The Investment Banking Job Hasn't Changed Since 1990. AI Is Changing That with Wall Street Prep CEO Matan Feldman
Matan Feldman Paul Barnhurst speaks with Matan Feldman, Founder and CEO of Wall Street Prep. Matan shares insights from his career in investment banking and financial modeling, discussing the future of AI in the industry, its impact on investment banking roles, and how it will reshape financial modeling education. He also reflects on his journey founding Wall Street Prep and the lessons he's learned along the way.
Matan Feldman is the Founder and CEO of Wall Street Prep, where he leads business development, course creation, and training programs for clients such as Morgan Stanley, J.P. Morgan, The Wharton School, and many others. Before founding Wall Street Prep, Matan worked at Chase Manhattan Bank and J.P.Morgan in investment banking and equity research.
Expect to Learn
How AI is transforming financial modeling and investment banking workflows.
The evolving role of financial modelers in an AI-driven world.
How AI accelerates repetitive tasks, enabling analysts to focus on higher-level judgment.
Why foundational financial modeling skills remain crucial despite AI advancements.
Here are a few quotes from the episode:
"A model can either be the most powerful tool or the most frustrating one." – Matan Feldman
"Judgment and interpretation will always remain important." – Matan Feldman
Matan Feldman shared valuable insights into the evolving role of AI in financial modeling and investment banking. He emphasized the growing importance of business judgment and foundational skills as AI tools continue to reshape the industry. Matan's journey from investment banking to founding Wall Street Prep highlights the ongoing need for both technical expertise and strategic thinking in financial modeling.
Follow Matan:
Website: https://www.wallstreetprep.com/
LinkedIn: https://www.linkedin.com/in/matanfeldman/
Follow Financial Modeler's Corner: LinkedIn Page: https://www.linkedin.com/company/financial-modeler-s-corner/
Newsletter - Subscribe on LinkedIn: https://www.linkedin.com/build-relation/newsletter-follow?entityUrn=7079020077076905984Sign up for the Advanced Financial Modeler Accreditation today and receive 15% off by using the special show code ‘Podcast”Visit https://bit.ly/497oAqW and use the code “Podcast” to save 15% when you register.
In today’s episode:
[00:00] – Introduction
[02:27] – Matan’s Modeling Nightmare
[05:07] – The Journey of Wall Street Prep
[08:21] – AI in Investment Banking
[14:06] – The Future of Modeling and Automation
[22:16] – Opportunities and Concerns in AI
[29:48] – AI’s Role in Teaching Financial Modeling
[36:02] – Rapid Fire Questions
Full Show Transcript:
Host: Paul Barnhurst (00:39):
Financial Modeler's Corner is the world's premier modeling podcast. It is brought to you by the Financial Modeling Institute, the world's leading financial modeling accreditation organisation. Welcome to Financial Modeler’s Corner. I'm your host, Paul Barnhurst. In this podcast, we talk all about the art and science of financial modeling with distinguished financial modelers from around the globe. The Financial Modelers Corner podcast is brought to you by the Financial Modeling Institute. FMI offers the most respected accreditations in financial modeling. That's why I completed the advanced financial modeling course this week. I'm thrilled to welcome Matan Feldman to the show. Welcome to the show. Thanks,
Guest: Matan Feldman (01:26):
Paul. Nice to be here.
Host: Paul Barnhurst (01:27):
Yeah, really excited to have you. So a little bit about Matan's background, and then we'll jump into some questions here. He founded Wall Street Prep, and he's their CEO. His responsibilities include business development, development of courses, and overseeing training programs, but he oversees training programs for clients, including Morgan Stanley, Credit Suisse, FBR Capital Markets, JP Morgan, Wharton Business School, London Business School, Kellogg, Booth, Stern, Cornell, and others. Prior to founding Wall Street Prep, he served in several capacities on Wall Street, first as an analyst in Chase Manhattan Bank's mergers and acquisitions group, and subsequently as an associate with JP Morgan's equity research group covering food and drug retail equities. So Matan, again, welcome to the show. We start every episode with this story. Tell me that horror story you worked in modeling. You have to have one. Unfortunately,
Guest: Matan Feldman (02:27):
My modeling nightmare is sort of a recurring dream. So it happened over and over again in my time at Chase as an investment banker. What it looks like, it's almost always the same thing. It's 5:00 PM This is, again, this was back in 2000 and that plays a factor in this recurring nightmare. It's 5:00 PM. I've been working on a model for five hours. I forget to hit save, the model's gone. I end up going till 10:00 PM to do it again. I probably forgot to hit save. I'm just going through that infinite death loop. Back then we did not have auto recovery or anything like that, and so you are out of luck. And so that was a very common theme in that first year where just hours and hours of your life are just gone. So we'll leave it there.
Host: Paul Barnhurst (03:08):
You break that unfortunate habit or that challenge because we've all been there. I've had that one as well. Or at least any of us that are older have been there for sure.
Guest: Matan Feldman (03:16):
Right, exactly. You break the habit by learning and hit control S just repeatedly throughout the day. Just a tick. Just control S, control S over and over again. But it actually goes to, I think a bigger point which helped me ultimately in my career, which is just how quickly the time flies and how locked in you get into building a model. I think to some extent that's either something that really turns you on or turns you off, that just the time flies and you forget to hit control S or you forget to do the things you're supposed to do because you're just so locked in. And I think for me, I actually realised early on that I enjoyed living inside these models and the time went by much faster than maybe I would've liked to go. So
Host: Paul Barnhurst (03:54):
Good lessons there. And it is interesting when you get into flow, it's amazing how fast time can go, whatever that is for you, but you get to that state where you're just enjoying something, you're just working, things are going well, and yeah, it's easy to forget, okay, I need to go eat, I need to hit control, save, whatever it might be.
Guest: Matan Feldman (04:11):
Correct.
Host: Paul Barnhurst (04:13):
I think everyone's dealt with that one. I think back to college more than once. How did you end up getting into training? I mean, you obviously built Wall Street Prep, you've built quite a business, so what's the story of how that happened?
Guest: Matan Feldman (04:26):
Very much a story of desperation needing to get out of investment banking and trying to find a reasonable excuse to do it. So the reality for me was that I had spent about four years Chase had been acquired or merged with JP Morgan. I had spent a couple of years in investment banking doing m and a and then in equity research covering food and drug retailers and sort of the underlying technical work I actually found really, really interesting, which is ultimately why I started Wall Street Prep, but definitely the lifestyle and I think where I was in my career, it just felt like it was time to try something out on my own. So it wasn't like a eureka moment where all of a sudden I sort of said, okay, I've got to do Wall Street prep. It was more like, what do I do to get out of here?
(05:07):
So maybe I will open a restaurant. At the time I was covering the food and drug retail side, I was covering supermarkets, drug retailers, whole Foods and Wild Oats back then. These were natural food grocers, super hot and interesting, and I was like, oh, maybe I'll start one in Manhattan. Just all kinds of ideas. I think scuba diving and I mean there was just a whole bunch of crazy ideas, and this one was probably the least crazy at the time. I think I'd asked myself, what skill do I have that I can leverage and didn't actually know how to scuba dive, didn't really know much about running a grocery store, but better or for worse, financial modeling is a skill that I did think I had that could be parable into some sort of competitive advantage. And that is ultimately why I ended up starting Wall Street Prep.
(05:51):
I didn't know at the time if I'd be any good at teaching it, but I did really enjoy the technical work that I was doing on the job and thought this would be an interesting avenue to try simply because at the time, again, this would've been 2004, so 22 years ago where really access to this kind of training didn't exist outside of the bank, and my vision for it was very much making it available to folks outside of that corporate veil. So it was a novel idea at the time that now anyone going online could get access to some of these lessons. But back then that wasn't the case. No,
Host: Paul Barnhurst (06:31):
I mean training is much more available today for better and worse I think, but we'll get into that a little bit later. I love honesty. One of my biggest drawbacks, I consider going into consulting, I always thought investment banking would be interesting, but I always thought about work-life balance. It's funny, I feel like as I've got older, it's almost in some ways got worse. I said I start my business late and when you start a business, it's a lot of hours.
Guest: Matan Feldman (06:56):
A hundred percent. Yeah. It's interesting. And even within that, so you think, okay, the work-life balance at the bank, I was working 80, 90, sometimes even more hours per week. And as that scaled down to equity research that scaled down to something like 60 or 70 hours a week, the catalyst, the thing that you're like, okay, I need to do something different, is in some respects about that work-life balance. Although it wasn't actually front of mind for me. I was like, I sort of expect it to work hard starting the business. But there was something about the investment banking work-life balance where because you're in a very regimented and sort of constant way for somebody else, for me that was like, Hey, I think I want to do something on my own. So I never actually had a problem with the work. In fact, to your point, you end up starting a company on your own, you end up working just as much and if not more. And in some respects, wall Street prep's been around for 22 years now. I'm sort of Benjamin buttoning my time here where I'm working longer as the years go by. But it's an enjoyable process because it's something that you're doing on your own. At least for me, it's something that I'm doing in my own free will.
Host: Paul Barnhurst (08:21):
I understand, like you said, the own free will. I mean, I've really enjoyed doing my own business. It's just kind of funny because a lot of things I avoided is I didn't want the hours found me either way.
Guest: Matan Feldman (08:30):
Exactly, exactly, exactly.
Host: Paul Barnhurst (08:33):
I want to focus most of the rest of our time on two areas. Where I want to start is what does AI mean for the investment industry and modeling? And then I'd love to get a few of your thoughts on the education space. Obviously AI is disrupting everything, but I think those are two areas you're thinking about pretty regularly. So let's start with what are you seeing around AI with investment banks?
Guest: Matan Feldman (08:57):
Yeah, so within investment banks, and this question has a different answer today than it did six months ago and we'll have a different answer in six months. So just appreciating how quickly things are evolving. But right now in investment banks, what's definitely starting to happen is the deployment of tools, whether they're copilot or chat GPT or others to help junior analysts with repetitive workflows. So anything from document summarization to helping on modeling tasks. And again, these are not full end-to-end workflow changes. That's not happening partly because just the technology is not there to make it possible. But certainly more and more banks have rolled out these tools and people and particularly in the junior ranks, are using them. What's not yet happening is, for example, senior bankers managing directors are not really using these tools in any meaningful way beyond the basic prompting for copy improvements and emails and things like that.
(10:18):
None of that sort of real complex modeling deck creation end to on the senior side, more real end-to-end workflows around communications are being rolled out. We are still in the early days of that. And so that's the state of affairs. And then I would say in parallel to that, what's happening is decision makers are within the banks, whether it's chief investment officers learning and development professionals, heads of divisions tasked with deploying these tools. It's almost like there's a Schrodinger's cat situation where at the same time nothing's really happening on the junior ranks and then the senior ranks beyond basic productivity enhancements.
(11:13):
While simultaneously there are some real existential questions around how this industry changes when AI does everything. And so everything from the entire pyramid where analysts are on the bottom and you have sort of small associates, smaller amounts of associates, then VPs and then managing directors at the top that's being questioned while at the same time not even one analyst could be removed at this point and because of these tools. So that's really the weird state of where everything's at. So it's super in flux, but I will say lots, it is the topic of conversation at banks by decision makers with a lot of fomo. What else are people doing? We're sort of sitting at the intersection of that and hearing how all of it is coming together at banks, and banks are coming to these rollouts at different speeds with different approaches, but ultimately there's just a lot of consternation around, okay, where does this go and where does it lead to?
Host: Paul Barnhurst (12:33):
First I have to commend you for shorting yours, cat reference. I haven't had one of those yet in almost 400 episodes, so well done.
Guest: Matan Feldman (12:40):
Well, thanks. I'm not sure how well it works in this particular context, but it does seem like two things are happening in parallel that are incompatible with one another that eventually, I'm sure they'll find compatibility. But it does create a very interesting dynamic psychologically for everybody that's dealing with this.
Host: Paul Barnhurst (13:02):
A hundred percent agree. I mean, right, there's people who want to adopt it, they want to learn it. Then there's the fear. Well, what if I get really good now that I did all my job. Am I going to lose it? There's also the, okay, are they going to be hiring as many people? Is this a good career to go into? I think a lot of people in college are thinking, the parents are thinking as they go in, they go into that career, are they going to be able to have a job in four years? And then like you mentioned, there's the reality of right now it's not saving us a job, but we all see huge potential here for disruption, how much we see, how far, we all have different opinions and everybody's trying to figure out how to deploy that and what that means. So it's an incredible amount of change. Like you said, fear of missing out. I mean, all you have to go onto is LinkedIn for five minutes and you see the kind of fear of missing out and this promise of the whole world's going to change. And it's like, okay, step back and let's take a look at where we're at versus what we hear people say, which often are incongruent. I'm sure you see that a lot.
Guest: Matan Feldman (14:00):
And actually I've never seen it quite like this. So never in my career have I seen and actually taken a step back. In reality, investment banking and our lens, I should take a step back. It's sort of our lens at this point. My background is investment banking and equity research, but really Wall Street prep at this point, our lens is a whole bunch of end markets. It's banking, it's asset management, private equity, hedge fund work within the investment management umbrella, fp and a, just to a large extent. And so our lens has broadened, but really within, I'll just kind of keep it focused on the investment banking lens that work, the investment banking skillset has not changed all that much since Excel came about and up until now where the conversations are really about a massive change in what the investment banking skillset looks like from the analyst all the way up to the management managing director.
(15:09):
That's a completely new experience for people really across now several generations. The investment banker in 1990 is doing almost the same thing that the investment banker in 2023 was doing. And again, caveats there, notwithstanding the skillset has not changed all that much. And now all of a sudden you have this profound change that nobody, that hasn't even happened yet and everyone expects it to happen. So it is the term I used the other day. It's crazy making for a lot of people that are trying to build organisations and restructure their organisations and support this change without understanding where it lands. It's actually quite somewhere on the spectrum of really challenging to impossible tasks to do right now,
Host: Paul Barnhurst (16:09):
Right? I mean, you're trying to predict where we're going in an era where we've been changing and moving almost faster than any time in history as far as what can influence a job. I mean, you think what the models were doing on helping build stuff in Excel six months ago, three months ago with four seven releases, they're even better and they continue to improve. But like I said, okay, we haven't displaced jobs yet. They're not able to do everything. And so it's like, okay, what's the bet we make for 12 months, 18 months, three years, five years? That makes a huge difference on how successful you are as a firm.
Guest: Matan Feldman (16:45):
And you've got to remember that the larger, the firm that kind of lead time, 12 months, 18 months, is also constrained by the fact that these are massive organisations with regulatory constraints. And they can't just roll out, oh, we were using chat GPT, now four point sevens out for clot, so we're going to use that now, we're going to roll this now we're going to move to that. That doesn't work in large organisations. So there's all kinds of constraints that have nothing to do with AI that are now coming in contact with this sort of new pace of change. And I think that that's really, really, really interesting and challenging for those that are tasked with how to manage it. And we see clients across the spectrum managing it in a variety of different ways, all of them really stressed out about how to make this right.
Host: Paul Barnhurst (17:37):
I agree with you. I mean, big organisations, I mean, right. JP Morgan, I worked for American Express and I was located next to a lot of the bank people that had to do CCAR and all those type of stuff. You can't just say, oh, I used AI to do it, and then next time I switched the model. Okay. And how much back testing did you do? Oh, a little bit,
Guest: Matan Feldman (17:57):
Right, exactly.
Host: Paul Barnhurst (17:58):
Yeah, that's only going to get you in trouble. So what do you see? I wanted to ask one or two more questions here and then we'll kind of move into the education area, but what do you see as maybe the biggest opportunities we move forward in this space? And then what's the biggest concern you have as we continue to roll out more and more ai?
Guest: Matan Feldman (18:18):
So the biggest opportunities in the space, let me just sort of narrow the narrow, what that means. I think again, for, I think there are opportunities in every,
Host: Paul Barnhurst (18:28):
And I think really talking to the investment banks to private equity, larger ones, what do you think is maybe the biggest there
Guest: Matan Feldman (18:34):
Within the investment banking world? I think that there are both opportunities on the demand side, and that's probably, that's the biggest unknown, and that will take some time, but that is just the reality of, okay, banks are in the advisory and capital raising game. They help businesses acquire one another, sell businesses, and raise capital. Well, in an AI driven economy, what new advisory opportunities are unlocked for investment banks that become economically feasible, that weren't economically feasible before? That's unknown right now, but I think it would be unreasonable to assume that the only thing that's happening is that there are supply side changes. The investment banking workflow becomes more efficient, but nothing happens in terms of the bank's ability to serve clients better. And all of a sudden more volume and more capacity is unlocked to serve more clients, and that those clients themselves are able to be served because of all the benefits that AI creates for transparency around their own financial systems and their own ability to sort of present themselves and essentially become bankable.
(19:55):
So the demand side, which is largely unknown at the moment, because again, ai, the tools aren't there yet to actually fulfil the promise of ai, but I do think that that's going to be a very, very important part of what happens to the investment banking industry. Those who are innovative are going to really look at the demand side of the equation and say, how else can we better serve our clients? And the same goes for capital raising. So there's advisory and capital raising, and the way that we raise capital today, whether it's through IPOs, bond issuances, loans, these are brittle and friction filled ways to raise capital. And there are many, many ways that you can imagine raising capital, being augmented, improved, facilitated by ai. And so I think that that's probably question number one and an exciting sort of question. The exciting question to ask about that industry, again, it's different from private equity that has its own sort of, but I'll sort of linger on investment banking because obviously the flip side of that is what gets a lot of our attention in a lot of our focus, which is the supply side.
(21:06):
What happens to the analyst, what happens to the modeler, right? We're on a modeling podcast, what happens to the modeler? And I think a lot of that is obviously more front and centre because the tools are now starting to get good enough where modeling itself is much easier to do than an agent with a copilot sitting next to you. And so those questions are probably going to come even sooner, which is okay, what is the role of an analyst now? Is an analyst really now an associate that has to use a lot more judgement and essentially become a quarterback for the work that agents do, similar to what an associate currently does with analysts. And I think that because we're on that road now where AI is getting good enough to do more and more of the sort of workflow of the investment banking job that you're going to have a lot of automation that changes the bank, and again, that creates pressure to sort of flatten the pyramid, whereas the demand side puts pressure the opposite way or relieves that pressure because all of a sudden you need more. It
Host: Paul Barnhurst (22:16):
Provides opportunities, it relieves some of that bottleneck. So it's like how much do you flatten if we have more opportunity?
Guest: Matan Feldman (22:22):
Exactly, exactly right.
Host: Paul Barnhurst (22:24):
Yeah, it makes a lot of sense. It'd be interesting to watch. Well, I'm curious. We know AI is getting better at building models. I think any serious model, any model of real concern is not going to be done completely by ai. You see people saying that, build it in five minutes, whatever. We're not there yet as you mentioned, but as we get better and better at that, how does a modeler stand out? Is AI doing more and more of that model build process, especially early in their career? I know over time, obviously you're doing a lot more deal management and the storytelling and the associate and higher level work, but any thoughts there of how early in that career as AI takes more and more of it, how do you stand out?
Guest: Matan Feldman (23:02):
I guess the simplest sentence, and then I'll get into probably a little more detail on it, it's like you have to see the forest for the tree. So one of the big challenges with modeling is that in that example I gave earlier on in the podcast where I'm sitting there for five hours deep in the modeling rabbit hole, a very common experience of modeling for many of the listeners quite familiar with this is that you've got so much work to do to build the model that you're sort of punting on the, I'm going to take a step back and then I'm going to look at what's actually happening here. So that's the process of I got a big job to do mechanically and then I can take a step back. So what AI does almost instantly is accelerate the time, time at which you can finally take that step back.
(23:48):
In fact, there's a point in the not too distant future where all you'll probably need the role modeler will probably no longer be the right moniker for what this is. It'll be someone who has to evaluate these outputs. And so right now, those two skills are completely intertwined and depending on how you like to work, you're stepping back, diving in, stepping back, diving in simultaneously. And the spreadsheet, and I'm probably preaching in the choir here, is just a uniquely wonderful way to understand a business to tell a story. I don't think that's ever going away in the sense that I don't think the spreadsheet ever goes away, but the role of a modeler is almost, there will be a point where that skillset is no longer, is no longer the right way to look at that role. It has to be somebody who can look at models and evaluate models and really understand the story that's being told.
(24:50):
This is actually not only a real gap in most young modelers today. So young modelers today may know every function and every shortcut and sort of pride themselves on the speed at which they're building models, but there's an entirely different skillset, which is going to be the key skillset. It's already a problem. So maybe taking a step back, one of the big challenges that, for example, our partners on the learning and development side that we see in training is that the modeling is there's a lot of folks out there, and I come myself in that group that just really love modeling just are into that as a practise fun. It's a problem. It's a puzzle that historically has been half of the puzzle and half of the skillset because the other half is really understanding how a business runs. And frankly, I didn't understand how a business ran until I had to start a business.
(25:51):
And so I was modeling for four years or whatever it was, and didn't really understand how businesses function, didn't really understand how to take a step back and understand the story around it today. Again, part of the occupational hazard here is I've been modeling less and running a business more for the last 23 years, but the challenge to me is about how the business run and how we think about how the business runs and using judgement around what is the story being told? I think that is not something, that's not something that's going to be relegated to ai. AI will certainly take a stab at telling you that story, but in the same way that every investor that you have, analysts, investment analysts coming to you could take 10 analysts and they're all going to come to different conclusions about whether Google is a buy or a sell, whether that's a good stock or not, and they all have access to the same information and they're all synthesising in a different ways.
(26:53):
And there's that human judgement element that with all the resources that are available, different analysts are coming to different conclusions you're going to, that's going to be the final layer of where AI doesn't exist, where human judgement comes in becomes really, really important. So long answer to your question, how does a grey modeler stand out? A great modeler at this point starts really leaning into what is happening. What is the story that's being told? I'm going to build a scenario analysis, I'm going to build a sensitivity analysis. That's great. Okay, when I'm sensitising revenue and margins and I'm getting all these different outputs, what's under the hood of how is that generated? And does the rest of my model, is my model segmented in the right way to sort of reflect that? Is my model generating these outputs from a coherent story around investment and returns on those investments? And I think that becomes super, super important. It also frees up the modeler who I don't think will be called a modeler. I think that job will probably be a much more judgement and an investor investment analyst type of title. But that frees you up to doing a lot of things that currently model X can't do. Just think about anything that an investor needs to do right now. Channel checks, interviews, site visits. There's a whole bunch of things that you need to spend time on that you just don't have time to do.
Host: Paul Barnhurst (28:22):
While my background is in fact, I am also passionate about financial modeling. Like many financial modelers, I was self-taught. Then I discovered the Financial Modeling Institute, the organization that offers the Advanced Financial Modeler program. I am a proud holder of the AFM. Preparing for the AFM exam made me a better modeler. If you want to improve your modeling skills, I recommend that AFM program podcast listeners save 15% on the AFM program. Just use Code Podcast. Great point on that. It will definitely free up time. I think you make a point that the value of judgement is going to be much higher. I want to do one or two questions before we move on to the rapid fire and some of the standard questions I have here around the education side. Obviously you're thinking about that a lot as it's disrupting the education space. So how are you thinking about AI in teaching modeling, in teaching people the skills that historically, a lot of what the teaching was around it is how to use Excel. Obviously there's the thought leadership and the principles and everything, but historically you see a lot of modeling stuff focused on how to build in Excel. So how are you thinking about AI on the modeling side of education?
Guest: Matan Feldman (29:48):
I strongly believe that in order to learn, well, taking a step back, I believe that Excel is a great canvas for learning, by the way. So as a canvas for learning corporate finance and accounting, there's nothing like it, right? If you want to learn accounting, there's no better way than to put it on that canvas and understand the flow of debits of credits in the context of a three statement model, for example. There's just no better way to do that. So you have these modeling tools, and historically we have taught modeling in a way that's different from how you actually build a model. You very rarely on the job start from a blank Excel spreadsheet and build a model. That's not a thing you typically do. You're reusing, you're repurposing, what have you. But from a learning perspective, it's incredible to start with a blank spreadsheet and learn from scratch.
(30:53):
And I think to some extent that that doesn't change with ai. So there's learning and there's the doing once you've learned the foundation. I think those two things now we have to be very explicit about as being distinct because once you have a strong foundation, you got to use AI as much as possible. These are incredible tools and you want to leverage them as much as you possibly can. But as an educator, as someone who's tasked with getting people up to a foundation around corporate finance, around accounting, around that foundation is ultimately the foundation for good judgement . I think everyone is, we're all coalescing around the idea that judgement will become extremely important when all these technical skills start getting automated. How do you get to good judgement ? Well, good judgement starts with a really solid foundation of these concepts, and I don't think there are shortcuts to that.
(31:50):
So I think it really does start with the same kind of pedagogy, the same kind of approaches that we've used. What does change is then you now have to layer on top of that this whole world of AI productivity tools and how to interface with those tools. So the job hasn't gotten easier from an education perspective, it's actually gotten harder. You need to both provide the foundation and you need to provide a real breakdown of how to use these tools constructively so that your own foundational muscles don't atrophy. That's really, really hard. It's also hard because you're operating in an environment where you still have the same number of days for training.
(32:37):
You still have the same time and to do the work that needs to get done. So we're spending a lot of time leveraging AI tools actually to create personalised coaching, to sort of leverage the tools on a pedagogical side. So whereas before you couldn't really have one-on-one training pedagogically, it's really, really important. If you can get one-on-one tutoring around concepts you struggle with, well, that would be a huge unlock for an educator, and now you have those tools. So AI is not only a new tool that you have to support in terms of how you train people, but you can also use it as a pedagogical tool. And that's going to be the way, that's the way we're using it to have our cake and eat to you have students, new analysts, new professionals who need to learn a lot more than they did before because you can't lose those foundations. But at the same time, we have tools that are enabling us to deliver that knowledge more efficiently, more effectively. You
Host: Paul Barnhurst (33:32):
Make a great point about both sides, right? There's more learning that has to take place. I've been talking a lot about this. Just take fp and a as an example, I talk a lot about that versus modeling. I'm like, it's already a role that has a high degree of technical and a high degree of soft skills. Well, it's like many finance roles, financial modeling, investment banking, all those things are often some of the higher ones. And now you're asking everybody to learn a whole new tool, which has its own technical learning curve in its own way. And it's really interesting to watch of, okay, how do you do that? How do you bring people up to speed and how much should you expect them to learn versus how much you go out and hire the consultant or the expert to set in that agentic AI and a lot of things in place and they just need to learn how to prompt or use it in the workflow versus in some places they're being expected to build it as well. And so it's fascinating to watch. I think you make a great point there of, Hey, we have to learn these tools. I think everybody's trying to figure out what does that mean and how deeply do people really need to learn? It's going to vary depending on your role, the company, a lot of things. I don't think there's a clear answer of how far you need to go with the tools.
Guest: Matan Feldman (34:38):
No, there's no clear answer. I will say the closest analogy, and it's going to be a poor one because there's nothing like ai, but the closest analogy is when I was starting out the business, Google AdWords, it was an ad platform where that was a really important way that we raised awareness for our business. And I learned that that platform is fairly easy to learn, is fairly rudimentary. Now that platform is really complicated and you need specialists to manage Google ads and run all that. But the foundations back then that I built up actually created a lot of leverage. Same with learning. I had to learn how to build my own website for the company. And so those basic foundations, these tools will evolve, but there's nothing but good that comes from learning these tools in their rudimentary forms. You will build a foundation for how they're evolving in the same way that today, if you use Opus 4.7, it's still not perfect, and you need to know how it's not perfect, in what ways? It's not perfect where the mistakes are happening. They're not happening in the same way that analysts make mistakes, they're happening in different ways.
(35:50):
All of those things, even though we're probably on the same page, that these tools will get better, right? That's not a, yeah,
Host: Paul Barnhurst (35:56):
I think anyone who says they won't, I don't know what they're thinking. They haven't been watching the last couple of years.
Guest: Matan Feldman (36:02):
Yeah, that's not like a hot take these things. Yeah,
Host: Paul Barnhurst (36:05):
That's not a major statement to make.
Guest: Matan Feldman (36:07):
No. So that said, if you, the more you're keeping up with them, the more you're aware of what's happening now that provides its own foundation for understanding where these things are going to go, and you'll be, if you haven't used them, you're like, well, I'm just going to wait until they're amazing and do full end-to-end work. You're going to miss the boat. You got to be on top of what's happening now.
Host: Paul Barnhurst (36:29):
Agree. I think you make a great point. The thought that came to mind is we often talk about first principles. There's a foundational level. Everybody has to learn. How deep you go is going to vary. But there is, like I said, there's a foundational level that needs to be in teaching and that we all need to learn. Alright, so I've enjoyed this conversation. I'm going to move into some of our standard questions. We're going to have a little bit of fun here. We're going to do rapid fire in a minute. But before we do that, what's your favourite shortcut? Is it control S?
Guest: Matan Feldman (36:55):
I will tell you, I'm boring when it comes to modeling. I like to control one and just navigate around getting it into my formats. I like F five for go-to special, nothing too earth shattering, but I find those, I use those a lot more than others.
Host: Paul Barnhurst (37:15):
Bad financial models can lead to bad decisions or worse. So how do you minimise the risk of a bad model? You make sure the models you build are great Financial Modeling Institute developed the Advanced Financial Modeler accreditation programme to help modelers like you. The AFM programme offers a step-by-step approach to building world-class financial models. The programme ensures that you know the best practises in model design and structure and will help you brush up on your Excel and accounting skills too. Be the one on your team to build great models. If you want to impress your boss and your clients, get a FMI accredited podcast listeners, save 15% on the AFM programme. Just use code podcast at http://www.fminstitute.com/podcast.
Guest: Matan Feldman (38:27):
Exactly.
Host: Paul Barnhurst (38:28):
Alright, I'd love to know what's the most unique kind of fun thing you've created a model for or you've built in your personal life with a spreadsheet?
Guest: Matan Feldman (38:37):
So this started out fun and ended not fun as most Excel endeavours for me go. So I was into Wordle and I played it with my daughter and I was like, and she was learning spreadsheets in her school on Google Sheets, not real spreadsheet, but Google Sheets. And I was like, not real
Host: Paul Barnhurst (38:57):
Spreadsheets, that's going to show up in a short, that's going to be a line going to,
Guest: Matan Feldman (39:01):
And I said, Hey, let's create Wordle. I'm going to show you that you can actually create wordle in Excel. It's going to be great and really, really hard. I ended up, I remember we were on a flight and I started it. It was a six hour flight. I started and like all Excel projects, first 80% feeling really good, everything's great. And then that last mile, oh my god, six hours in, it's kind of working. I'm hacking some of the corner cases and then I we're on vacation. I think we're in Florida or something now, instead of hanging out, I'm running out to finish this damn Wordle spreadsheet. I think by the flight back I had gotten it just about wrapped up. It still wasn't perfect. I never actually got it perfect. So that's the honest take. So how long
Host: Paul Barnhurst (39:53):
Ago was it? Could you use AI now to finish it?
Guest: Matan Feldman (39:57):
Actually, it's a good question. I still have it in my shame. It's sitting in some folder. 99%.
Host: Paul Barnhurst (40:04):
You should open up Opus 4.7. See if you can finish it. We'll publish it with the episode. If you send it to me,
Guest: Matan Feldman (40:08):
Actually have
Host: Paul Barnhurst (40:09):
A little
Guest: Matan Feldman (40:09):
That's that's a great idea. I'm going to try that, see if that's alright. If you do that
Host: Paul Barnhurst (40:13):
Number,
Guest: Matan Feldman (40:14):
I'll
Host: Paul Barnhurst (40:14):
Put it on the website with the episode, so fair
Guest: Matan Feldman (40:16):
Enough.
Host: Paul Barnhurst (40:16):
We'll see what the final version looks like with ai. Alright.
Guest: Matan Feldman (40:20):
Alright.
Host: Paul Barnhurst (40:21):
Yeah, I've been there where you're like, I'm just going to get, then you finally, okay, close enough. You get to that point where you're like, I'm done.
Guest: Matan Feldman (40:28):
Yep, exactly.
Host: Paul Barnhurst (40:30):
So Rapid fire, here's how it works. I have a list of about 15 questions.
Guest: Matan Feldman (40:35):
Your
Host: Paul Barnhurst (40:35):
The job is to give a quick answer, take one side or the other, yes or no, or if there's multiple options, you pick one, no elaboration. Then at the end you can take one or two that you know you want to elaborate on because I recognise there's nuance to all of these. So none of the classics, it depends. Consultant answer.
Guest: Matan Feldman (40:54):
Okay,
Host: Paul Barnhurst (40:54):
Ready?
Guest: Matan Feldman (40:55):
I'm big, it depends on Guy. So this is going to be hard for me, but go ahead. Well, you
Host: Paul Barnhurst (40:59):
Laugh, you know in snore, obviously he sponsors the show. First time I had him on, he couldn't do any of these. He had ad context to every single one. It's just not in his nature. So I understand.
Guest: Matan Feldman (41:08):
Let's do it.
Host: Paul Barnhurst (41:09):
So let's see if you can do this. Circular references, yes or no?
Guest: Matan Feldman (41:12):
Yes.
Host: Paul Barnhurst (41:14):
VBA.
Guest: Matan Feldman (41:15):
No
Host: Paul Barnhurst (41:17):
Lambdas in financial models.
Guest: Matan Feldman (41:20):
No
Host: Paul Barnhurst (41:21):
External workbook links?
Guest: Matan Feldman (41:23):
No.
Host: Paul Barnhurst (41:24):
You said that modelers are easy modelers. Should they be keyboard warriors?
Guest: Matan Feldman (41:29):
Yes.
Host: Paul Barnhurst (41:31):
All right. Should all models be print ready?
Guest: Matan Feldman (41:35):
No.
Host: Paul Barnhurst (41:36):
Are merge cells ever acceptable?
Guest: Matan Feldman (41:38):
No.
Host: Paul Barnhurst (41:40):
Should financial modelers learn Python in Excel?
Guest: Matan Feldman (41:43):
No.
Host: Paul Barnhurst (41:44):
What about Power query?
Guest: Matan Feldman (41:46):
No.
Host: Paul Barnhurst (41:47):
Power bi?
Guest: Matan Feldman (41:48):
No.
Host: Paul Barnhurst (41:49):
Alright. Should every financial modeler be able to build a fully integrated three statement model?
Guest: Matan Feldman (41:57):
Yes.
Host: Paul Barnhurst (41:57):
Okay. Will Excel ever die
Guest: Matan Feldman (42:01):
On a long enough timeframe? We all die.
Host: Paul Barnhurst (42:04):
That's my answer,
Guest: Matan Feldman (42:05):
But let's put it this way, not in our lifetime. Okay,
Host: Paul Barnhurst (42:08):
Fair enough. I had someone say yes, just please don't let it be in my lifetime. And I laughed, so, alright. I'm pretty sure I know the answer to this one. Bo ask, have you used AI to help you build a model?
Guest: Matan Feldman (42:19):
Yes. Okay.
Host: Paul Barnhurst (42:21):
What financial statement is most important for modelers? Income statement, balance sheet or cashflow statement? It's
Guest: Matan Feldman (42:27):
The income statement and the cash flow statement. If I had to pick one, it's the income statement.
Host: Paul Barnhurst (42:30):
What's your favourite? A LLM.
Guest: Matan Feldman (42:33):
Claude.
Host: Paul Barnhurst (42:34):
Okay. If you could pick one and only one for all of your models, would it be sensitivity analysis or scenario analysis?
Guest: Matan Feldman (42:43):
Sensitivity analysis.
Host: Paul Barnhurst (42:45):
Okay. Do you believe financial models are the number one corporate decision making tool?
Guest: Matan Feldman (42:51):
That's a yes.
Host: Paul Barnhurst (42:51):
Okay, we'll go with the, yes. Then I'll have to ask the follow ups. What's your lookup function of choice
Guest: Matan Feldman (42:56):
X lookup?
Host: Paul Barnhurst (42:57):
Alright. I could tell you struggled with a view of those you were thinking for a while. So pick one or two you want to elaborate on.
Guest: Matan Feldman (43:04):
So a lot of these, the no. So for example, for VBA and Power query, power bi, python, these are all massive. It depends on the project. So if you're in investment banking, the answer is no. Right? Almost certainly not. In investment banking, if you are within a finance function and an fp function and you are tasked with a tonne of data analysis and you have to do that, it's definitely a yes. So that's why it depends. The problem is in investment banking, you have people that go crazy with some of this stuff and it actually gets in the way of what their job is. So a lot of these are, it depends. I would look at those from the investment banking context and it really is different in other contexts, I think. I mean, I'm trying to think what are the other ones that I was struggling with? I think corporate decision making. So it's really, really difficult to look at a model and say this is the number one corporate decision making tool on strategy. So it's a great facilitator of strategy. It's hard. I'm now clouded by, I think I would've said yes with no reservation up until five years ago. But I think as the years have gone by, there are things that a model cannot surface and there are more of those than I would've probably appreciated five years ago.
Host: Paul Barnhurst (44:18):
I say yes, but I'm a little more nuanced on that one than probably I used to be as well. My favourite answer I've got on that one all the time was someone said, no politics. I'm like, how do you argue with that?
Guest: Matan Feldman (44:28):
Exactly. So there are things, politics is just one, but again, there's things that a model doesn't surface. So return on investment, return on capital, it does surface, and that's an important strategic sort of decision-making tool. What it doesn't surface is obviously politics, but there's also external things like competitive dynamics, a whole bunch of things that obviously influenced the model because they influenced the return on investment and some of those decisions. But those are really, in many ways exogenous to the model and have their own considerations and are actually really, really important corporate decision making tools. So that's why, I mean, I did actually, I think I switched. I said yes at the end.
Host: Paul Barnhurst (45:06):
You did end up paying. Yes. You went back and forth. You left it at Yes, but it's,
Guest: Matan Feldman (45:10):
I'm just giving the counterpoint. I think it's ultimately so valuable that I brought it back to a yes. But there are lots of other considerations
Host: Paul Barnhurst (45:17):
And I like how you clarified, right? Obviously depending on your work, a certain type of work, you have to learn Power query. I say yes to that. I came from fp and a, you said, no, you came from investment banking. A lot of that is coloured by our experiences. And then there's some that you get very strong opinions like circular references and VBA, and again, coloured by your experience. But it's always interesting to see because there's certain ones that you're like, I could go either way. I don't know that I care. And there's others where it's like, that matters to me.
Guest: Matan Feldman (45:45):
Well, circular references are a great example of, it's kind of a 50-50 call in banking. It's probably a no if you're just looking at the banking lens, but there's no project finance model, for example, that exists without circular references. I mean, it's a core feature of those models, so I can't say no to those. And it really is, goes back to my favourite line. It depends. Unfortunately that line doesn't work year round. Yeah,
Host: Paul Barnhurst (46:07):
It doesn't make for a great podcast if I let you say. It depends 15 times in a row.
Guest: Matan Feldman (46:11):
No, no, sir.
Host: Paul Barnhurst (46:13):
Alright, well I know we're up against our time here. So last question. If our audience wants to learn more about you, potentially get in touch, learn about the resources, wall Street Prep offers, what's the best way for them to do that?
Guest: Matan Feldman (46:24):
Easiest ways to go on our website. Wall Street prep.com can also check out my LinkedIn page. That's it. Those are probably two easiest ways.
Host: Paul Barnhurst (46:30):
Perfect. Well, thank you so much Matan for joining me. A pleasure chatting with you. I love what you've built at Wall Street Prep. I know you got a lot of great programmes out there and I've been fortunate enough to be instructor with the fp and a one, so keep up the great work. Thanks
Guest: Matan Feldman (46:43):
Paul. It was nice talking to you.
Host: Paul Barnhurst (46:44):
Thank you.
Guest: Matan Feldman (46:46):
Bye.
Host: Paul Barnhurst (46:47):
Financial Modeler's Corner was brought to you by the Financial Modeling Institute. This year, I completed the Advanced Financial Modeler certification, and it made me a better financial modeler. What are you waiting for? Visit FMI at https://fminstitute.com/podcast/ and use Code Podcast to save 15% when you enroll in one of the accreditations today.