Searching for Signals: BlackRock’s Raffaele Savi on the Future of Systematic Investing

0:05Welcome back to another episode of Goldman Sachs Exchange's Great Investors. I'm Raj Mahajan, a partner at the firm, where I'm responsible for helping to build out the firm's capabilities to serve the systematic investing community. Today, I have the great pleasure of sitting down with one of the stars of our industry, Raffaele Savi, the global head of BlackRock Systematic, which is celebrating its 40th year in the business. Raffaele leads BlackRock's quantitative investing teams across fixed income, equity, and factor investing, and oversees more than$317 billion in assets under management. He also sits on the firm's global executive committee.

0:40We're going to dive into his impact on quantitative investing and explore his perspectives on the growing role of AI, and quite timely, get his perspectives on the current tricky moment for the quant industry, given some of the drawdowns that we've experienced.

0:59Raffaele, it's a pleasure to welcome you to Great Investors and congratulations on the 40th anniversary. Thank you. I might clarify, I have been with the team for 20 years, so half of this 40th, but the team itself has been serving clients for 40 years and thank you for having me here. Let's just get right to it. Let's start early in your career. You started your career in Italy at Capitalia. I did. And you eventually rose to become the CEO of Capitalia Investment Management. Take us back to what drew you to finance in the first place. and how did those early experiences in Europe shape your investing philosophy?

1:33Yes, it was a little bit of a mix of destiny and intent. I'm an electrical engineer by training and actually I was interested, I was doing my dissertation in remote sensing and then a set of circumstances, I ended up taking a class on derivatives and this was sort of the early 90s and I fell in love with finance. And so I decided to do my dissertation on the math of derivatives. There was some coding involved. And back then, that was enough to sort of being considered a quant. Today is much harder for young people entering the field now, I feel. And then I ended up working in the sort of nascent investment management firm in Italy.

2:19And so we were all young and it was another element of luck. They say that you want to be starting your career in a new field sometimes to have the best experience. And that's what happened to me. And then in 2006, you joined Barclays Global Investors, or BGI, which merged with BlackRock in 2009. What motivated you to move to BGI? And then later, you stayed at BlackRock post the merger. Tell us more about your thought process in both of those inflection points. Yes. So I left Italy when I was 35, which is a little late. I like to say that sort of I do everything 10 years after the best practice.

2:58A lot of people leave Italy to study, and then they come back after 10 years in the industry to enjoy the quality of life. And I did a little bit the opposite. I worked the first 10 years of my career in Italy, and then I moved. It was a personal set of circumstances. I met this woman. She's my wife. She's American. I was in love, and she moved to London. And so I started cold calling a number of companies and not even in my wildest dream. uh you call call bgi i call call headhunters i call call companies and then and then i got this offer from bgi and i couldn't believe it because that's you know for people that are in my field grinald and khan is the book that that we all studied on and so the idea that i could join uh that profound impact on our firm our industry and work with richard and ron that sort of you know ron has been a mentor ever since he's still the head of research of blackrock systematic It's like not even in my wildest, wildest dreams.

3:54And when BlackRock acquired BGI in 2009, in the middle of a lot of change, there was this incredible energy at the firm that is still the same, if not more intense these days. And I think what's special about BlackRock is a firm that constantly reinvents itself, constantly brings together different ways of looking at investing, active and passive, public and private, fundamental and systematic. And I think if you do investment research and you're trying to figure out how markets work, there's no better place to be. What were some of their early innovations at BGI that you brought to the industry or that BGI pioneered?

4:36You know, it's interesting. I joined January 2nd, 2006. So at late innings of the first big sort of growth decade for quants, and if you think about some of these strategies, our strategy, our oldest strategies you were mentioning with the four years was launched in August 1985. So there's a big celebration coming. But these strategies, in fairness, didn't really have a lot of tremendous commercial growth for 15 years. And then when the 90s were the years of sort of highly concentrated portfolios, the dot-com boom, definitely not quant strategy, traditional quant strategy material. But then these strategies that were based on race control and sort of diversifying insights did really well in the dot-com bust and in the years after.

5:28And so they enjoyed tremendous growth up until 2006, 2007. I think a lot of the first version of quantity investing at BGI is what today you would call factor investing. And so it definitely has a place in investors' portfolio, definitely has a place in long-term allocations, but sometimes it doesn't have the dynamism to navigate what could be crowding periods like 2007 or a crisis like the GFC. And one thing that we've been focusing on with the team was, okay, how can we make these strategies less dependent

6:03Raffaele Savi:on long-term risk premium, more dynamic, ready for at least to try and take on any environment that one finds himself in. And then that was, if you want, the defense part. The offense part was the explosion of big data, machine learning, predictive analytics, and the excitement in the team to sort of move in that direction in 2009, 2010 was hard to sort of contain. And so that's what happened. In that second wave, when post the acquisition by BlackRock, how would you characterize that? Would you say there are refinements to the first set of factor strategies, or was there a new class of quantitative strategies that you guys built?

6:46I would say more of the latter. It's interesting. In some sense, sometimes when you use a new approach, whether it's these days is large language models, or in the old days, it was when big data appeared on the horizon. The first way you convince yourself that you're not delusional is to actually try to replicate prior generation strategy and improve them a little bit. And it's only after you gain that conviction that you start really using technology in a 3S format and you realize that you can do so much more than build a better value signal or a better quality signal or a better momentum signal, those effects are prevalent.

7:34We call them common factor for a reason. They're common in markets. They're common over time periods. They're common in investors' portfolio. But some of these newer technologies have a broader aperture. And it took us a while to sort of fully understand that and use it. But when I look back now, It's a very different type of thinking and behaviors for the strategies we run. Well, you cracked the door open on large language models and AI, so I'm going to now drive a truck through it and ask you to expand on how is AI impacting your investment approach? There's$320 billion of assets under management.

8:12How would you envision over the next few years AI playing a role in either enhancing that strategy or disrupting parts of it? Right. Well, I think, I mean, if you hang out with a lot of the people I hang out with, and there's a fair amount of old-time practitioners that are embarrassed by the hype around AI. And so this would be the graph people in your team that will tell you, oh, this is statistics, or this is optimization. Bunch of regressions. Right. And that's true to it, right? The same time, what I think is very interesting are maybe three concepts. So the first one is this concept of scale.

8:53There's been this, there's a great paper that's called the Bitter Lesson that basically tells you, hey, you know, you can spend so much time trying to understand everything about a particular problem. Maybe it's linguistic, maybe it's predicting sort of company's fundamentals. And then you just realize that, you know, if you double the amount of data that you're training your models on, or you multiply by 10 or 100, you can beat any sort of smart sort of adjustment to your model. Just more data, more compute leads to better outcome. That's something that I think has to be kept in the background.

9:28I think our industry relative to many other industries still does not understand this concept in full. And I think this concept will play out over the next sort of decade. And that's, to me, the first thing about AI is it's bringing scale as a big driver of success in the investment industry. Okay. Scale? Scale. The second part that I think is also interesting is I'm a big fan of generative AI and large language models because I find that they are universal. And if you're thinking about deep learning, in its previous pre-GPT moment, it was by experts for experts. Even if you have a business person that wanted to apply deep learning in their teams, it was difficult for executives to really feel it.

10:15You would ask someone and they would do something complicated and maybe - Use some jargon like neural networks. Maybe the results were better, but the beauty of this language model is that they are interactive and sort of they speak our language. You can, I can prompt them, you can prompt them. You can get a visceral sense for their capabilities. And that I think has broadened the appeal and the applicability. And so you can use them in way more parts of your investment process. It isn't just about building a signal, but also you can have a lot more people, a lot of more smart people with great domain expertise.

10:56They didn't know how to put it in numbers, but they can do it with an LLM, right? So where I think this is going is that we've seen systematic succeed, you know, the more liquid the asset class and the shorter the time horizon. I think what AI in this generation in the future architectures will do, will sort of open the door for a systematic approach to investing in longer time horizon and in less liquid asset classes. So I think we'll see systematic in private. We're already doing a lot of work systematic in credit. We're using systematic to sort of translate from macro insights to micro insights.

11:33And that's a big trend for me. And the third one, I like to think about it as safety engineering. If you're thinking about cars, automotive industries or aeronautics, the cars today aren't any faster than 20 years ago. Airplanes aren't any faster than 20 years ago, but they're much safer. And so here you have two industries that used all the technological innovation in the last two decades to create more safety. And I think sometimes, you know, as soon as we find a new data set or as soon as we find a better algorithm, our first instinct is, you know, can we generate more alpha, right? But there's also can we build portfolios that are more resilient to shocks?

12:18Can we build portfolios that get closer to client desired outcomes, no matter what markets throw it at? So I don't think AI will get us any closer to the crystal ball, but I think it will create layers of safety so that, you know, events will be what they are. They're ultimately unpredictable because the world is beautiful, because it's unpredictable. But you give me some money today, you want more money one year from now, three years from now, five years from now. How can we solve that problem better? That's the other application of AI. That's fascinating. So what I heard is scale, safety, and then that second element is moving into less liquid asset classes.

13:00Can you just maybe elaborate a little bit more on how systematic approaches to investing are going to be applied to private markets and private investing? Yes. I think it's one of the most exciting areas. And I think it's an area that over the next 10 years is going to become a big market. And, you know, it's interesting because in the earlier generation of quantitative models, the basic ingredients used for these models were prices, volumes, returns, fundamentals, and analyst estimates. And none of them is readily available for a private company, right? But you fast forward 15 years and the state of the art models today are using a lot of different sources like product reviews and search trends and Instagram posts and interviews and various forms of sort of human capital network.

13:51And all this data is actually readily available for every company. Actually, we realized that we were filtering out a lot of private companies out of these new data sources that we sort of acquired. And so I think that as the models become more data rich, you can build forecasts for any companies, including private companies, right? Then I think the second aspect, again, is there's a lot of interest in portfolio construction and risk modeling challenges that I think are very exciting about what you can do in privates. And if you think about the arc of development of these strategies, again, like quantitative strategies started in equities, shortened for institutions, right?

14:35So it was a sophisticated strategy in liquid instruments for institutions. And I think where systematic is going is going equities and fixed income and is going for longer duration assets. And AI is the unlock, is the bridge to this when you don't have a time series and volume but you have more language and text. Text and language, images, and AI is a great way to extract information out of these sources that are rational or not tabular. That is what we have in spades for public companies, but we don't have it for private. And are you seeing in the midst of all of that data that you're accessing and then applying these AI techniques, are you seeing a signal?

15:13Yes. We have products that we developed with colleagues in BlackRock that are in the private equity team that have a traditional fundamental investment approach to private equities. And we've been able to partner and build something new. I think we'll see much more of that in the market in the next 10 years. And there's a lot of assets there. So you could see this have a big impact on your AUM if you get this right. Yes. And I think as we are, as more and more investors and investors have private assets in their portfolios, it will become a more interesting and a more sort of rewarding challenge to figure out how to build an optimal public-private portfolio.

15:54You know, what kind of risk models we can build that tell us how the whole portfolio will behave relative to a certain shock? What's an optimal deployment strategy? And, you know, how do we optimize sort of liquidity profiles while investing across all these asset classes. And these are problems that quants are very well positioned to attack. And so I think you have, it's more relevant in clients' portfolios. And from a data and an AI perspective, you can build models that are as information rich as the ones you have on public. It's just that the data is of a different nature. That's very interesting.

16:33We're recording this on July 29th, 2025. Then I'd be remiss if I didn't at least take a moment to talk about the past seven weeks. It's been a relatively rough stretch for systematic investors after a bumper start to the year. And we've heard lots of theories from retail surge, broad-based rally factor, pressures with high vol, high beta names, short interest. you oversee$320 billion of these assets. You probably have the best panoramic lens into what's really happening right now. Tell us a little bit about what may be causing this drawdown, which by our numbers could be as high as say 40 % from peak to trough.

17:21Give us a sense of what you're seeing and enlighten us on what may be causing all of this. Raj, I thought you had the better seat. And that's what we talk about these things. So I'm trying to think what's a good way to share the way we're thinking about it now. And I think one way to do so is to say, what is the primary feature of this drawdown, right? So if we were sitting back here a year from now, if you invite me again, and we look back now, you know, however painful, however long it's resolved, how would we describe it? I would say that the key feature is this sort of high vault names that have been ripping.

18:01People have been losing money mostly on their short positions or underweight. And it's been a bit of a collection of some fundamental beats, some meme stocks, some technological breakthrough, some deregulation hopes. But sort of there's been a high number of seemingly unconnected high-volved names that have done really well. And I think for a variety of reasons, the length of time in which that happened created correlation between slower players and faster players that have been trying to get in on that trade with reversal strategy. And so by mid-July, by July 10, and sort of the 10 days after the second decade of July, a lot of people were on the wrong side of these high-vol names ripping for whatever reason.

18:51And usually those events tend to lead to risk management. Sometimes it's mechanical. You might have some leverage limit. You might have some stop losses. Sometimes it's an overlay where people say, whoa, we don't want to lose that much of the year-to-date alpha. And so I think what you've seen in the week before this one was more easily described as sort of selling down positions, closing shorts. That usually happens towards the end of these moves. It's very hard to say is it in six, is it in seven, but usually you have a shock, positions adjust, and then sort of markets find bad and equilibrium.

19:32Could there be another shock along the same dimension that happens? Yes, but I think if you just go by history, it does seem that we are towards the end of this move. Yesterday, today, Friday as well were much better days. I hear people have recovered a third to half of that drawdown, and you think you've in everything, and then something new happens. There will be another one. I mean, you made a really interesting point, which is during this period, if you're in a drawdown and the markets rallied 10%, you have to mechanically degross just to hit a leverage, to have constant leverage. And so you could be just seeing the duration of this be the result of a number of what might be in absolute terms, small degrossings, but in aggregate terms, quite consequential.

20:20Right. And I think when you look at sort of some of the technical signals that we sort of developed to try and follow this, again, they weren't particularly allowed up until July 15. And then they start picking up that kind of behaviors that, again, tends to be signaling that we enter the last legs of this move. To your point also, I think what's interesting is, and I'm an optimist, as you probably figured out by now, it's been interesting how many shocks we've seen over the last five years and nothing happened. And so I was actually thinking, have we all gotten so much better in risk management?

21:04You know, COVID, the inflation shock, you had that big one day down 14 % move in Nikkei last August. If you have any of these shocks happen in the 90s or the 2000s, we would have read about hedge fund going under and massive disruption and very little happened, right? But I think this move of the last five to six weeks is a good reminder that there's never a good time to be complacent in our industry. And you got to be worried every day. You got to work every day to make your strategies better to sort of fulfill the promise you made to clients. Well, you mentioned risk management. Just tell us a little bit about the risk management approach within BlackRock Systematic.

21:45Are there certain lessons learned that you hold everyone to account to and make sure that they're being implemented and monitored daily? Yes. I think risk management is a combination of two things. One is the systems that you put in place. So one is very - Baked into the code. Yes, it's very numerical. And how much do you invest? How sophisticated are they? How multifaceted are they? How much environmental awareness you're working to build in this system? How intelligent these layers are? But then there's also a philosophy. So how do you feel about trading off, for example, alpha in a good years versus drawdown protection?

22:29Those are design choices. And I think that the better these systems become, the more sophisticated these models become, the more the philosophical part becomes important. You're investing in making your tools and your infrastructure better and better so you can pick up things, you can design products that have certain behaviors with a higher probability. But what behaviors are you after, right? And there's two quotes. I like these little quotes. One is Richard Grinault that I quoted earlier one day told me, the best way not to gain 20 pounds is not to gain five pounds. And so if you're worried about your fund sort of having a big drawdown, then trying to avoid a small drawdown is a good way to start.

23:10And that to me means dynamic, action-oriented. I think sometimes there's some sort of more dogmatic approach in the quantitative industries that we don't subscribe. We, you know, our view is like, hey, what happens, happens. Our job is to deliver alpha to our clients in any market circumstances. So dynamism is important. And then, you know, recently I was talking with a pilot and he said something along the line. So it's better to be on the ground wishing to be in the air than to be on the air wishing to be on the ground. And that made me think about sort of safety first. You know, if it doesn't feel right, if you don't have all the facts, you know, there's always another day.

23:52Our clients have a very long time horizon. People are saving for retirement. People are saving for college. You know, if you have a team and an infrastructure and systems that are good at forecasting, are good at delivering outcomes, you want to have a chance of using them tomorrow, in a week, in a month. So never do anything that will prevent you from doing that, right? And so I think that those choices, as technology becomes better and better, what's going to shape risk management sort of processes and philosophy is more what you want to do, what you think is right. You mentioned clients. I know that you spend a lot of time with clients day to day.

24:36Can you tell us a little bit about why that's so important to you in your role? And what are some of those discussions like? What are some of the conversations that you're having with clients? But I think, you know, I know it sounds a little sappy, and my team jokes about the fact that sometimes I'm a little cheesy and sappy. But, you know, this is true. Anyone that is in this industry knows that you get smarter every time you have a client interaction. I never had a client interaction where I didn't learn something, sometimes a lot, sometimes a little. Some are very pleasant. Some are less so, but they own the use case.

25:14Ultimately, we manage money. It's not our money. It's their money. And many times it's their ultimate principle money. And that's one big advantage. They really understand what they're trying to accomplish. The second is they see everybody. And no matter how good of a year you had and how sophisticated your team is, there's a lot of really smart people out there that are doing really well. They're doing cutting-edge work. they develop organizations with great cultures and you know to try and be additive to that win business um you know they you need to raise your game and i think that type of interaction is absolutely necessary and uh yeah i love every second of it that's well said you've worked with a lot of the leading minds in the industry over the years can you tell us about what are some of the common properties that the very successful systematic investors have had?

26:13There's various ways of being successful. You know, when I was younger, I really thought that there was this sort of top cognitive abilities, peak speed, was everything. And, you know, I still think it's very important, especially for certain parts of what we do, of the investment process, and the forecasting process, building systems. I think you have to match that with curiosity and open-mindedness. And so I found some of the best people, the smartest people, I know the most accomplished people I know are bizarrely open-minded on topics they know all about. They've written the books. They've trained generations of sort of professionals.

26:59professionals, and yet they can have a discussion on that topic, listening to a new opinion, inviting new opinions. That to me is remarkable. That's really cool. Yes. It's that open-mindedness combined with the intellect that I think makes miracles. Speaking of getting smarter from spending time with clients, at one of our recent lunches, you gave me the, you were describing the industry as a West Coast and East Coast philosophy, and you weren't talking about rap music or anything like that. But maybe for the listeners, maybe elaborate a little bit on that model for how you've thought the industry has evolved and what are some of the implications of that?

27:44Yes. So this is a little bit along the lines of we shape our destiny, but it also matters where you're coming from. And when you look at quantitative investors, systematic investors, and I'm half jokingly, I say that if you could do a genetic test of everybody in our industry, us or our competitors, the new entrants, the big ones, the small ones, you essentially find two routes. One is trading desk of investment banks. And that's where sort of the first start-up strategies are born from. That's where modern HFT comes from. And it was from the origin long short. It was from the origin numerical.