials. So we had a period of about a decade where it seemed like everyone's policies were aligned. The world looked like a reasonably safe and ordered place and interest rates were zero. And volatility came down substantially in foreign exchange. Now it looks like we live in a dangerous world. Policy diverges and interest rates have come back and the carry trade has come back. Last time we had events like this you start seeing some major dislocations in currencies. Dollar Yen has gone from 90 to 150 odd, you can see that off 30% if they change intra trade policy in Japan. So yeah it should be fun markets.
Jamie: Gentlemen, I will say thank you so much for your time. It's been great talking to you. I'm sure people can try and track you down on LinkedIn somehow, but for now I want to say Tan, Tim and Alan, thank you very much for your time.
Alan: It’s been a pleasure thank you.
Jamie: The information contained in this podcast does not constitute a recommendation from any LSEG entity to the listener. The views expressed in this podcast are not necessarily those of LSEG, and LSEG is not providing any investment, financial, economic, legal, accounting or tax advice or recommendations in this podcast. Neither LSEG nor any of its affiliates makes any representation or warranty as to the accuracy or completeness of the statements or any information contained in this podcast, and any and all liability therefore, whether direct or indirect, is expressly disclaimed. For further information, visit the show notes of this podcast or LSEG.com.
