The inner workings of a High-Frequency Strategy ft. Anant Jatia

[00:00:07] Anant

We really think about this world as physical commodity traders first and how they might be reacting to these markets, and then look to systematically trade around it. That’s held us in really good stead over the years as we think about taking this program further.

[00:00:29] Intro

Imagine spending an hour with the world’s greatest traders. Imagine learning from their experiences, their successes and their failures. Imagine no more. Welcome to Top Traders Unplugged, the place where you can learn from the best hedge fund managers in the world so you can take your manager due diligence or investment career to the next level.

Before we begin today’s conversation, remember to keep two things in mind. All the discussion we will have about investment performance is about the past, and past performance does not guarantee or even infer anything about future performance. Also understand that there’s a significant risk of financial loss with all investment strategies, and you need to request and understand the specific risks from the investment manager about their products before you make investment decisions.

Here’s your host, veteran hedge fund manager, Niels Kaastrup-Larsen.

[00:01:24] Niels

Welcome to another episode in the Open Interest series on Top Traders Unplugged hosted by Moritz Seibert. In life, as well as in trading, maintaining a spirit of curiosity and open-mindedness is key, and this is precisely what the Open Interest series is all about. Join Moritz as he engages in candid conversations with seasoned professionals from around the globe to uncover their insights, successes, and failures, offering you a unique perspective on the investment landscape. So with no further ado, please enjoy the conversation.

[00:02:00] Moritz

Hello, this is Moritz Siebert, and I’d like to welcome you to another episode of Open Interest on Top Traders Unplugged. Today, I’ll be speaking with Anant Jatia who is the Founder and CEO of Greenland Investment Management, a hedge fund focused on commodity arbitrage strategies. I met Anant about five years ago at a conference in Miami and we’ve stayed in touch ever since.

About two years or so ago, I asked him, for the first time, if he’d like to speak about his trading on a podcast or webcast, an offer he politely declined at the time as he didn’t want to speak about his trading strategies on the air. But then three weeks or so ago, I ran into Anant again in London at a conference and asked him again, and this time around, luckily, he agreed to come on, which is great because, as I’ve said, Anant has never been on a podcast before, so this is a true first.

The main subject of our conversation today will be systematic commodity arbitrage and spread trading. We’ll speak about location arbitrage trades, substitution spreads, as well as the importance of Greenland’s high-frequency trading tech setup for the efficient implementation of their strategies. So without any further delay, Anant, welcome to Open Interest.

[00:03:10] Anant

It’s a pleasure to be on, Moritz.

[00:03:13] Moritz

Well, it’s been a long time in the making. It’s great to have you on and we’re looking forward to that. You’ve been very quiet about the things that you do, so this is going to be interesting, I think, I’m sure actually. But hey, before we start, Anant, I know you’re based in Mumbai these days, but could you please give us some additional background on yourself, Greenland commodities, the firm that you run?

I remember when we met in London a couple of weeks ago, you mentioned that your family runs or used to run a commodity merchants business focused on steel scrap and that physical commodity trading conversations used to be pretty much normal around the dinner table at your house when you were growing up. So, please, if you could, tell us a little bit about how you grew up, your connection to the commodities world and all that.

[00:04:01] Anant

Sure, happy to. As you said, I grew up in a family that was a family of commodity merchants. My grandfather actually, back in the ‘80s, had started a commodity merchant business trading a host of commodities across the world. My dad and my uncle took that over. Now in the third generation, 40 years later, my brother is running that part of the business.

So for me growing up, being around physical commodity trading or conversations around physical commodity trading was just second nature. Terms like CFRs, bill of ladings, letter of credits are things that I knew of as a teenager because we used to hear about all the real-world challenges that physical traders actually face as these conversations in the family business were ongoing.

So I find myself quite fortunate that all these years later, I’m still involved in commodities trading, not necessarily from a physical aspect. But we can learn a lot from what goes on in those physical businesses and implement that in our strategies. A little bit about myself. I graduated with degrees in finance and computer engineering from the University of Pennsylvania.

I started out at AQR actually, not on the commodity side, but on the global stock selection side as a researcher and PM over there in the early 2000s and then found myself back in India. I co-founded a hedge fund onshore with two other colleagues. It was actually the first hedge fund onshore in India.

Finally, a few years later, I found my calling in trading global commodity arbitrage strategies and found that the insights that we could involve from our physical business in implementing our strategies was just tremendous. The mix of taking a systematic structured approach to investing in commodity location spreads plus the information and insights from our physical business was a really powerful mix and a different take on trading these commodity markets.

[00:06:39] Moritz

Thanks for that Anant. Actually, I’d like to go back to the AQR days now that you mentioned it. You said you joined them in the early 2000s. I think they got founded in the late ‘90s or mid ‘90s, something like that. So you were at AQR at a relatively early stage. How did you find it there? As much as we all love Cliff Asness, actually, I like him a lot. I think he’s funny. He’s absolutely great listening to.

You were working on equities. Was that something you found out that’s just not for me, you’d like to go back to commodities and equities is not your thing? Tell us a little bit about how it was at AQR back then.

[00:07:14] Anant

Sure. The week I joined AQR was actually the week of the quant crisis back in August of 2007. So definitely a really interesting time to see how a firm reacts with a tail event taking place and I was thrown right into it. I think I’m really fond of my time that I spent there at AQR. I think it was some of the smartest people that I was fortunate to have an opportunity to work with.

I think my takeaway from there, in terms of how it shapes my investing today, is really about how AQR used to think about risk management, really careful portfolio construction. We actually use a lot of those insights and ways of thinking in how we even manage the book today. Whether you’re doing that within equities or applying a similar concept to other asset classes, I think the takeaway still remains the same. I was really fortunate to have been there.

At that time, AQR was a much smaller shop, still at around, I believe, 20 billion or so while I was there. It’s obviously significantly larger now. But I think a lot of my thinking about systematic investing is really driven by my time and experience at AQR.

[00:08:46] Moritz

Cool. Awesome. One of the things that I’d like to touch on, and this is maybe a distant memory for me, but when we first met in Miami, I think a couple of years ago, you mentioned that you had your start actually in the high-frequency trading FX market-making business, so kind of different to commodity arbitrage trading, what it is that you do today. So I think your firm, maybe you can speak a little bit about that.

How has it evolved? How did you start the business? Which focus did the business have initially and how did it evolve over time?

[00:09:23] Anant

Sure. Actually, when I started Greenland, we originally started out in 2013 as a prop trading firm. We were just trading my family capital at that point. We were doing two things which seemed a bit distinct from each other but are still highly related. The first was really doing high-frequency market making in FX. We used to trade FX forwards versus FX futures, make markets in the interbank markets and then hedge ourselves in the futures side.

We still do that today, and that was actually what led us to go out and build high-frequency trading systems, low latency systems, co-located servers across the globe. It was actually primarily driven by that business. But even back in 2013, even at that point, we were also running these commodity arbitrage strategies at the same time. Then along the way, you realize there are actually a lot of synergies that you can get having those HFT systems in place as how it applies to our commodity portfolio as well.

[00:10:47] Moritz

I have no connection to the HFT space whatsoever. How difficult is it to set something like that up? Were you capable of doing that with your degree in computer science? Do you need to fly around the world in order to connect with these exchanges and co-locate with them? Tell us a little bit about how that part of the business works, which really is still a bit of a black box to me to be quite honest.

[00:11:12] Anant

It’s not an easy business to say the least. It does take a lot of effort to get the right systems in place. You’ve got to get your, as you mentioned, the physical servers in each of the locations that you want to be co-located at. You’ve got to work with telecom providers to find the fastest lines that can connect to different servers.

You have a whole tech stack that you’re building out internally in terms of connecting into the exchanges, processing orders, or processing market data, and then on top of that, you’ve still have got to build your alphas and your actual trading signals that should hopefully go out in the markets and make money. So it’s a pretty large undertaking that you have to do.

You can start out by working with some providers that can provide you some hosting to get started with. You can work with some low latency algo system providers as well. But then over time, if you really want to be competitive in the space, you pretty much have to in-house all of that and you can’t really rely on third parties. I’d say in that space in any case, we were not working with the objective of trying to be the fastest HFT player in the world.

We ran those strategies in a manner that allowed us to take meaningful amounts of overnight exposure so we didn’t have to go on flat at the end of the day. It allowed our signals to be differentiated enough from the other HFT market participants so we didn’t have to purely compete on speed. But if you did not have at least the base case co-located systems and still operate in microseconds, it would not be a game that you’d be able to play.

Those systems are just table stakes if you want to play those games. But there again, how much you want to push that speed boundary is something that you can decide in relation to the types of alphas that you’re looking to capture.

[00:13:31] Moritz

This is a part of the business that you set up in 2013 and you were still a prop shop back then I think you mentioned, but you have decided to maintain it to the present day. Later on in the conversation, I think we’ll speak about why that is important for the things that you actually trade at the moment, right?

[00:13:51] Anant

Absolutely. That is something that we set up way back then. We still run that program today. It is a highly capacity-constrained program as you would imagine for anything that’s high-frequently traded. But we still continue to run it. It’s worked well for us and we continue to invest in those systems as well, because not only it benefits that particular program, but it benefits the rest of our investing that we do as well.

[00:14:34] Moritz

Maybe let’s dive into the things that you’re doing these days. By the way, for the benefit of the audience, we won’t be speaking about performance numbers or anything but I think it’s fair of me to share that when I first met Anant, you were I think something around 100 million or maybe slightly below that, I don’t remember, but you’ve scaled quite successfully, shall I say, to north of 1 billion in AUM at the present time, which is just great.

I had a look at your returns. Obviously, people can find you through your website which we’ll share in the show notes and contact you. But your returns are really amazing, I should say, and very differentiated and uncorrelated to what other systematic traders do. So framing it, you are a systematic commodities arb trader.

What you’re doing is not based on fundamentals or discretionary analysis of one market versus another, but you’re using your technology and your tech stack in order to find out arbitrage opportunities across a number of different commodity pairs and relationships that you have defined. Then you will essentially trade them, sometimes hold them for days or weeks, and sometimes scale up and down positions and exposures even on an intraday basis. Is that a fair intro before we start getting into some of the details?

[00:16:05] Anant

Yes, absolutely. But I would also add to that that our entire thinking and philosophy is still driven around the fundamentals of these markets; how different agents in these markets, whether it’s the commodity merchants or the producers or the consumers might be thinking about holding or trading of these commodities, and then we look to just apply a systematic decision-making process on how we go around betting on these fundamentals that we’ve looked to model out, and then trade it in this best execution manner through these high-frequency trading systems.

[00:16:48] Moritz

Clear. Yes. I remember, like you say, it’s all around transportation costs and these types of things which are near and dear to your heart. You were essentially brought up in a family that was focused on transportation costs and getting ships to run from one continent to another. So this reflects in your systematic trading positions today. Maybe let’s take an example.

We can start with something that let’s just say as an easier example of a location arbitrage trade where you’d be arbitraging the price of gold or gold futures in one location versus the price in another location, say, COMEX Gold versus Shanghai Gold. That is something that you do?

[00:17:32] Anant

Yes. That is something that would fit right within our wheelhouse.

[00:17:38] Moritz

Shall we run through this? How does such a trade come into existence? What are you looking at? What’s important there?

[00:17:48] Anant

Sure. That’s a fairly simplistic trading example there. If you really look at the quality of gold that trades on these different exchanges, the first thing to think of is, is the gold actually fungible? Is it the same grade of gold? Is it 99.95 gold or 99.99 gold? Can that actually be picked up from one location and delivered to the other?

If you’re looking at say a COMEX Gold or an LBMA Gold and looking to sell it into Shanghai, questions that will come up is what is the cost of transporting gold from one of these markets? If I have gold in a vault in London or in New York and I want to actually ship it to Shanghai, what’s that end-to-end cost going to look like?

Gold goes by air freight service, so you’ve got to figure out the cost of actually moving it over there. You’ve got to price the gold in Shanghai in appropriate currency-adjusted terms because the local gold there trades in RMB. You have to price it appropriately in USD taking into account when you would actually be converting your USD to RMB or when you would receive RMB for delivering your gold into China and then see if, given the pricing differentials that are trading across these two markets, if net of this end-to-end logistics cost, is there an excess spread available or not?

Based on that, you can say, if let’s say Shanghai Gold is trading at a premium net of all of the end-to-end logistics costs, there is a genuine physical arbitrage that’s possible that a gold importer in China could take advantage of and is very likely going to take advantage of by actually going and physically doing that particular arb, and that’s really what we’re going to look bet on.

So we’re going to go short on Shanghai. We’re going to go long the global gold market, whether it’s London or New York that we’re pricing against, and look at what that pricing differential is, effectively betting that as the physical gold traders move the inventory from one of these markets into Shanghai, that’s going to increase the supply in Shanghai, it’s going to reduce the supply offshore, and that’s going to get these markets back into an equilibrium causing convergence of the spread to eventually play out.

The way we really think about this, from a first principles philosophy, is that what is the genuine physical arbitrage that a commodity merchant involved in this commodity would be able to take advantage of? That’s really what we’re going to bet on. We’re only expressing our views through the paper markets or the derivative markets. But the underlying thinking is, what is a physical player going to do given this particular scenario of relative pricing across these two markets?

[00:21:19] Moritz

So you will take positions in the futures markets in this example. You wouldn’t be seeing the trade to its end or see it through on the physical side, right? In your example, the Shanghai-based physical gold dealer may actually be taking advantage of the arbitrage opportunity here by engaging in a physical arbitrage trade; actually ordering the gold from New York or London, having it shipped, paying for the freight, have it arrive in Shanghai where the local business needs it, and then essentially, they were able to source it for a cheaper price. But you don’t go through all of that trouble. You take the position in the futures markets and you wait for that spread to close and narrow and come back into equilibrium for preferred pricing.

[00:22:08] Anant

That’s correct. We’re not looking to actually hold the physical gold or actually ship the physical gold. Given our physical business, commodity merchant business is part of the family. If we really wanted to do that, we could probably get it done through that business. But that’s not the intention through our investment programs to do any of that.

We’ve never had to do that in our history. But as we are relying on the commodity merchants to actually do that trade and get that market in equilibrium, we are very sensitive to events that might disrupt that particular flow or that trade flow of that commodity.

So the arb only works, or that spread only converges if, one, physical traders can actually physically move that material and take advantage of that excess return and that there are enough of those physical players such that enough inventory can move over to address the excess demand that is there, say, in the Shanghai market in this example.

We are very cognizant about specific events that might be taking place that might lead to some distortions in those trade deals.

[00:23:35] Moritz

When we go one level down with that trade, gold has a very stable forward or futures curve. It’s essentially a funding trade. It doesn’t have the backwardation and contango dynamics or seasonal dynamics which some other commodities, say the agricultural commodities have.

But in order to get that, in order for you to get interested in the trade, do you actually keep track every day or periodically of what the cost of air freight, insured air freight of gold from London to Shanghai, or from New York to London, or from New York to Shanghai actually is and what the insurance premia are? Do you have your market intelligence and your feelers out to get updated prices on that? Because it sounds to me that without that information, you can’t really know whether the trade is attractive or not.

[00:24:28] Anant

Correct. With gold, there are two important aspects over here. Absolutely, we need to know, and we do actively track on a daily basis, what the insurance premiums are, what the air freight is for transporting the gold across each of these routes that we care about. We price what the cheapest delivery path for the gold is going to be and that’s what we’re going to look to define as our fair values.

But as you said, because gold trades a lot more like a currency than a pure commodity and very different from an agrigultural commodity, the spread that we’re looking to trade or the volatility of that spread is quite muted. You’re not going to get, most of the time, extremely large deviations that you can take advantage of. You’re really going to play for some bips when you’re looking to take advantage of the spread movements across that fair value.

So here is where our high-frequency trading systems come into place. Because if you don’t have really good execution systems that can take advantage of a few bips deviations in the spreads, in something like gold in particular, it’s not a market that you can actually participate in for most of the time. At times, in stressed environments, that spread can widen out and widen out meaningfully where you can participate in that spread without having high-frequency systems in place.

But let’s call it 95% of the time, you would actually not be able to put any exposure on because you’re sure that slippage and market impact that are just entering and exiting the positions will eat up any expected return that you would have from trading that particular spread or trading the deviations of that spread from your arrived at fair value.

So your ability to then get a really tight estimate on that fair value and then having the right execution systems that allows you to trade small deviations but frequent deviations from that fair value is what will allow us to actively participate in that trade and put on meaningful risk where we’re not just waiting for that two or three sigma event to occur before we can put risk on with a trade like that.

[00:27:20] Moritz

This shows why your HFT setup is so important. Actually, about a three-sigma event, I want to get to that in a second. There’s something I have in the back of my head that I want to bring up. But before we go there, there’s also I guess a component that, with respect to the specific trait we’re talking about, you need take into account is, if you’re trading in Shanghai and you’re trading say on the COMEX or in London, you presumably don’t have any margin netting benefit there because you’re facing two different clearing houses, one in China, the other one in New York.

I’m not sure what your PB setup type, but you probably deposit margin in these two different venues independently without any netting effect and there are always with brokers (unless you self-clear, I’m not sure what your setup is), there’s always a funding spread that you pay on that margin which you also have to take into account here.

[00:28:13] Anant

That’s correct. Because we’re trading across different exchanges, we are funding margin on each side. Our typical margin to equity that we run tends to be a lot higher than say a lot of other CTAs or other spread traders that might be there. You’ve really got to pay attention to your cash management and ensure you have sufficient liquidity given that you are running a higher margin to equity.

We do work very closely with our clearers to minimize the impact of funding costs on the margins that we are placing. Particularly now as rates have risen, the team spent a lot of effort ensuring that we can get not entirely up to risk-free on all cash that’s sitting with us or on the entire equity, but a very reasonable spread to cash to risk-free on the entire equity that’s sitting with us. So the net safe funding costs as such is not as intense. But you do also have to think about the funding cost that the physical trader is going to incur.

So now the physical trader is moving that gold physically from New York or London. Earlier when rates were zero, they could move it with a lot smaller margin because their funding costs were so low. But as that funding cost rises, they also have to think about the margin that they are baking into their trade and that has to reflect in our models as well.

[00:30:00] Moritz

In the back of my head, because you mentioned that five sigma or whatever sigma event, during COVID times, and earlier you said, your trade can also depend on whether the physical world is actually able to move the commodity from A to B and affect the arbitrage. But during COVID, some ships stopped sailing and the flows didn’t work anymore. I don’t remember the specifics, but there was this moment in time where people were very focused on arbitraging or possibly arbitraging gold between London and New York. Remember what I’m talking about?

Apparently, there was a big disconnect in prices and you could ship it from or fly it from London to New York. I think this was the direction if I’m not mistaken. But you couldn’t really get it done because you couldn’t move it, or it was difficult to move. Was that the trade?

[00:30:54] Anant

Exactly. When COVID hit, all the planes were grounded. If you recall, countries didn’t want people coming in from other countries in the world, so all planes were grounded. Most gold in the world actually gets shipped on normal passenger airplanes, in the cargo holds of passenger airplanes. It’s not that it’s going in some dedicated air freight. It doesn’t go by ship because it’s too expensive and it takes too long.

Gold being a high value product, it really just flies in normal passenger planes. As those planes went offline, you got to a point where you actually couldn’t close the arb. As New York started moving into a premium, you couldn’t get gold from London to New York and close the arb. Given the spec differences between those two golds, the actual arb is that it actually goes from London to a Swiss refiner who then shapes it into a form that’s acceptable on the COMEX and then it gets shipped out or flown to New York for delivery.

But as those planes were not flying, the arb broke down and that led to a large dislocation in that particular spread at that point and really widened up. We did not get caught up in that fortunately. Again, because we take this truly fundamental view as to what’s driving these markets, we were able to say, okay, look, planes are no longer flying, this arb no longer makes sense because you can’t physically close the arb.

We were able to, from a risk perspective, override our models to say do not participate in this arb right now. So we got saved when that particular arb extended out. Only when we started seeing air traffic pick up again did we allow the models to turn back on and actually start taking risk-on on these trades again and we could benefit as those spreads converged.

So it is very critical to understand what is going on in the physical markets because if you’re taking a purely statistical approach to this and trying to trade some reversion to some rolling mean, you can have these tail events which are driven by the underlying physical economics that make that trade no longer useful for a particular point in time.

[00:33:54] Moritz

That’s right. We see this over and over again. A couple of weeks ago, I had another firm on and we were speaking about grain arbitrage trade between South Africa and the US and Europe. Again, during COVID, these spreads completely disconnected. By the way, they’re also like you. They’re looking at the markets, really figuring out what’s going on. They’re not just using a say, Z score or a rolling deviation from the mean in order to figure out whether a trade makes sense or not. But still, you can easily get caught in these things and this can be extremely painful.

Maybe we leave gold aside for now. We used this as a relatively simple example. Before we maybe speak about one or two others which I thought are really interesting, how many markets are you actually looking at? We just mentioned Shanghai Gold, so there’s already a Chinese market there. I know that you’re trading the majority, if not all of the onshore Chinese futures markets, but maybe give us a background on how many line items do you have in a portfolio, how many spread relationships do you consider and all that.

[00:35:14] Anant

Sure. Today, in the program, we trade 25 unique relationships in the program. That translates into around 40 to 45 line items that we might be trading. If I’m, say, trading gold across four different markets, that’s one relationship for us. We run a fairly diversified program across the different commodity sectors. It’s pretty much across precious metals, base metals, agricultural commodities, as well as energy.

We do want to run a diversified exposure to these relationships that we want to be betting on. Part of what we try and do from an operations perspective is actually try and get market access to as many markets that we can get where we have liquid commodity markets trading. Recently, we went live on trading on our Q3 license in China.

So we can now access China onshore futures as a qualified foreign investor and trade it like any other global futures market with its pretty complex operational setup, time-consuming setup to get in place. But we think that that gives us a meaningful edge in terms of just having maximum market access and allows us to diversify the program quite meaningfully.

[00:36:50] Moritz

Got it. You just mentioned copper. I presume, correct me if I’m wrong, that like a copper arb trade between say COMEX Copper and LME Copper, it kind of works in the same way that we’ve just chatted about the gold trade where it’s like you’re factoring in transportation costs. Maybe copper now actually sails on the ship and doesn’t fly through the air. But is it fair to say that it would be the same rationale behind it or the same driver that you’re using for the gold trade?

[00:37:22] Anant

As you move on to something like copper, that’s where these trades actually get a lot more interesting because complexity actually goes up more exponentially as when you’re looking at copper or you can look at Shanghai Copper, LME Copper, COMEX Copper, and we can trade a combination of all of these against each other. Now, when you’re thinking about the transportation costs, you’re really looking at every possible warehouse to every possible other warehouse in an opposite market.

The LME, for example, has global warehouses not only in London. It’s got warehouses in Rotterdam, in Malaysia, Singapore, in the US. China has its warehouses in Shanghai for Shanghai Copper. So now when you’re actually looking at these transportation costs, it’s no longer one or two locations that you’re trying to price or one or two routes that you’re trying to price.

You actually have to price this from every single possible warehouse of the LME to the other market that you’re looking at. Whether it’s any possible COMEX warehouse, which again has multiple warehouses across the US, or to, say, the Shanghai set of warehouses for Shanghai copper.

[00:38:45] Moritz

How do you get about obtaining these transportation costs? Is there a specialized broker market out there that you can phone up and say, hey, what would it cost me to deliver from one of the LME warehouses, say the Malaysia warehouse of the LME to Shanghai? Can you just call somebody or e-mail somebody and they come back with okay, this is what it costs today, Anant.

[00:39:12] Anant

That data is actually really hard to come by. You can’t really go out and subscribe to that level of granular data out there. The data providers on shipping data for example or shipping cost data give you very coarse data. You can get it from one continent to the other. But if I want from a specific warehouse to another or I want inland trucking costs, you really can’t go around getting that data by simply subscribing to it.

For this data, on our end, we lean very heavily on our commodity merchant business that we have, which is the family business. That business, due to its size and scale, has contracts with all the major shipping lines, the truckers, warehouses, insurance providers, so we can actually get from them every day for every single route that we are interested in the cost end to end of what it takes to move a commodity from one of those warehouses to another warehouse that we’re interested in.

That becomes this really key proprietary dataset for us that we can use to define our fair values, have a really tight estimate beyond what level physical arb traders will actually take in and use that to then trade systematically any deviations from that fair value.

[00:40:50] Moritz

That is a huge edge. You have that underpinning, that background, that connection to the physical world. Then you combine that with the alphas that you have, which are good as well. On top of that, you have the HFT setup which allows you to very efficiently get into these arbitrage trades even for a few basis points where other market participants essentially need to stay on the sidelines because they don’t have that trading efficiency that you can bring to the table here.

[00:41:20] Anant

Exactly. It is about the integration across each of these areas when put together. We think you get the best result when it comes to trying to trade these particular markets or these particular spread opportunities. Of course, we have a tight estimate around that fair value. You can then think of trading small deviations from that fair value. Because if your fair value itself had a very large error term, automatically, you’d have to wait for a large deviation from that before you can even think about participating because you’re just not sure as to what that end-to-end transportation cost might look like.

So now because we have a tight estimate of fair value, and then we have the high-frequency systems in place, we can configure our portfolio construction to say, okay, I can look to rebalance on a real-time tick-by-tick basis if there is sufficient movement in the spread where my view changes even by a few percentage points to go in and trade a marginal change in my exposure where I can either size up or size down my exposure to that spread because we can have certainty that yes, this fair value is highly accurate, it’s actually executable and now I know that I have HFT systems in place that is going to limit the amount of market impact and slippage that I’m going to face when I go in and trade into these spread markets.

The number of legs that you are looking to trade at one go in one shot could be anywhere from 2 to 5 legs depending on the complexity of the trade and the spread that you’re looking to capture. So the amount of slippage and market impact that you will face can be a very significant percentage of the P&L that you then give up if you’re trying to trade small deviations. It’s really the combination of it all put together, and that to trading a diversified set of these opportunities across all the major sectors, this is, we think, a really good value proposition.

[00:43:38] Moritz

Great that you mentioned the multiple legs that you may trade. During our prep call, you mentioned that those trades, I think you mentioned that (I don’t want to say favoring maybe is the wrong word) they play a large role in your portfolio, essentially sometimes like a substitute substitution arbitrage trade, but that trade then includes a location arbitrage as well.

One example would be, for instance, New York cocoa versus London cocoa, or even more complex, you could go, okay, we’re trading New York heating oil, arbitraging that against London gas oil and in some crack spread type of way, it goes into Tokyo crude. That is stuff that you do, right?

[00:44:28] Anant

Those are all things that we are happy to do. The underlying principles that we’re looking to bet on is that is there a strong fundamental economic rationale as to why these spreads, if they widen out, should converge back to some sense of what that fair value looks like? Even when we’re saying we’re happy to trade substitution arbitrage over here where we’re okay with slightly different grades (so like London cocoa versus New York cocoa), we like those opportunities more so when there is a geographic transportation cost element into it because we think we can bring some edge to the table when we’re trading that. More importantly, what we’re looking at is saying are these genuine substitutes for one another? Can a consumer who’s consuming one grade quickly switch to consuming another grade if the primary grade that they’re consuming gets too expensive?

For example, something that we would not look to trade would be Brent versus WTI. Even though they are highly correlated grades of crude, it’s really hard for a refinery to suddenly start processing Brent if they have been set up to process WTI. That spread between TI and Brent can keep widening and narrowing, but it’s not necessarily going to trigger a genuine substitution effect in the physical markets because the cost of reconfiguring the refinery to process the other grade is too expensive.

So we would stay away from those trades and really focus on only those sets of relationships that are going to be very tightly configured. Similarly on cracks and crush spreads, is there actual economic flow that’s going in from that one market into the others where you’re going in from your input commodities to those output commodities? Or are those markets genuinely just distinct from each other and having their own supply-demand dynamics with no way of actually bridging the flows between them?

Because if you don’t have true trade flows bridging these markets, then these spreads can actually dislocate and remain dislocated. So really diving into the fundamentals of these markets, as a starting point, to understand where the consumers are, where the producers are, and what are the linkages that you have between these markets is really critical before we even say here’s a relationship that’s part of the universe that we’re even good to go and research any further.

[00:47:36] Moritz

One market that I’d like to bring up, it’s just coming to mind as we speak, is nat gas. Because you have a molecule of natural gas, used to be fairly constrained around where that gas is produced, and then delivered to Henry Hub, TTF. But now that market has changed. LNG is growing. A gas molecule can now, I don’t want to say relatively easily, but it definitely can travel on ship on a specialized LNG tanker from one location to another and essentially find a home where the price is highest. There is a global market for LNG.

Have you become more active in this market given that it has in a way structurally changed to become more global or is it still a market that you stay away from, calling them, like you would with WTI and Brents, like, well, they’re really distinct where Henry Hub is one thing and TTF is really another and we’re not looking at any relationship there?

[00:48:42] Anant

We won’t comment on our specific involvement in any of these particular gas markets. But I think to your larger point, I totally agree that these markets, which historically have been actually quite distinct markets, are now actually getting integrated as global. Because as LNG capacity ramps up, both in terms of liquefying and regasifying on both ends, as LNG shipping capacity ramps up, you are effectively moving to a world where you are going to have a large single gas market where gas is going to go effectively to the highest bidder, barring any import-export restrictions that countries might choose to impose.

That becomes a market, over time, that historically we would stay away from, and more and more as these capacities come online, they become integrated and a true location arb kicks in over there. We’re not all the way there. We’re on that spectrum right now. How we specifically get involved today, in that market, is something I would not be able to comment on.

But I think this is an important aspect that players in these markets are pretty much aware of now and will be a significant driver for how gas is traded going forward. That’s no longer say a local Henry Hub market with various basis trades or locally and geographically only in the US. How export capacity is and how exports are going out is going to be as key a driver to pricing even for Henry Hub than any of the other markets.

[00:50:52] Moritz

Moving away from specific trade examples in markets, when you put it all together in a portfolio context, how do you think about that? Is there a rule that says, I need to have at least a certain number of spreads on at any given point in time in order to have a certain level of diversification which should be able to also run a very concentrated book because say the only thing that happens is actually in the gold market and the other markets aren’t interesting or nothing is really moving enough for you to get interested in it.

How do you go about putting all of these, I think you mentioned 25 unique relationships that you’re monitoring, how are you going about putting all of that together?

[00:51:36] Anant

Sure. When we think of each of these spread relationships that the system is trading, we like to run it as let each relationship run completely independent of the other. If, say, there are a host of spreads that are not on, we’re not going to take the risk that’s allocated to them and load them up on spreads which are on or have attractive opportunities at that point.

The nature of this game is that you’re anyways taking time-varying risk on individual spreads in the portfolio. If you are going to actively reallocate risk and load up on other interesting opportunities just because something is not on, you’re going to be doubling down on that risk timing that you’re doing anyways on these spreads. So we like to run them independently.

We like to size the overall exposures or the max risks that spreads can go up to such that an individual spread can give us a certain level of volatility contribution to the portfolio over time; we will go and statically allocate how much risk we want then, over the long term from individual spreads, taking into account what the covariance across the spreads are, taking into account the tail risks that similar spreads might be exposed to whether it’s sector risk or producer country risk or consumer country risk where you want to diversify those risks as much as possible.

But then once we do that static risk allocation, we don’t go and look to shift risk from one to the other just because a particular opportunity is looking attractive. We do try and target a constant volatility on the portfolio over time. But because we’re taking time-varying risk, what it means is that at any point in time, we might be running well below our target level of risk or our constant vol target.

At other times, you’d be running higher than that vol target. It is a function of how many opportunities are attractive and to what extent at any point in time. But that said, again because we trade in this continuum where we are putting small amounts of exposures on, even as spreads deviate minimally, it means that we tend to have a lot of positions on, on average.

It’s not going to be the case that the portfolio is going to only hold 2 relationships out of the 25 at a point in time. We’ll typically have some positions on and around 13 or 14 of these at any point in time. The size of those positions could obviously vary depending on the attractiveness of the opportunity.

[00:54:43] Moritz

Maybe a final question on that, on risk management. Is there a point also where there’s a stop loss or you’d say this is enough, the spread has moved too much against us? It has widened even more to a point where you say, we’re calling it quits, we’re leaving it, or which you really if you have determined an arbitrage opportunity that you think will play out, you would then essentially stick to that trade and see through to its end.

[00:55:15] Anant

When we’ve spoken of the arbitrage opportunities for starters, we’ve spoken primarily of the deviation from the fair value or the transportation cost. The models that we run are actually multi-factor models where we say that, look, this physical arb is the primary call it value factor. But then we do want to explicitly and quantitatively account for supply-demand, relative inventory levels, market sentiment factors, that can help determine how quickly this arb might comeback or whether it’s going to be stretched for longer periods of time.

But let’s say even in this multi-factor approach, if you get to a point where the spreads are continuing to expand against you, we do run automated drawdown controls where after a point, we gradually start taking risk down, and at further predefined points, we would automatically be 100% out of the spread. We think it’s really important to build in these automated drawdown controls when you’re trading the spreads because everything in finance is fat-tailed and negatively skewed.

Spread trading, in particular in any form that you construct it, is going to be even more negatively skewed. So you’ve got to build for negative tail events that are going to happen every now and again. It could have been because there’s a policy change by a government or suddenly duties or taxes have been changed, imports or exports have been restricted from a certain country that’s an important player for that particular commodity.

You can get tail events in these spreads as a result of that. So it’s truly important that you are in a position to very quickly and automatically reduce the losses that you face in those event scenarios, and hopefully, because you’re running a diversified program over here, the impact of that event is fairly muted on the portfolio as a whole.

[00:57:28] Moritz

Speaking about the skew and maybe in preparation of closing this conversation down, which by the way, I think is absolutely fantastic. But on spreads, there’s more of a tail risk, like you said, something happens, policy change, import duties, weather risk materializing, COVID, all these types of things. But is it fair to say that it’s really negatively skewed?

It does have the tail exposure on both ends. It simply is, I would say, more tail exposed, a more leptokurtic distribution. Whether the tail is negative or positive remains to be seen. It can go both ways. Would you say it’s really a negatively tailed exposure that you have there?

[00:58:16] Anant

It is going to be negatively tailed because principally, you’re still trading reversion over here. So if you are crossing fair value, you’re anyways exiting out of your positions. So highly unlikely unless there’s a huge gap event in your favor, it’s highly unlikely you will benefit from a tail move.

[00:58:43] Moritz

Absolutely. No, you’re absolutely right. You are more the mean reverting fair value reversion type of trader, whereas some other spread traders, they may actually look to say trade momentum in these spreads and they may have a different tail experience than what you’re describing.

But you’re absolutely correct. Kind of like with any mean reversion strategy, right? You’re getting in at a point where you think a spread is stretched, and that by itself because you want to go back to the mean, but if it doesn’t go back to the mean, you really only have the tail on the other side that can sting you.

[00:59:18] Anant

Exactly.

[00:59:20] Moritz

Anant, fantastic. Is there anything that you’d like to bring up, anything that you think we should speak about but didn’t?

[00:59:28] Anant

No. Thank you, Moritz. I think we covered all the key facets of how we think about the world. I think the key takeaway, if there is one, is that we really think about this world as physical commodity traders first and how they might be reacting to these markets, and then look to just systematically trade around it. That’s held us in really good stead over the years as we think about taking this program further.

[01:00:00] Moritz

Fantastic. Thank you a million for doing this. I know it’s been a long time in the making. We wanted to get you on for a long time, we finally made it happen. Listeners, please make sure you follow Anant and Greenland.

As ever, should you have any questions, be that to me or to Anant, please email us at info@toptradersunplugged where we’ll pick it up and absolutely respond. Thank you big time for listening. It was really great. We’ll soon be back with another episode of Open Interest, so please stay tuned.

[01:00:32] Ending

Thanks for listening to Top Traders Unplugged. If you feel you learned something of value from today’s episode, the best way to stay updated is to go on over to iTunes and subscribe to the show so that you’ll be sure to get all the new episodes as they’re released.

We have some amazing guests lined up for you and to ensure our show continues to grow, please leave us an honest rating and review on iTunes. It only takes a minute and it’s the best way to show us you love the podcast. We’ll see you next time on Top Traders Unplugged.