Episode Transcript
Dr Joseph Simonian : We have to make a usable product. So if the chair is mathematically very sophisticated and complicated, but you sit on it and it breaks and you fall down. Well, it doesn't matter how beautiful and elegant the actual design was. It didn't actually achieve the goal was supposed to achieve.
David Sol : Welcome to FTSE Russell Convenes. I'm David Sol, and in this session, we'll dive into portfolio construction approaches. In an era of systematic strategies. I'm here with Dr Joseph Simonian, head of Equity Portfolio Design and Analytics at CalPERS and the author of many academic papers such in the Journal of Portfolio Management. And I'm with Sebastian Lancetti Head of Equity Index Design at FTSE Russell.
Sebastian Lancetti : Now, before we dig into the details of portfolio construction, let's take a step back. Portfolio construction I'm talking about optimization. A lot of math is very complicated. Why are we doing this? Why make this so complicated?
Dr Joseph Simonian : Well, it doesn't have to be complicated, but I think the reason we do it is because we have to have a formalized process, typically in order to defend the ultimate allocation of weights in our portfolios, which is what portfolio construction is. Ultimately, it's the allocation of weights. It can be separated from security selection. It can be combined with it, but it can be separated from it as well.
The reason we use somewhat formal or mathematically rigorous methods is because we benefit from the clarity and the objectivity that an optimization approach, for example, brings to the process. If we didn't use some kind of a formalized process, we would essentially be artistically deciding the weights on our portfolios. And that's not defensible from a stakeholder standpoint or from an investment theory standpoint, if you will.
Sebastian Lancetti : So we're doing it to get some kind of framework, some kind of structure. And our model is thinking portfolio construction. Is this art or is this science. I know Sebastian, you do a lot of backtesting. Is it art? Is it science? So what's what's your view?
David Sol : It's definitely some science with a lot of art. I think adding to what was just said, very important from an index provider standpoint. Also to consider the fact that everything we do needs to be systematic, needs to be put in rules and portfolio construction techniques can be formulated in a way that are replicable, and so that our clients and the marketplace knows exactly what is happening in our indices, too.
Sebastian Lancetti : So let's dig into some portfolio construction. Well, give me some examples. You want to kick off first, maybe with some.
Dr Joseph Simonian : Sure. I'll list some of the basic ones. Of course many of our viewers know mean variance optimization. Modern portfolio theory. Markowitz is approach, and this essentially involves considering expected returns, volatilities or variances of assets and covariances or correlations. But then you have other techniques. You have frameworks that incorporate multiple objectives. You have approaches that are specifically attuned to, liability driven investing.
You have some approaches like black liniment, which allow the investor to inject their subjective views into the process. So there's been a decades of research, of course, that is continuing today. In this area. A lot of it is very mathematical. But I think from my standpoint, the important thing is to, take the practical applications of any framework, if they're suited to your particular use, if they're not that, it just become the theoretical construct on paper.
So there's a list mile-long of techniques that we can talk about for hours. Those are some of the major ones. But, there are very highly tailored approaches that are coming on all the time that speak to one specific aspect of investing. So there's a whole inventory.
Sebastian Lancetti : So, Joseph, so what you're saying is you have an investment process from kind of asset allocation to estimating returns to the way you implement your trades. And for those different parts of the investment process, you'll use a particular, technique from portfolio construction. Is that how you should see it as a kind of a menu that you pick from?
Dr Joseph Simonian : There can be each part of your investment process may very well incorporate different optimization techniques, but they don't have to. Right. I think what I was referring to is that at the beginning of your investment process, you asked yourself, okay, what are we doing here? What are the most important inputs? Are there correlations that are very important, or is it more of the liability asset relationship that is most important?
Of course, these things aren't always easily separate from each other. Often, the same considerations apply to different contexts. And an LDI or liability driven investor is going to think about correlations as well as an asset only investor. So I don't want to make it as if every optimization framework is so specific. However, there are fine tuning things that you have to do along the way to essentially apply any given framework fruitfully to the the application that you're looking for.
Sebastian Lancetti : And you've written many different papers on portfolio construction. How do you see the evolution? I mean, you were writing, about things 20 years ago, ten years ago. So, so what are the type of things you're writing about now in, in your journals?
Dr Joseph Simonian : On the optimization side, I think what many of us had written about fairly recently, are ways of allocating amongst factors versus assets, factors which we know they they inject a little bit more flexibility in how you can do the allocation of assets. And so understanding, you know, what are the differences between an asset class based optimization or a portfolio construction process versus a factor based one is important.
Now again it's not necessarily a hard dividing line right. You can look at the walls of the lens of assets asset classes but also factors. But I think, you know, a factor approach does open up some more vistas if you will, or possibilities to allocate unique and interesting ways. Again, I've also written about the liability driven investing side quite a bit.
And in terms of trying to develop portfolio construction frameworks that speak to the LDI or liability driven aspect of investing, and that's always important various kinds of pension and insurance applications. And then I have also ridden, you know, a lot on the black Litman model extensions to it. I've presented quite a few extensions to it. And again, the black aluminum model is derived initially from the mean variance framework that many folks are familiar with.
But it overlays the ability to inject the subjective views of the portfolio manager, which allows it to incorporate more real time information, which obviously benefits us because we have seen events like Covid and and other things where you can't rely on a historical data set. So it may behoove you to actually utilize more real time subjective inputs. And there have been models like that that are coming out all the time, again, for very specific applications.
Right? It's like the gift that keeps on giving. You can never get enough optimization. And then of course, on more of the academic side, there are very highly mathematical papers that come out, and our papers are mathematical as well, but they are first and foremost intended for practitioners to be used. I think a lot of the academic papers a practitioner has to really sort of peruse these papers pretty in-depth and tease out the useful aspects of them, because they have a lot of, I'll call it a quote from.
All right, additional elements to them that may not always be applicable. So I think that's the one caveat I would give in in looking at optimization techniques from the academic world, is that there's going to be a lot there. Not all of it is applicable to practical portfolio construction.
Sebastian Lancetti : And Sebastian, you work a lot with clients, asset managers, asset owners on building indices, constructing them. How do you approach portfolio construction? What type of discussions do you have and what's your philosophy about building in robustness in in strategies?
David Sol : Yeah, we have a wide range of approaches to build the exposures. And the portfolio said our clients are looking for on top of everything going back to the point, academic versus practitioner, it's a very important aspect is just a lot of our clients have constraints. They have maybe exclusion list. They might have all our targets on our purely return based targets.
And so we need to incorporate that in the portfolio construction process and do so efficiently. So we want to make sure that the index we produce ultimately reflects the intention of a client, but also is somehow optimal in terms of enforcing the constraints of the client's specific request. So that's another dimension to it. One thing is the performance, the alpha, the behavior, the risk behavior for portfolio, but also the enforcing enforcing the constraints as well.
Sebastian Lancetti : Now it's interesting you have this theoretical performance to the to the modeling. You on the other hand, Joseph, you deal with the practical implementation of really big portfolios. We're talking about thousands of securities. Can you go a bit into the detail? What type of things do you have to think about when you deal with the big scale? In terms of implementation versus the theoretical, approach that, that we can do on that kind of the indexes on.
Dr Joseph Simonian : Yes. Well, I think whenever you're dealing with an optimization framework, first and foremost, you have to understand that at the base, it's a mathematical construct. And it may very well be the case that the individual who has built this framework did not have experience or practical portfolio management implementation in mind. That's not to say anything negative about their background or their intentions, it's just that often a framework is presented saying, look, if we actually changed the math in this way or that way, we get better outcomes using a specific set of historical data.
When you have a larger scale portfolio, what happens often is that the implementation issues often present challenges to you that you have to confront, and you may have to either change or relax or loosen or somehow modify a chosen framework. Right? Because as mathematical constructs, what will happen is that these optimizers optimization frameworks, what they'll want to do is allocate in a specific way that is more or less rigidly loyal to its framework.
But then when the output comes, you may say to yourself, wait a minute, I understand why it's doing that mathematically, but from a practical standpoint, I can't have this kind of allocation or this concentration or what have you. And so you have to think about that. And you don't want to just artistically, you know, smudge the numbers. You can't do that, obviously. So what you have to do is you have to, in a formal, rigorous way, modify the framework or come up with a new framework that actually is much more practically attuned to what you're doing.
Sebastian Lancetti : So you have your framework, you then use some kind of overlay, common sense almost. I would describe it to see whether or not you go to extremes in your, in your, in your output. Is that, is that how we should view to.
Dr Joseph Simonian : I'll give you more specific analogy. I think in finance we often think of ourselves as an especially quants. As scientists. I think that's incorrect. I think we are more similar to engineers or maybe even cabinet makers. We're craftsmen. We have to make a usable product. So if the chair is mathematically very sophisticated and complicated, but you sit on it and it breaks and you fall down, well, it doesn't matter how beautiful and elegant the actual design was, it didn't actually achieve the goal was supposed to achieve.
Wouldn't they say we're building chairs? I know it's not as glamorous as think of ourselves as a scientist in a laboratory. And yes, the chairs may be complicated, right? They may be sophisticated chairs, but ultimately we're a craftsman. And so we have to think about the end product in mind, which is a portfolio that speaks to, you know, stakeholders and anybody else who has a vested interest in the good performance of our portfolio.
That goes for any kind of real investor, whether they're a large institution or a small wealth manager. You have to think about the end product has to be a practical, usable thing. And if you have to deviate again in a in a rigorous and formal way from an established framework or a framework that you liked on the merits on paper, then you have to do that. It's just how it goes, because you have to deliver something that works. Ultimately.
Sebastian Lancetti : I like your analogy of the cabinet. Maker. Sebastian, I know last year we spoke a lot about concentration issues was a big theme in the, in, in the building of portfolio control. How how is concentration, now, at the forefront of of investors. How are you thinking about it?
David Sol : Yes. I think if anything, things got a bit worse compared to last year. So we have still very, very concentrated markets, specifically in the US if we think about the growth segment. And that's a perfect example where you can do a lot with even simple portfolio construction techniques, does need to be an optimizer all the time. But just to make sure that ultimately we build also indices that are more diversified and also less prone may be to what happens to individual names.
So that's still very, very, much a big topic in indexing. So we actually, about to launch a new methodology to do that. And in general, again, going back to what was just said, also very important when we think about optimization, portfolio construction, however, stakeholders and especially from an index standpoint, we are very public in what we do. And we must make sure that, what we do is understandable to all the stakeholders. So one thing is our elegant or mathematically sound solution is but we also need to be tractable, transparent and understandable to a broader audience as well.
Sebastian Lancetti : So it's very much about dialog, with the end users trying to understand what they want to achieve and of course, discuss the assumptions, with them in depth so that they know what they are holding. This has been a fascinating discussion. There's more to portfolio construction, and we'll be keeping an eye on your papers. And of course, the indices that you'll building out of, FTSE Russell, thank you very much for this interview.
Dr Joseph Simonian : Thank you.
