View Full Transcript

Carl James   00:00:00

Welcome to the second video that we've done between Chris and myself. Today, what we want to talk about is how do you -- how can you succeed in capital markets. Chris and I have been thinking about this quite deeply as we both had quite long careers in capital markets. Chris, can I turn to you and just ask if you can just give a little bit of a potted history of your time in capital markets.

Chris Fenske   00:00:31

My career started back in fixed income research quite a long time ago, focusing on asset-backed securities, mortgage-backed securities, a lot of writing. So there were a lot of writing skills involved. Back then, the ability to program in visual basic for applications, being very Excel were just based standard skills for everybody. After that actually went on to a trading desk. I was trading distressed asset-backed securities, CDS and asset-backed securities. And then following that, actually managing a portfolio of about over $1 billion in distressed asset-backed securities before coming to market, which is now S&P.

Carl James   00:01:12

So I go back a little bit further, Chris, where I started my trading on the floor of the London Stock Exchange, where I literally had a book and a pencil where I would put my positions down once I traded verbally face-to-face with the broker. And then from there, the ride and adoption of electronification, the implementation of things like fixed protocol. And then you've seen the rollout of the globalization, which is predicated on technology. I did want to ask you, Chris, actually, when you said you started as a financial analyst, I think you said, how did you get your information?

Chris Fenske   00:01:58

So coming from mortgage-backed securities or S-backed securities, it's very data intensive. You're not talking about 20, 30, 40, 100 rows of data analysis. You're talking about sometimes millions of rows of loan level data, securitization performance data. So initially, we had a lot of unstructured data. I mean this is before the days of official data scientists. So I did actually work directly with -- which are considered modern data scientists, the prepayment modelers back then were the ones that are actually doing data science to predict prepayment rates for different types of mortgage-backed securities and asset-backed securities. So when I first started, I mean, the #1 priority when they were hiring team members and when I was hiring people was being able to handle a vast amount of data. This is before the days of Python. So Excel was the #1 tool, being basic understanding of SQL, structured acquire language was just mandatory for anybody that started on the team. And then some of the actual stronger skill sets that we look for, things like visual applications, being able to program Excel or Microsoft Access, creating very sophisticated macros. A lot of these macros actually drove some of this trading strategy and research that we published. That was quite a long time ago. I mean things have changed significantly since then. Now you're looking more for people with that Python type skill set, more skill sets in R and fast forwarding today, being able to actually write very good accurate prompts for different type of large language models.

Carl James   00:03:37

Yes, I think it's very interesting. I mean, from my own perspective, a lifetime ago to trade on the floor of the exchange, it was very small and cluby. So you had to have high emotional intelligence in the sense of understanding, getting on with people. There had to be a bit of aggression in the system because occasionally, you had to kind of force your way through a crowd of people to get the price that you needed to. And mental math, that was something that you have to be very quick at mental math. Now what I wanted to do is move ourselves forward to today because things have changed extraordinarily. However, I think there's some basic behavioral elements that are still required. I mean, from my own perspective, being curious or inquisitive. I think that's a basic thing that people need across any facet of the capital markets. Obviously, I come from my background of trading. I think you need to be inquisitive to understand why does a price move or when an event happens, you're curious to find out what do I think will happen. So there's that curious or inquiring nature that you need to have. What are the sort of some of the -- a couple of basic behavioral elements that you think have sort of stayed the same over the years or decades?

Chris Fenske   00:05:07

I mean, even today, I mean -- and one of the things I focused on back then a lot is just trying to fill in data gaps where a very, very small percentage of actual useful valuable data comes in a format that's absolutely perfect for making investment decisions. Part of my role in research, even on the trading desk in addition to managing money was trying to find disparate data sets that nobody really looks at or nobody was able to connect to financial market activity that I was able to connect before them and create some sort of investment insight, whether it's some sort of technical information, who's trading what, some sort of fundamental information, which with mortgage-backed securities, we're always looking for new types of bespoke fundamental data, updated credit scoring, different type of valuation models for properties, roll rates, analysis. So a lot of it was investigative type work. And on a trading desk, I mean, believe it, I mean, as you know from your experience, it's 90% sales. You're trying to convince someone to pay a higher price when you're selling. You're trying to convince someone to to charge you a lower price when you're trying to buy and you're basically just presenting your case. You're basically just presenting a bunch of facts that actually lead towards either one of those sides of the case to get to the best possible execution for that trade you're doing.

Carl James   00:06:35

I mean I really like what you said earlier about we need to have an informational edge so that you feel, okay, I'm going to commit to this. And so again, that comes back to that curiosity, that drive for digging away at different elements and trying to find that informational edge. I do also like what you said about the idea of sales. I'd also wrap it up as well as I said earlier about high emotional intelligence, which you could frame as being able to read people. How how much do they want to buy whatever is the position you've got or vice versa. And there's that bluff and double bluff that you're trying to put in place. I always thought that was kind of interesting. And then if I think about in today's world, I think that element has been reduced. The utilization now of algorithms or even auto execution logic engines, which we've seen in play, it's a machine talking to a machine. And it's the numbers that actually are the critical and important element rather than say, you and I, if we traded together, I'm trying to figure out, does he really want to sell this or am I going to -- how much am I going to have to push him and et cetera. So I do -- I think that's a very nice observation. Regarding the use of data, I thought it was interesting, you said maybe millions of elements within the data sets that you had. That took a huge amount of time. I assume it took a lot of patience as well. We roll forward, we know that the power of computing has gone up exponentially. The prices dropped through the floor. So I sort of think there's a bias now you don't need that patience anymore because the computer can crunch through. And I think it's about finding those nuggets within all of this data. Does that create a different skill set because you've got to identify going back to that information edge, you've got to be able to identify those pockets of information that are useful.

Chris Fenske   00:09:01

I think it -- I mean, it does take a human or a skilled analyst to point the computer or the analytics in the direction of a data set that makes sense. It's not something a computer could just do by itself. And you get these type of insight from doing a lot of reading, listening to webinars, attending conferences, talking to clients, talking to market practitioners. And you take these -- all these bits and pieces and nuggets and it sort of puts you in the direction of what type of data sets could actually make sense. In addition to every single day, there's a new vendor producing a new data set to try to sell to some sort of financial market that's going to add some sort of alpha to the portfolio or improve their risk management. So sifting through all of this, trying to determine if this data overlaps, it's something that's already widely used. So it amplifies the power of the existing data set. In addition, I mean it's something that's going to have a quick decay where if I start using this data and everybody starts using this data, the data's usefulness or alpha is going to just complete the decay because it's crowded and people of the system. That's why even -- I think even investment banking analysts and a lot of the people within the financial markets capital markets, they're focused on making hires that also have a lot of skills handling this mass amount of unstructured data or data that needs to be cleansed or filtered. You need pypeline for that. You can't do that in Excel. Excel has its limitations on what you could do in addition to basic programming skills, too.

Carl James   00:10:40

It may surprise you, Chris, but I'd like to be a bit controversial here. So I read a white paper by a professor at Warton, and he's called Ethanullk. And he did -- he wrote a white paper about he did a test with a company with 776 employees, and it was the use of AI, a group of people trying to get to a solution, no AI, a single person with AI and a group of people with AI. So the 3 groups. The team without AI, they came with a solution with marked at a certain level. The single person with AI really outshone the team without AI and the team with AI was just above the single person. And one of the observations he made was that with the use of AI, it allows you to bridge a knowledge gap because you're able to say, please tell me a little bit more about whatever it happens to be. AI can answer that for you and bring you back and then you say, actually, I'm not sure I understand that concept. please tell me some more. And I just wonder, as we look forward into the future, is there something there that -- when you said the conferences you attend to, the colleagues you speak to, the number of times that you do look for this information, it creates this layering of information, which is so valuable, I think, the experience that you build in. Is there a part that AI will diminish actually all those years of experience? I wonder what your thoughts are of that.

Chris Fenske   00:12:28

It definitely has an efficient way of speeding up the learning curve for anything, especially a topic that you're not particularly familiar with. It definitely can accelerate things. I mean the hallucinations continue to be an issue. It's also only as effective as the questions you ask it. I mentioned before, prompt engineering is the Excel of 25, 30 years ago. So if you're very skilled with prime engineering, and it's not a straightforward skill. It's something that's learned. It's got to be practiced. If you're very good at it, you can get the right answer quicker with less errors, less hallucinations and be more effective with the work you're doing in your workflows and the analysis. So that seems to be like just another skill set. But keep in mind, too, these skill sets aren't each one is replacing the next. So back in the day, Excel was the hot skill set. People were good at Excel, Excel much faster in financial markets, financial services. You still need a very strong Excel foundation. You still need to be able to definitely handle data. So Python is still relevant, statistical software like our also very relevant. So these are all compounded and you're just adding prompt engineering onto that list of skills that you need today to prosper in the financial markets.

Carl James   00:13:57

So I think from my perspective, and again, please chip in and see where I'm going wrong here. The foundational element, behavioral -- the foundational behavior, be curious, have that inquisitive nature, wanting to look forward, and it's about being committed to the role that you have. I mean it sounds a bit of a platitude, but I think it's worth saying that -- then you get the next layer, which for me is about understanding what the technology that you're working with, how it can be utilized and how it can help you get to that informational edge that you're looking for. So I'm just trying to think carefully when we say to people, what do you need to succeed, there's those foundational steps. On top of that, it's the utilization of the technology that's coming. And I think you said it very well, Excel, all those years ago, people used it and if you became master user, you did better. And I think you're saying now that with Python, you should really know how to code in Python. Is there something -- if you look to yourself 3 to 5 years in the future, what should someone be thinking about to start learning a skill maybe on a 3- to 5-year future view?

Chris Fenske   00:15:25

I think -- and this is something I practiced in the past. Every year, the new year, I would try to take on a new type of skill set. I think like 10 years ago, I was trying to learn just basic machine learning function, how it works, the shortcomings of each different type of analysis. And then the following year, try to learn a computer language, something like Python. I think you should, throughout the course of your career, make a point of learning something completely new out of your comfort zone that is somewhat relevant to what you're working on at work on your career path. It's a formula that actually works very well because it keeps you in a classroom, keeps you reading. keeps you practicing something new and something that you could add to your current job. I mean, as far as 5 years from now, it's hard to say what that scale is going to be. 5 years ago, prompt engineering would not be the first thing anybody said or very few people would have said something like that would be such an important skill set in 2025. But yes, it's hard to predict. I mean things are moving very fast now. And the AI itself accelerates the speed at which direction things are moving.

Carl James   00:16:38

I mean to wrap up this video, Casta, I think that's such great advice. keep learning, keep yourself current. And I have to say I didn't know that about you, Chris, but I do like what you say every year, you take on this new challenge. I think that's great advice to give to people. I've seen too many people who keep stuck in their ways and this is how I've done it for 10 years, and I'll carry on doing it, which that is a strategy. I think a more sort of likely winning strategy is exactly as you do, which is taking on new skills and being very proactive about that and making it something that you build in. I want to say thank you so much for sharing some of your insights. I think they've been really useful. I think many people will take that away and I hope they take on a new skill every year. Thank you, Chris.

Chris Fenske   00:17:37

Thank you. Always a pleasure to chat. I'll do it again soon.