The Deutsche Bank team's inaugural podcast, 'Stochastic Conversations,' showcases the intersection of quantitative analysis and market insights, as highlighted by the role of leaders like Vivek Anand in cross-asset quant research. Per the full note source, this series aims to explore the stochastic aspects of market movements by discussing varying backgrounds in quantitative finance, emphasizing the randomness inherent in market dynamics. As volatility persists, understanding these quant strategies becomes critical for positioning portfolios effectively across asset classes, particularly in uncertain market conditions.
What the desk is arguing
The podcast series illustrates the importance of quantitative analysis in navigating financial markets amidst growing uncertainty. The discussion hosted by Kayo with Vivek Anand highlights how blending different quant strategies can meet client objectives and adapt to dynamic market forces. This realization underscores the necessity of a diversified approach in portfolio construction as volatility remains a hallmark of current market conditions.
Vivek's background as a mechanical engineer, which he transitioned into quantitative finance, reflects a broader trend towards incorporating diverse skill sets into financial analysis. Fluctuating market dynamics require a fresh perspective, making the discussions in the podcast particularly relevant for traders looking to adapt their strategies. Parameters, such as risk management and diversification strategies discussed in these conversations, are essential as firms aim to navigate and capitalize on market volatility.
Where it sits in our coverage
Following our coverage of currency movements, we maintain a consensus target of 1.075 for the EUR/USD pair, with a range from 1.04 to 1.12. Notable firm targets include:
The current desk view aligns closely with jpmorgan, placing it within the upper end of the range while diverging from bofa, which suggests a more conservative outlook.
How other firms see it
Market sentiment is corroborated by firms such as jpmorgan and goldmansachs, which are aligned in their bullish assessments. Conversely, bofa takes a more cautious stance, indicating that downside risks are more pronounced in the near term.
In this context, fluctuations in the EUR/USD pair offer an essential barometer for assessing the broader implications of central bank policies, particularly from the ECB as they continue to navigate inflationary pressures amidst varying economic signals.
01The 'Stochastic Conversations' podcast enhances understanding of quant finance in current volatile markets.
02Quant strategies are critical for portfolio construction in uncertain conditions, aligning with the current market emphasis.
03Vivek Anand's engineering background exemplifies diversification of skills leading to innovative financial strategies.
04Market signals, particularly from the EUR/USD pair, reflect broader economic trends and central bank policies.
Market implications
Focus on the EUR/USD pairs as they reflect central bank policy shifts and market volatility; a significant break above 1.10 could trigger a bullish sentiment shift in investor positioning.
Risks to this view
A change in central bank policy or unexpected economic data that deviates from current projections could challenge the effectiveness of quant strategies and compel a reassessment of market dynamics.
All right, hello everybody, good afternoon, good morning, good evening. I don't know where you are, but we are here in London. It's a fine, beautiful morning.
It's raining, of course, because it's London, but then again, otherwise it's very fine. My name is Kayo, and this is the inaugural session of our Stochastic Conversations podcast, and I suppose one of the important things about the Stochastic Conversations podcast is the stochastic nature of it, which means that there will be some randomness associated with that. It is also, I suppose, by design, a term that is widely used in the quant community, and this is in order to make sure that you understand that it is a quant podcast, and we're going to be talking about quant stuff.
In this inaugural version, I have with me Mr. Vivek Anand, who is a senior member of our quant research team. He's going to be telling us about himself, what he does, his background, and then also later, he's going to be answering some stochastic questions from us.
So why don't I start with the questions here as the host. Vivek, could you share a bit about your background? You're the director of cross-asset quantitative research at Deutsche Bank in London.
What does that role entail, and how did you get there? Sure. Thanks, Kayo.
Yeah, I'm happy to share my journey. So as I mentioned, I'm currently director of QI's research team at Deutsche in London, which essentially means that I lead my portfolio construction effort here. So in simple terms, my role involves bringing together different quant strategies across various sector classes and blending them in a way that meets our client objective.
Now let me answer how did I get here, because it's a quite interesting journey in my view. So I actually started out in a very different field. I graduated from IT Bombay in 2006 with a degree in mechanical engineering.
Oh, you're an engineer. How unusual for a quant. Yeah.
It was a dual degree program for five years, and with a specialization in computer-integrated manufacturing. So there's a word computer embedded in it, but it is related to mechanical, I would say. And early in my career, I had a brief instant in different companies like ITC, Siemens, which basically give me a lot of experience in the beginning.
But my first significant step into finance was my stint at SunGuard. It's a financial software company, as you might know, based in London, where I worked in credit risk analytics. Then during my two years there, I gained valuable experience in my risk management.
And then in 2009, thanks to you, Kyle, I managed to join Deutsche Bank in FX's sales team. We were young and innocent back then, weren't we? I had more hair at that time.
You had more hair, indeed. So at Deutsche Bank, I was really doing into building these systematic quant strategies for currency markets, exploring intraday trading patterns, reversal, breakout strategies. And I remember we had a lot of discussion debate at that time when we started.
We had lots of discussions and debates after that, too, over the years. Yeah, that's correct. So after a few years focusing on currencies, I had opportunity, again, thanks to you, Kyle, to broaden my horizon.
So in 2013, I was moved to UK, London, to join a new team, which is a cross-asset quant strategy team. So the cross-asset means we were not limited to one market, I would say. We looked into opportunities across equities, bonds, commodities, currencies, you name it.
That was a very big shift to me because coming from a different background and joining a London team was a huge opportunity for me. So as… It's a bit cold, though, no? Yeah, it was a bit cold, but I enjoyed it because… You enjoyed the cold?
The weather, everything was… the air, I would say. You enjoyed the air? Yes.
That is, okay. That's the first time I hear that. Yeah.
So as I got deeper into this research, I became very interested in portfolio construction and risk management modeling, essentially, where we put all the strategies together in a portfolio that can weather all these different market environments. So I worked on designing portfolio that could serve a specific purpose, for example, a defensive portfolio that helps to protect during market drawdown or market neutral portfolio that can tailor to various macroeconomic scenarios. So to sum up my journey, Kyle, I would say I started as an engineer, transitioned into finance as a quant, honed my skill in FX markets, and then expanded to cross-asset perspective.
So over the last decade and a half, in fact, this year I am… Today, in fact, I'm celebrating my 16th year of… with the Deutsche Bank. Oh, fantastic. Yeah.
We're going to get you a cake. Right. In the UK, the practice is that the person celebrating is the one that buys the cake.
You buy the cake, and then we eat the cake that you bought to celebrate you. Yeah, that's my 16th anniversary today. So yeah.
Congratulations. Thanks, Kyle. And thanks to you as well.
So, yeah. Okay. So now tell me about your team.
What does cross-asset mean in practice? I mean, just tell us a bit about that. Yes, sure.
So let me start with what cross-asset mean to us and in generally to the quant community. So in a nutshell, cross-asset means that our team's research spans the entire spectrum of asset classes. So we look at equities, single stocks, rates, effects, commodities, and also look at more niche area like credit, volatility as well.
So I would say it's actually one of the fun part of the job where we are not siloed into one domain. We diversify our research across multiple asset classes. Now about our team, I think it's important to understand how our team is built and grouped together.
So we are essentially people of eight-person team, all based out of London. It's good to sit all together, right? Oh, yes.
It gets noisy a bit, doesn't it? Sometimes, yes. And some off discussions also are very useful, I would say.
Yeah. Especially on cricket or maybe something else. For a quant to be very opinionated about things.
When they have an opinion, they're very opinionated. So let me break down the rules a bit because we have a bunch of people with a mix of skill sets, right? So out of eight, six of us are quant analysts focusing on either developing signals or doing portfolio construction.
And two of us are quant developers. And quant developers are absolutely integral to our operation. So think of them as an architect or engineers who build the infrastructure that keeps our quant investing running.
So they are very essential part of our job and they help us to make our life easy in productionization everything and make it running all the time. And other six of us are quant analysts. They also have a whip that in case we're coding wrong, they come in there and they hit us with a whip, right?
Yes. Best practices and all that. Best practices and they help us in a lot of areas.
No spaghetti code, none of that, right? Yeah. Anyway, sorry, I interrupted you.
No, that's right. So and other six of us are, of course, we do quant. We are quant analysts and our role is more on research and strategy design.
So we have a bit of unique philosophy in our team. So each of us is a specialist in some area, but also a journalist in others. So that's a very important aspect which I like and which we have been developing, I think, over the years.
So it means that if, let's say, a person or colleague primarily specializing in cash equities, for example, so he might spend a lot of time in developing a stock selection model, like equity factor strategies, things like that. But he's not working in a vacuum. He also stays involved what is happening in other areas, for example, portfolio construction and so on.
And if there is a client meeting where he needs to talk about portfolio construction or options and things, he can do that. So that's why he specializes in equities, but he has a broad understanding of other as well. So that is very useful.
Another thing I would like to emphasize is that we are diversifying three into three dimensions of our research team. So one is asset class, strategy style and time horizon. So basically three areas, three dimensions of diversification, to put it that way.
OK, I like that. And by asset class, I mean we want to cover as much as possible. So as I mentioned, we cover equities, fixed income, FX, commodities, credit and so on.
By strategy style, I mean we explore different kinds of strategies. Some are based on fundamental ideas like value or carry, and others are technical like the price based momentum, mean reversion and so on. But we also look at strategies driven by alternative data.
I think you are spearheading that front, Kyle, in bringing a lot of alternative data as well. So I think that's another area where we are moving and diversifying. We're looking to a lot of alternative stuff.
Yes. And then in time horizon, of course, we have done a lot of work on short term frequency as well, including some longer term frequency models as well. So just to summarize everything, we cover a lot of ground and we work as one cohesive unit, kind of an assembly line that turns very raw materials, data and ideas from all of the market into a finely tuned product.
Right. Very well. You talked a bit about signal research.
What do you mean by that and what is your approach to it? And going back through the years, I do recall you being involved in signal research across multiple asset classes and so on. So tell us a bit about your approach to it.
Yes, I think that's very important because as an important part of time we spend is to do signal research, not only me, but other part of our team members. So when I mention signal research, it is mainly about the work we do to create systematic trading strategies. In quant investing, I mean, there are two big stages, right?
So first, you develop the strategies to understand certain patterns on anomaly. And then second, you figure out how to combine those strategies into a portfolio. So signal research is basically the first part where it's identifying and building the strategies themselves.
Then the second part I'll talk about later. So what is a signal in this context? Put it simply, it is just a systematic indicator, a rule that tells us when and what to buy an asset versus another.
So it's a secret source behind a strategy, the pattern or anomaly that a strategy is trying to capture. It's like a command, so to speak, right? Your signal.
You've got to do this. You've got to do that. It could be.
Yeah, it is a kind of command where you understand what to sell and what to buy and when as well. But our approach to signal research is to. But those are those commands so-called are rule-based systematic strategies, which means that everything is defined, predefined, and we don't have discretionary on those signals or those command.
That's the important bit here. So did you like that? Not having discretion.
You just have to use. Yeah, I think it is very important because then you can able to diversify, right? Otherwise, you cannot diversify and extract all the signal possible.
So that's the difference. I think signal can come from many places. That's it could be something well-known and grounded in financial theory, for example, value premium or carry and so on.
But it could be observation of the market behavior as well, for example, like trend following momentum and so on. So if an asset is going up, it tends to go up. And that's could be a strategy as well.
There are also more niche or newly identified anomalies people explore, like certain pattern in options market or effects around investor behavior and so on. But key for us is that any strategy we develop should be backed up by sound economic rationale, which is very important in our view, some sort of evidence in academia. It should also show persistence over time and it should work historically in different periods, especially that we should be able to explain why a strategy is working and also why it is not working and what model we are capturing and what are the risks associated with those as well.
So these are the important element which we want to put into it. And I always emphasize that importance of having a clear economic intuition for a signal is paramount and because strategies usually work either because they are compensating for a risk or because of some behavioral bias in the market. And there are various examples.
So, for example, trend following, why would buying things that have gone up or selling things that have gone down yield a profit? So one argument is behavioral because investors often underreact to new information or are slow to change their belief. Right.
And so trend persists longer. You would expect that there would be some sort of behavioral hurting or some sort of hurting happening. Right.
Where people tend to choose or tend to chase performance causing trend to overshoot. On the other hand, consider a carry strategy like borrowing a low interest currency to invest in a high interest currency. The intuition is more of risk based here.
So you are earning that interest rate differential as a reward for taking on a risk. So so if nothing bad happens, you keep on earning this interest rate differential and if something goes wrong, then, of course, you will pay the price. So that's how strategy is done.
And they complement each other. Right. Yes, that's correct.
And these strategies, they have a lot of diversification. They complement each other. And that's why the portal construction become very important.
So our signal research process tries to identify such pattern and test it if we can exploit them systematically. And funnily, I mean, we call our approach a cookbook style. Sorry, what?
A cookbook style of research. A cookbook. Yes.
Like you're a cook. Yeah, that's correct. And that's an analogy we give here that we are a chef and we follow a certain recipe to do that strategy.
Do you cook, Vivek? I cook. I can cook a few bit.
Yes. What do you cook? I can cook pasta, for example.
You cook pasta. Yeah. Okay.
And pizza. Okay. Pizza as well.
But of course, I bake pizza, not from the scratch. I see. You're a fan of Italian food.
And I can cook some Indian food as well. I know. Of course.
Yeah. Anyway, cookbook approach. Yes.
All right. So in this style, let's go step by step. Okay.
So first, we want you to understand the driver of the market or research class we are looking into. So we are deciding what cuisine we want to cook, right, in the first place, which asset class. So we look at a variety of tools to understand and study market data and figure out what really makes price move over the long term.
So sometimes we use technique like PCA or regression, which can help to discern the main factors driving the set of asset. Other times we might use machine learning to uncover non-obvious pattern or non-linear pattern. And goal at this stage is to identify broader themes or factors.
For example, in equity market, key driver could be factors like growth, quality or investor sentiment. In bonds, it might be inflation expectation or central bank policy and effects could be interested differential or terms of trade and so on. In other words, we want to understand what drives the market and we use our statistical tool to do that.
Now in the second part, I mean, from those drivers, we construct signals. So that's the second thing. And once we have a hypothesis about, OK, about a driver or pattern, we translate them into a specific measurable signal.
It's important that we can quantify those drivers, right, to a signal. This is akin to taking the ingredients, right, and preparing them in a recipe. For instance, if data analysis suggests that value is a strong driver, say cheap currencies tend to appreciate against expensive currencies in the long run, we build a value signal.
That could be something like, let's say, ratio of purchasing power parity or a model estimated fair value to the current spot prices, right? And the strategy might go along the undervalued one and short the overvalued. If we think that driver is a momentum, then we might construct a trend following signal for say commodities, measuring the 12 month price momentum for each commodity future and go long those with a positive momentum and short those with a negative momentum.
So at this stage, it's important to determine whether a signal works on its own or in a relative sense, because that's another idea which have to look. And some signals works on absolute basis, for example, trend following might say that if gold is trending up, then we are long gold. It doesn't necessarily need to correspond in short, right?
On the other hand, a value signal usually works on a relative basis, what we have seen. So you buy a cheap asset and sell the expensive one, because value is a more of a fundamentally about relative mispricing. So we carefully decide best expression of each signal.
Is it market neutral, long, short within a sector, group or asset, or is it a directional position that goes long or short entire asset class as well? So I think a lot of very holistic and yeah, it's very holistic. And I think I remember we had a lot of debate on this area as well, when we did, yes.
And still do. So yeah, these are, I think, has helped us to do our research more efficiently. Yes.
And talking about research, let's move on to portfolio construction. You had a very interesting, well, you built some pretty interesting things on that front in portfolio construction, especially with the flagship Portfolio 365, right? Yes.
Which is your baby, so to speak, right? And it's a cross-asset systematic portfolio that you've built. It's been now live for over a year.
I read that you're, I mean, I understand that you group strategies together, but why don't you tell us more about Portfolio 365? What was the thought process behind it? And yeah.
Yeah, no, I think. And feel free to get very quanty on this. I think Tiare is- No, I think this is a very important bit where I spend most of my time into the portfolio construction.
So in the earlier section, we talked about signal research and how we build strategies across different asset classes, trying to capture anomaly pattern and so on, right? But the second stage, as I mentioned, it's mainly about portfolio construction, where we combine all these strategies in a most efficient way to meet certain objective. So as I hinted earlier, that once we have this menu of these individual strategies, the next question is that how one should combine them, right?
And that's basically looking to a full course meal, right? That is to satisfy a particular taste or a nutritional need in this case. So the Portfolio 365- Yeah.
So the Portfolio 365 is like a full meal. Is a full meal, right? You have the entree.
Yeah. And it's a- The appetizer and then the dessert. That's correct.
And that should meet a certain taste of the person who's eating as well. I see. If someone likes spicy, then you have to add spice.
Then you have to add spice to Portfolio 365, if the person is, I don't know, vegan. Yeah. You have to remove certain things, right?
Remove certain things. I see. That makes a lot of sense.
Yes. That's how we design. How you think about it.
All right. Okay. So the key idea behind this Portfolio 365, again, it's strategies, as I mentioned.
So first is to group the similar strategies together into buckets, so you can have different bucket or sub-portfolios. And then after that, you blend these buckets into a balanced way to suit that objective. We find that most strategies can be categorized into kind of market behavior in which they perform well and conversely when they struggle.
So in practice, we ended up three broad categories of strategies, each corresponding to a different part of market cycle or distribution of your market outcome. So these are, I mean, these three categories are firstly defensive strategies. So they are the convex strategies that act as a hedge or crash protection.
They tend to perform well during market drawdowns, recessions, or period of high risk aversion, basically focusing on the left tail of your market distribution. Okay. Defensive strategies.
Okay. Yes. So one of the features with this method of strategy is that they might lose a little bit of money during strong bull market or normal market scenario, but their main value add is obviously give you convexity during the stress episode.
That's their value add. So during bad times, they shine. Yes.
That's correct. Okay. The second category is pro-cyclical strategies.
They think of them as a mirror image of the defensive one. So they do best in the rising market or during normal benign market situation, conditions, but they might be, but they suffer from time to time when there's a certain stress in the market. Right.
So they have a negative skew, I would say. The third category is the balanced or so-called market neutral strategies. They are the one that aims to generate returns regardless of the market environment, typically with a minimal correlation to traditional assets and so on.
So they do best when things are calm, nothing happening, and they do well most of the time. But they, in other words, they have usually zero or low correlation and low skew. But they suffer again from time to time when either market goes suddenly up or down because of misestimation and risk, the beta, which basically makes their portfolio neutral.
I see. So it's basically you've got one class of strategies for good times, one for bad times, and then one for other times, if I can put it that way. And so it's a bit like when you're a cook, it's like having the carbohydrates and the proteins and then the- Yes.
So you make those food? You make a food group and you're putting them all together in a way that gives a very nice balanced meal. Yes.
Okay. So, yeah, you're right. So with these three buckets, I mean, the idea is that each plays a distinct role, as you just mentioned.
So defensive does well when the world is falling apart, pro-cyclical does well when your markets are doing well, and balanced grinds out the return in many periods in between. So that's how. Cool.
Now let's also talk about how we actually construct, because this is where I think the main philosophy comes into it, how we build these three sub-portfolios and then how we aggregate them. Yeah. So here we follow a systematic framework, being a systematic quantizer.
So we have to do systematically. So first step is that we define scenarios and we estimate the strategy behavior. So we start by explicitly defining the set of market scenarios we care about.
And we use here the primary factors like growth or inflation or equities and bonds to define these scenarios. So using growth and inflation, we can define scenarios such as a recession, stagflation, recovery or deflation and so on. So first is to define the scenario.
But in order to define scenario, we need to quantify them. And in here we use our set of macro risk factors. We have identified about 11 key factors that model the financial world holistically.
And these include MSCI world equities, inflation expectation, interest rate, credit spreads and so on. And we proxy. And the important thing is that we proxy these factors with the market data, not with fundamental data and so on.
And in doing so, what we do is that we first shock the system. So for each scenario, we paint a picture that how these factors would move. So for example, in a recession scenario, we think that equities would drop significantly, the credit spreads would widen, the government yields would fall as the investor feel free to safety and inflation expectation may decline as well.
However, in a stagflation scenario, which could be opposite, here the equities fall, but inflation expectation rise and yields also might rise too. Which means that these are the both scenario in which growth is down, but your other macro factors are different. So we need to model them differently.
So you need a lot of flexibility to model these things, right? Because it's not just a risk on, risk off world. And this has become more apparent after 2020 when we see other market drivers coming into the picture, especially inflation and how central bank had controlled inflation and how people are expecting how inflation would go, how central bank would react.
And also when we see a deflation trend in 2024 as well, right, where we see a lot of changes in expectation of the market every day or every week, right, when we see some news coming out of on the. And then in addition to that, you have other, maybe some shocks that are coming from some of the other factors of risk that, you know, are not necessarily directly tied to growth or inflation, right? And then you have to model that too.
So it's interesting that you have basically a macro risk factor, right? A way of modeling all the factors of risk at the macro level, because that then allows you to construct scenarios, right? That whereby you're shocking one factor or shocking another, or maybe shocking three factors at a time and then evaluating what is the impact of that on strategies.
So once we do that, Kayo, then the second step is the selection, that how we should select strategies for each sub-portfolio. So we apply, again, rules to select these strategies that should go into each sub-portfolio. We want each sub-portfolio to purely represent its intended role.
So for example, for defensive bucket, we want to have those strategies that show significantly positive performance in the bad scenarios like recession, equity bear market or spike involved, etc. But also we want it to show flat or negative performance in good scenarios as well. This is very important because something to be considered a hedge would obviously do well when that scenario happens, but it should relatively underperform if we enter into a different scenario altogether.
So that is another thing which we take into account and we use a lot of statistical tests, hypothesis testing, and so on, in order to do the selection process. So and similarly for post-cycle basket, we do the opposite, right? We want to pick strategies that tend to do very well in the good time, but perhaps underperform or lose in a bad scenario, right?
So we use this statistical rigor to do all the selection. So based on the rule-based selection strategies, we have strategies which could be part of these support for use, right? And to give some example, let's say in pro-cyclical, we generally have PRP strategies, asset classes.
We have FX carry, credit curve and carry, and some equity long-term dividends. The carry world. The carry world, which are more pro-cyclical in nature.
Then for the defensive, again, we have three set of strategies or sub-cluster. So first cluster is long vol strategies, where you long vol across rates, equities, and so on. Okay.
Okay. Second is your price-based trend following strategies across different tenors. So short-term, medium-term, and intraday as well, that can give you- Some convexity.
Some convexity in a cheaper way. And the third are usually the risk premia, which is defensive in nature, especially FX value or quality in equities or credit long-short and so on. So these are the strategies which goes into defensive.
Okay. So that's interesting because they also, even inside the defensive portfolio, sub-portfolio, you have things that are, you have these blocks that are complementary to each other. One is the price section block.
The other is the options block. And then the third one is the defensive risk premia block. Yes.
Okay. And they basically complement each other because during bad times, generally they tend to show higher correlation, but during normal time, they would have lower correlation, which means that they help to mitigate the bleed, which we generally see it with. So when having high correlation during bad times, it's actually a good thing in a defensive portfolio because that means that they're all rallying, they're all defending you.
Okay. And during good times, they de-correlate, which means that some of them may be losing, some others may actually be winning as well. Yes.
Yeah. Okay. Okay.
And then last, which is obviously the most important, is the balanced portfolio. And here we have usual multi-factor in equity space, then commodities and fixed income as well. And then we have the, I would say, it's let's say the long, short world.
Long, short world. Market neutral world, RVs and so on. Okay.
Okay. Would you say that most of the market neutral strategies in equities, that they fit in that balanced? Yeah.
So we have built these multi-factor strategies, especially in cash equities space, where we have used various approaches, but the approach which we like, and we have this portfolio is a dynamic way of choosing the factors which work in a particular environment. So we use machine learning-based strategies or machine learning-based rules in order to select the factors which is performing well, and we put more weight to those factors which is working in a particular, in the current scenario. Which means in other words that you're kind of capturing a factor of momentum.
Kind of factor of momentum, yes. A diversified factor of momentum kind of thing in cash equities, which is a thing in equities, right? Yes.
And also this works in equities because the breadth, right, of both asset and also the factor that help us to do this machine learning algorithm fit better than other asset classes. Yeah. It must be interesting for you coming from a non-equities world to suddenly look at thousands of stocks and find all these patterns.
Yeah. And it has done really well. Especially 2021, I mean, I would say when things were mainly driven by the fundamentals, not driven by just a low interest rate or cheap money.
I mean, the age of fundamental came into after 2021, right? And this is where our, this multi-factor dynamic strategy have been doing well, outperforming most of the benchmark in the industry. And in accurate long shot, yes.
Okay. Very good. Okay.
Now the third bucket is mainly, I think we alluded a bit, is allocating capital because this is where we, so once a strategy is selected into each sub-bucket, then the third is how we allocate capital. And here we mainly follow some sort of a cluster-based risk parity approach because we, in other words, what we do is that within each bucket, as you mentioned, that we have some sub-blocks as well. So we first allocate capital within each sub-block and then between the blocks.
So this is our way of allocating capital. And last, but of course, where the thing comes into is aggregating these sub-portfolios. And this is where the taste of the person who's eating the food comes into the picture.
So someone who wants more of a, let's say, defensive strategies, we would put more emphasis on defensive portfolio and add a bit of your balance to mitigate its bleed. But if someone wants more, let's say, all-weather portfolio or more of a market neutral portfolio, then our emphasis will start from the balanced portfolio and add a bit of pro-psychical defensive to mitigate some of the stale. And someone wants, let's say, very spicy food, then we go with more the pro-psychical portfolio, more carry, and then you add a bit of balance to make it less prone to the- If somebody likes curry, you'll carry.
That's correct. Curry usually spicy a bit, yeah. So this is how we do, yeah.
Okay. Now, do you track the things that you build? How does it work?
I mean, I understand that you're a research analyst, so you're not actually trading, but at the same time, it's important to see how the things that you've built are doing for various reasons, right? How does, not least the fact that there's usually investor capital committed to some of the things you've built. So how do you- No, I think that's a very valid question, and the answer is yes, we do.
Because designing these portfolios is just half of the job, I would say. But equally important is to track them and understand how they're behaving in real time. Because when we build portfolios, we look at the data in the past, and we may have a habit of some sort of overlooking things, right?
Because there's so much of data. So that's why understanding the performance, monitoring it on the real time, adds a lot of value. Not only it help us to track the performance, it also give a lot of learning lesson as well, which can put into our portfolio construction or exercise, which help us to enhance the model, I would say.
Yeah, yeah. It's the exposed world that you then compare to the ex-ante world, and come to conclusions. Okay.
So what we do is that we do a weekly update. So we do some sort of podcast and a report, and the podcast is called Quantables, where every Monday we recap what happened in the markets the previous week, and how our strategies, portfolios have responded to it. What has worked, what has not worked, and why it has not worked.
And we also talk about what's going forward as well, looking ahead in that week. So that help basically keep us understand what has happened, keep our clients informed, and it's one thing to create a portfolio that should perform across the NIL, right? But it's important to verify it continuously, whether it is performing expected or not.
Sometimes it does, sometimes it not, then we have to understand why it is not happening. So that then leads to the question, how has it been doing since launch, right? Just tell us a bit about the performance of Portfolio 365 relative to your expectations, and why have been the main drivers of that.
You're our synthetic portfolio manager, so how has that been going? Yes, I think we have seen a lot of some performance and underperformance as well in the portfolio. But one thing I would say that in terms of what we expected, the correlation and the skew profile of these strategies to behave, they are performed in line with our expectation, which means that the long-term correlation of my balance portfolio is sub 5 to 10%, which is a more of neutral, statistically insignificant.
And ProSequel has a correlation of around 50%, which shows that it has positive correlation and two equities. And Defensive has minus 40% equities over the past two years. So I would say in terms of that, but if you look into the performance, at first, I'd like to talk about how the markets have been doing, right?
Because that's what drives these strategies. And we see that since early 2024, markets have been shaped by a mix of fundamental drivers, right? And in sharp, exogenous shock.
So on the one hand, like this inflation trends in 2024 and shifting central bank expectations, they supported risk assets with U.S. equities led by large cap tech and AI related names. So they did really well last year with U.S. outperforming the other market. But however, the news coming from the tariff, these policy uncertainty, and also some of the geopolitical tensions which impacting commodities price swings, they have caused a reversal in the correlations and they have challenged these, I would say, these portfolios, which is based on...
Challenged the quant world altogether. Altogether, yes. It's not just...
Okay. And so, again, this backdrop, I would say Portfolio 365 has modestly negative since launch in early Feb last year. So it's down about 1.3% with a vol of 7%.
Though the performance this year is slightly positive, around 40 basis point year to date. But if we look into the performance since April 7th, it is up by 6% since we have the news of 90 day pause after tariff. So these, the main...
So that's also interesting. So you maintained the 7% volatility, which if I remember correctly, was your initial target as well, right? So that stayed on top performance, up and down, I mean, that's usually the case, but in terms of risk estimation and contribution, it's been going in line with expectations in this year and a half of a live performance.
Yeah. And the main drag, I would say, has really come from pro cyclical sleeve, which has suffered various episodes, like you would remember Yen carry and wine trade in August last year. And then a sudden episode in December last year, we had tariff headlines in April, which caused a lot of shock.
And we also see sharp commodity swings in Brent and Copper due to tariff due to war and so on. So all these shocks have caused repeated drawdowns or negative skew to these pro cyclical strategies, be it commodity VRP, be it equity VRP and so on, or FX carry. And that has led the performance down.
At the same time, defensive portfolio, they have done well. In other words, they have given positive or shown negative correlation, but the expected convexity has not been in line with what was measured in the history. So which means that they were not able to compensate the loss given by the pro cyclical.
However, the balance portfolio has done well last year and this year as well. This year, we have seen some mixed performance from commodities, mainly driven by natural gas and oil and so on. But if you look into the equities long shortage strategies, it has done well in the past year as well.
This year, it is up by 4% as well. So in terms of, if you look into the performance of individual buckets, they have been in line with what it has been, it's just that it's the pro cyclical one, which has dropped more, but the defensive has not moved much as we would have expected. I see.
But in a way, we haven't seen a crazy market crash either, right? That would have otherwise triggered the defensive portfolio to really start performing, so as you can make progress. Yeah.
No, that's correct. And this is one of the lessons we learned in August last year, where we see this huge mis-specification in the risk, right, of the expectation versus what was realized. We wanted to understand how we should measure risk and this is where our paper on X post versus Exante come into the picture, where we think that the risk is just not based on the long term, but it has to look into certain areas where your portfolio is more prone to such stress.
I remember you published a paper on that in October. Due to that change, we see that some of the episode which happened December last year, April this year, and so on, our portfolio was more resilient what it was in August last year. I see.
So that has changed. Another thing which we also- That's actually good, because it means that you also make changes to the portfolio as you learn more about it, right? So it's not that you just publish something and then let it be static forever, right?
Yeah. Okay. That's good.
And this is one of the important element that we have to continuously monitor what is driving and what is happening. And also, we need to review or overhaul the portfolio as well, because a lot of strategies comes into the picture over that one year and so on. And that's my next goal, I would say.
Let me ask you a spokastic question then, okay? Now that we're coming close to the end of our podcast. What's your favorite color?
Yeah. I would say my favorite color is blue, though you are wearing blue, but I'm not wearing today, but usually wear blue. And there, of course, it is my favorite color.
Why blue? Yeah. I mean, various reasons.
That's a very common answer, blue. Yeah. No, and they are both scientific and some crazy answer, I would say.
So scientifically, yes, it is associated with calm, but still you are awake. So that is a very important aspect of a quant to do, right? One has to be calm and attentive as well to do the job.
Okay. I see. We also associate blue with both the team which I favor in cricket, Indian team and England team.
Okay. Both of them wear blue. Okay.
And then, of course, the sky is blue as well. The sky is blue. Okay.
That was a stochastic answer. Yeah. Mr.
Vivek Anand, thank you very much for this wonderful conversation. We got a lot of insights from you, not just in terms of your background, but also your thought process, how your brain works when it comes to designing signals and then when it comes to portfolio construction, implementation, and of course, the monitoring of that. This has been our inaugural version of the Stochastic Conversations podcast.
Thank you very much for listening. Should you have any questions or you want to know anything more about what we do in the quant research world here at Deutsche Bank, feel free to reach out and we'll come back to you. And with that, I stop here and we'll see you again in the next episode.
Yeah. Thanks, Skyu, and thanks everyone for listening. See you next time.
No representation as to its accuracy or completeness. Opinions, estimates, and projections discussed constitute the current judgment of the speaker at the time of recording. They do not necessarily reflect the opinions of Deutsche Bank and are subject to change without notice.
For further important information, please visit research.db.com.