Litigation Risk Modeling for Finance Professionals with David Perla
In this episode of Financial Modeler’s Corner, host Paul Barnhurst (aka The FP&A Guy) welcomes David Perla, Vice Chair of Burford Capital and a pioneering voice in the legal finance industry. David discusses how financial modeling plays a central role in evaluating and funding high-stakes litigation, offering a behind-the-scenes look at how his team monetizes legal risk with structured financial strategies. From managing long litigation timelines to leveraging Monte Carlo simulations and AI tools, David shares how Burford builds sophisticated models to guide investment decisions and mitigate risk.
David Perla has an extraordinary background at the intersection of law, entrepreneurship, and finance. He’s a former president of Bloomberg Law, co-founder of legal outsourcing pioneer Pangea3 (acquired by Thomson Reuters), and was named a top 50 innovator by The American Lawyer. At Burford Capital, David is helping reshape how the legal world accesses and deploys capital.
Expect to Learn:
What legal finance actually is and how it works
How Burford structures deals and evaluates legal risk
The modeling techniques used to forecast legal outcomes
Why Monte Carlo simulations are central to their underwriting process
How AI is used (and misused) in legal finance
Here are a few quotes from the episode:
“Behavior that’s irrational but predictable? We can model that. Idiosyncratic behavior? We avoid it.” - David Perla
Modeling helps us decide not only how to price a deal, but how to structure incentives.” - David Perla
Everything we do depends on modeling.” - David Perla
David Perla gave us a fascinating look into the world of legal finance and how modeling plays a critical role in managing complex litigation risk. His knowledge revealed how finance principles, AI tools, and human judgment all intersect in this unique investment space. From non-recourse funding to Monte Carlo simulations, this episode showcased the depth of modeling required in high-stakes legal decision-making.
Follow Dave:
LinkedIn - https://www.linkedin.com/in/davidperla/
Website - https://www.burfordcapital.com/
Follow Paul:
Website - https://www.thefpandaguy.com
LinkedIn - https://www.linkedin.com/in/thefpandaguy
YouTube - https://www.youtube.com/@thefpaguy8376
Follow Financial Modeler's Corner
LinkedIn Page- https://www.linkedin.com/company/financial-modeler-s-corner/?viewAsMember=true
Newsletter - Subscribe on LinkedIn-https://www.linkedin.com/build-relation/newsletter-follow?entityUrn=7079020077076905984
Sign up for the Advanced Financial Modeler Accreditation Today and receive 15% off by using the special show code ‘Podcast’.
Visit https://bit.ly/3AK5dYo and use the code “Podcast” to save 15% when you register.
In today’s episode:
[01:12] - Welcome to the episode
[03:37] - Defining Legal Finance
[04:53] - Typical Legal Finance Timeline
[07:56] - Structuring Deals with Non-Recourse Funding
[11:17] - Modeling Risk with Monte Carlo Simulations
[16:32] - High-Stakes Legal Deal Sizes
[21:34] - Predictable vs. Rational Behavior in Modeling
[26:02] - Data Transparency in Insurance vs. Legal Finance
[32:31] - Tools of the Trade: Excel, Python & More
[39:44] - Future of Excel, AI, and Model Confidence
[41:32] - Strategic Decision-Making and Wrap-Up
Full Show Transcript
[00:01:12] Host: Paul Barnhurst: Welcome to Financial Modeler's Corner. I am your host, Paul Barnhurst, aka the FP&A Guy. This is a brand new podcast where we talk all about art and science. Of financial modeling with distinguished guests from around the globe. The Financial Modeler's Corner podcast is brought to you by the Financial Modeling Institute. FMI offers the most respected accreditations in financial modeling, and that's why I completed the Advanced Financial Modeler. This week I'm thrilled to welcome on the show David Perla. David, welcome to the show.
[00:01:51] Guest: David Perla: Paul. Thank you. I'm thrilled to be here and I appreciate you having me.
[00:01:55] Host: Paul Barnhurst: I have to ask, as a lawyer, did you ever think you'd be on a modeling Podcast.
[00:01:59] Guest: David Perla: I absolutely did not. And on the ride over here, I was thinking, I hope I can keep up with your audience in terms of our modeling and our financial acumen.
[00:02:09] Host: Paul Barnhurst: Well, I think it'll be great for audiences, something different. We like to cover a lot of different things. Let me share a little bit about Dave's background and then we'll get into the questions. So David is vice chair and a member of the management committee in his role at Burford. Mr. Perla is responsible for marketing, public policy, industry affairs and public relations. An entrepreneur and legal industry leader with expertise in building high growth, legal and technology driven businesses. He was named a top 50 innovator of the last 50 years by the American Lawyer. Prior to joining his current company, he was president of Bloomberg BNA legal division Bloomberg Law, where he oversaw the legal and related products, including its flagship Bloomberg Law Enterprise legal news, information and tools platform. He also co-founded and was the co-CEO and a director of Pangea3, the top ranked global legal process outsourcing provider. Pangea3 was acquired by Thomson Reuters in 2010 and grew to over 1000 employees globally under his leadership. Before launching Pangea3, he was vice president of business and legal affairs for Monster.com. He began his career in the New York office of Katten Muchin. Welcome again to the show.
[00:03:33] Guest: David Perla: Thank you.
[00:03:34] Host: Paul Barnhurst: Alrighty. Well, let's start with this question. You guys are a legal finance company. Can you tell our audience what that means, what exactly you do? And just give us a little background of your guys' company?
[00:03:44] Guest: David Perla: Certainly. Legal finance is a solution for both law firms and the clients they serve. And in addition, Paul, it's a solution for some of the third parties that operate in what we think of as the legal industry, as opposed to just the legal profession. So the easiest way to think about it, for Modeler's or for people in the finance world is we help companies, we help law firms, and we help intermediaries to understand legal risk, to value legal risk, and to monetize legal risk or use legal risk to offset expenses. So we can do that either via paying fees and expenses to pursue meritorious claims on behalf of companies or law firms. Or we can use it to create liquidity by monetizing claims or by monetizing successful claims that have become what we call judgments, but that take oftentimes many years to collect. So we are in some ways an investment bank for law.
[00:04:53] Host: Paul Barnhurst: That's an interesting way to look at it. It makes a lot of sense. So investment kind of bank for law. And what's the typical time that someone may need help? Like your company comes to you. They have a claim that's legit. You know, from when they started that whole legal process to when a judgment happens and they get paid out. How long is that? Typically? I'm sure it varies a lot, but what kind of range are we looking at?
[00:05:17] Guest: David Perla: So it does vary, but most clients come to us early in the process. They're either considering a claim they've been harmed in some way, usually to the tune of many tens of millions of dollars. There are businesses, commercial, legal, finance, so we only work with large commercial types of entities and the law firms that represent them. They may come to us very early, they have a claim and they are working with a law firm, but they haven't yet filed the suit all the way to. They've already litigated, but they have a judgment and everything in between. And that in between can be years. Usually it's at least a few years from the time you file until the time that you actually have some sort of a resolution. They may come to us to fund the expenses of litigating that. And those expenses include law firms. They can include economic experts, other types of experts, or they will come to us because they want to create liquidity. They want an advance against the potential outcome of that litigation. Once we're involved, Paul, what we have found over the course of our 15 plus year history is from the time we get involved, on average, it's about two and a half years from the time we deploy funds or deploy capital against a claim or a portfolio of claims. Until the time we see a recovery for Burford, and that's in our public filings. We're a public company. but some of them, because that's an average, some of them are much faster where we are advancing funds against a judgment, and we have to help collect that judgment. Some are much slower. There's examples where we've been, had our capital deployed for five, six, seven years and we're still working with the client in, in helping them get an outcome. The whole idea, though, is for the client to bearing that duration risk. And that's part of the modeling we do, is modeling for that duration risk. We bear that risk and we take that risk off of the client's hands.
[00:07:26] Host: Paul Barnhurst: Yeah. No, that's true. That duration risk if it's six seven years before they collect. So client comes to you, you think they have a meritorious claim. Sometimes you're paying them out a fraction of what the judgment may be. Sometimes they just want the legal fees. How at the end do you collect your money? You collect when they get paid, and do you just put a kind of a fee on top of it, or kind of how does that work for you guys to recover your money and profit. Like, what's kind of the model there?
[00:07:56] Guest: David Perla: Everything we do is on a non-recourse basis. So your questions the the absolute perfect question which is if for non-recourse and we're not debt, we're not equity, how exactly do we make our money. The short answer is up front, when we, determine that we are going to finance a case or a group of cases, what we in our business call a portfolio, we will come up with a structure. We'll agree with our client as to how we get paid back and on what terms. So there are really three ways that we can get paid back. And oftentimes it's a combination or a hybrid. So normally the way the business works is most companies like us will get what's called their investment back. You'll hear the term IB. So we'll get our capital back plus either or both of some multiple of that investment back or some amount of money determined by the recovery that the client obtains. Okay. So we'll call that a percentage or a hybrid. And in both of those cases, either the multiple or the percentage of the outcome that the number can rise over time or it can rise with milestones. So it may well be that we would get our investment back plus a multiple up until the point where the case goes to trial, for instance.
[00:09:23] Guest: David Perla: But if it goes to trial and it doesn't settle, that multiple might go up. Okay. Or if it goes to trial and it doesn't settle, we would get then either a higher multiple or a percentage of the ultimate award. And that that percentage also might go up either over time or because of milestones. And all of that is modeled upfront and all of that is agreed to up front. So, you know, I laughed at the beginning of our conversation about, did I ever think I'd be on a podcast about modeling. Everything we do depends on modeling, because we have to think about what are the types of milestones and what are the types of outcomes, and how do we get paid back. But also, Paul, how do we make sure that our client is incentivized properly to behave in a rational, economic way? Right. We want to make sure if the case can be settled at an appropriate number, that it will be settled, but also, if a case can't be settled at the right number, and the better outcome would come from a trial that the client is incentivized not to settle too early or for too low a number. So we're thinking about all of that at the very beginning of the potential matter.
[00:10:38] Host: Paul Barnhurst: A lot of unknowns, a lot of uncertainty in that. So I can totally understand that modeling is really important. And in particular managing that risk. You have no guarantee T if there's no settlement, they lose the the trial. You may not be getting anything at the end. So how do you know let's start with how you manage risk. I mean, obviously you're doing modeling. Are you doing Monte Carlo? A lot of sensitivities here. How do you think about that outside of the contract? I'm sure there's legal things you do to manage that risk. Like you mentioned the milestones and the way you set up payments. But just in general, how do you think about modeling? And the risk.
[00:11:17] Guest: David Perla: We think about is a set of possibilities or a set of nodes. And we have dozens and dozens of nodes in our financial model, and that helps us calculate how we want to obtain a return and negotiate that with the client. But those nodes are also important for how we value the case for carrying purposes. Because we're public, we have to carry these on our books as an investment for purposes of the SEC. And so we have to think about both of those purposes for the model. So our model, which is dozens of nodes, looks at all the elements of risk that go into, on the one hand, pricing and the other hand valuation. We've learned that over the course of our history. And we then have to weigh each of those. We have to sort of put not just a value, like, what is the percentage likelihood that the case will settle at a particular point in time? But what happens to that percentage if certain other things happen in the case earlier? So later nodes get impacted by earlier nodes, which is not a surprise to anyone who's taken statistics or modeling. You know, we then add on Monte Carlo modeling. We want to layer in a more objective form of modeling.
[00:12:34] Guest: David Perla: And then we want to compare. We want to compare against other cases. And that goes into the model. So we may come out with a model one particular way and say well how does that model compared to what we've done on similar cases and what facts are different. What's changed? Maybe the court is different. The particular judge may give us a different outcome within the model might might give us different weighting or a different percentage. It could be that the lawyers, you know, it could be the defendant who's on the other side. Is it is it a company that settles early and settles it at a number that is appropriate for an early settlement, or is it a company that will only settle early for a low number and will always try the case? That's going to impact not merely the duration and that node, but it's going to impact where we wait the earlier nodes and what's going to happen. So all of that, you know, is done by our, our modeling team, which are, you know, traditional quantitative analysts, you know, people with math degrees, statistics, degrees and in some cases, physics degrees. But people you would think of as quants.
[00:13:42] Host: Paul Barnhurst: Sure. Yeah. I was going to say you have to have some pretty strong quants with the Monte Carlo. The number of variables. Everything you're doing. It reminds me sometimes a little bit like underwriting and some of those type of things, some similar.
[00:13:54] Guest: David Perla: It is underwriting, Paul. And what's fascinating about it is our quantitative analysts have learned enough law that they can sit with our underwriters, who historically are lawyers, and the lawyers have learned enough about modeling and the mathematics and the statistics to be able to create something that's greater than the sum of those two parts. In terms of how the model works, in terms of how we think about the case. The other thing that it's allowed us to do is it's allowed us to take data from public sources and our own proprietary data and use both in terms of how we model.
[00:14:33] Host: Paul Barnhurst: Yeah, I got it now. Interesting. So let's say a client comes to you like how often do you accept the you gotta do the modeling. Make sure it makes sense. How often do you accept the client or are you rejecting a certain percentage saying, hey, this is just way too risky? We've run the numbers and it doesn't make sense for us to fund this.
[00:14:54] Guest: David Perla: A few things happen. The short answer is we end up funding a significant minority of the opportunities that come to us, whether we seek them out or they seek us out. And by that, I mean, if you add up things like inbound emails, the phone rings, all the meetings we go to where people say, I'd like you to look at something, and it goes into what we think of as intake. We're funding less than 10% of that.
[00:15:21] Host: Paul Barnhurst: Okay. So that's a pretty small number.
[00:15:23] Guest: David Perla: It's a very low number. And that is the industry standard within the complex commercial litigation finance industry to be funding less than 10%. However, once something gets into the formal underwriting process, meaning we've sensed that it meets certain criteria. We like the merits of the case. We like the facts, you know, facially we think the counterparty is credible. We think they're in the right, for lack of a better way of putting it. We think the defendant is creditworthy and we don't want to be in the credit risk business. We want to be in the litigation risk business. So we think the defendant is creditworthy. We think it's in the right forum, in the right jurisdiction. Things that lawyers think about then it's a higher percentage of cases that we will actually fund. But we're going to undertake a rigorous underwriting process, which can take anywhere from a few weeks to a few months. That includes all the legal research, all the financial modeling, all the negotiation, all the structuring of the terms. It looks like any other financial transaction. Underwriting and negotiation.
[00:16:32] Host: Paul Barnhurst: Yeah. You mentioned investment bankers. Obviously a lot of M&A there, but a similar type thing in the sense of I'm sure there's some back and forth discussion. You're talking large dollars. I imagine, you know, these cases are millions, tens of millions, maybe even hundreds of millions sometimes.
[00:16:49] Guest: David Perla: Yeah. Generally speaking, we are only involved if the amount in dispute is significantly in excess of 50 million USD. Makes sense. And so the way to think about the business is legal fees and expenses. The cost of litigating is often about 10% of the value of the case, or the likely damages of the case. Any number higher than that. And it's very hard to use legal finance. So if $100 million likely damage is going to cost $20 million to litigate to get to a successful outcome, that's not a bad case, Paul. It's just not one that likely supports financing. When we account for the fact that we need to make a return, and the counterparty needs to behave in a way that's economically rational, we will use in the businesses the term headroom. There has to be enough headroom for Burford to make its money. For the law firm to get paid, and for the client to recover enough money that the client is happy, and that that headroom number tends to be about 10 to 1 from an investment perspective.
[00:18:01] Host: Paul Barnhurst: Got it. That makes sense. So just to make sure I got it right. You know, you have $100 million. Let's just say that's what you think you're going to settle at. That's what you estimated legal costs are going to be in that 8 to 12 million range. So it's roughly ten times that's something you would consider. You want that headroom at least ten times. So you want 10 million or less ideally.
[00:18:23] Guest: David Perla: That's correct. And the reason we operate at a higher number in the market is twofold. Number one, we're very large on a given year. We deploy give or take about 1 billion USD. And our portfolio is about $7.4 -1 billion. The total value of what we've invested, plus the kind of the markups of it, the value of the portfolio. The other reason is in litigation, we have learned that the settlement dynamics and the resolution dynamics from a modeling perspective look different as the case numbers get very small. So when people are arguing over $100 million to follow your example, that tends to be an economic decision. There are business decisions in terms of how they run their business. But those are economic actors. When people are arguing over, for instance, 3 to $5 million. Those are often smaller businesses. Those are often people less familiar with litigating, um, the dynamics of settlement and how those litigate tend to be very, very different, even if that ratio holds up. So it's not merely that we'd have to do a lot more of them, it's that they behave differently than the larger cases.
[00:19:41] Host: Paul Barnhurst: While, my background is in FP&A. 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 modeling 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 the AFM program. Podcast listeners. Save 15% on the AFM program. Just use Code Podcast.
[00:20:22] Guest: David Perla: There's more uncertainty and hence more risk because you're often dealing with people that have less experience, and so they may not behave as rational as someone who's been through it 15 times. You have a better idea of how they're going to behave and can model more of that risk and uncertainty. Certainly.
[00:20:39] Guest: David Perla: That's correct. And they also made a larger claim is often brought by a party that is treating it as an economic matter without emotion tied up into it or without other inputs tied up into it. We can model for really any economically driven decision making.
[00:20:58] Host: Paul Barnhurst: Sure. Yeah. When people behave rational, it's much easier to model it.
[00:21:03] Guest: David Perla: That's correct.
[00:21:04] Host: Paul Barnhurst: You know behavioral economics. Right.
[00:21:06] Guest: David Perla: Even if we think of, you know, in traditional behavioral economics really assumes there's no rational actor, or if we if we go back to sort of the old Twerski and Kahneman, behavioral economics, you can still model for predictable, irrational economic behavior, but you can't model for is idiosyncratic behavior. And so we try and avoid anything where we think people are going to behave. Idiosyncratically.
[00:21:34] Host: Paul Barnhurst: Totally agree. Right. There are certain things where it may not be rational, but it's predictable.
[00:21:38] Guest: David Perla: Correct?
[00:21:39] Host: Paul Barnhurst: Like overdraft fees from a start. Smart financial standpoint. There's no rational reason you should be constantly using overdraft fees, but you can predict a certain amount of the population for various reasons will use them.
[00:21:52] Guest: David Perla: Correct. The piece we can't do for individual cases is we can't undertake any actuarial modeling. We can model industries on a thematic basis. So I'll use the term actuarial Paul. We can model general trends on an actuarial basis. But in traditional complex commercial litigation, it's nearly impossible to undertake actuarial modeling on an individual case, which makes case predicting very, very difficult to do with something like artificial intelligence. Everyone's talking about AI, and AI has tremendous value in what we do. But when people say, well, it can predict the case. Our cases are sufficiently large and unique that to try and tailor a large language model to predicting the outcome would be to say, well, we could create an actuarial model and you just can't with large scale, complex commercial litigation.
[00:22:56] Host: Paul Barnhurst: Well, I would imagine each of them are so unique. There's not a big data set either out there, right? Large language models want to be trained on a huge amount of data, and there's very limited data. And each case has its own idiosyncrasies to it.
[00:23:12] Guest: David Perla: That's correct. So number one, there aren't 1 million or 10 million of the types of cases we invest in. But even if there were, the facts differ from case to case, right? There are no slam dunks in terms of it being, you know, 100% the claimant is going to win or 100% the defendant is going to win. Usually when they're that clear at the beginning. They set up long before anyone comes to talk to Burford about financing. If they're that obvious, there's occasionally a claim where a client comes to us and it's simply a matter of a very well-heeled, very well capitalized defendant and a claimant that doesn't have money. But it has a slam dunk case. Usually it's not quite so clear. And the underwriting involves facts, and the application of the law to those facts. And the law may vary in jurisdiction to jurisdiction. So even if we had a large volume of cases, you'd have these idiosyncratic facts. If I were to contrast that, let me give you an area where I think AI is going to play an enormous role. Yes, it is personal injury. So if you look and the reason how do we know that if you look at insurance in the property and casualty area and in and in the personal injury area insurance for decades has been using actuarial models to figure out what the claim value is and what to pay and what to offer, and I did.
[00:24:46] Guest: David Perla: Insurance defense work for a summer when I was in law school. We did airplane crashes. We knew what injuries were worth depending on where the person lived, their age, what they did for a living was known. 30, 35 years ago, already long before we had artificial intelligence tools. But it's easy to to, to sort of see you have a large volume of cases. Many of them are known. Not all. Many settle, but it's not terribly hard to get the settlement data. And even anecdotally, it's relatively straightforward to get that data versus in the commercial world, if we look at settlements in something like 80 to 90% of complex commercial claims filed end up settling, so they end up resolving without the resolution being public. The only settlement data we can look to is our own data. There is no public database of settled commercial cases precisely because they're confidential.
[00:25:47] Host: Paul Barnhurst: Yeah, they don't want that information out there. They don't want you to know the amount they settled for and all the details and what the client can say and can't say. And because often it doesn't look good, sometimes it might. But, you know, it's just private. They don't want the numbers to be known.
[00:26:02] Guest: David Perla: It's just private. If you flip that, if you look at insurance, the reason insurance was able to do it for so many years at the industry side was every insurance company globally has always known what it settles for. And they have a massive volume. But even the personal injury firms, some of them are very, very large. Some of them are doing thousands of cases a year. They have a fairly robust set of data around not only what a case is, settle for what's the value of an injury or a harm in a particular place, meaning particular criteria. But they also know what that value is by insurance company. So if you talk to personal injury lawyers, they could tell you here's what you know. Here's what a Geico will settle at versus here's what a progressive will settle at. They know that information because it's such a high volume business and there are things in between. But the more idiosyncratic the claim becomes and the larger it becomes, the less likely that we'll be able to model it on an actuarial basis, and therefore, the less likely we are going to be able to run AI driven models. It's not to say we can't use AI. We use plenty of AI, just not on outcome based, not on predicting the outcome.
[00:27:19] Host: Paul Barnhurst: Sure. You know, and something you said I thought was interesting when you talked, you know, personal injury and, you know, the geicos the insurance of the world, right? I can get information going out there to know who's good at paying claims and who usually offers an amount, right? There's enough information out there. You see, even Consumer Reports and different people saying from a customer service supporting your claim, these ones are best. You generally get the highest percentage and you can make your choices accordingly. Where you know you don't have that in your industry. It's amazing how much I can even just figure out going out on the internet about that kind of stuff today.
[00:27:52] Guest: David Perla: That's exactly right. And property and casualty insurance is the best example which you used. Most people know they're insuring. In the case of my house, I live in a very old house. I'm insured with a company that will pay, but it's also a company where you don't file a claim for the average damage to the house you're using. The insurance company I use in case a major problem happens. Because I have a very old house that's very hard to rebuild. There are other, you know, car insurance. Most people view it as sort of a cost of owning a car. And so they're looking for the least expensive. They're not worried about the case. Well, what if my car gets totaled, right, because they have to insure their car. And so they're saying, how do I get the cheapest price, even though I think I'm really not going to use the insurance? It's very different than owning a house.
[00:28:43] Host: Paul Barnhurst: Yeah. No. All great examples. Right. There's a lot of difference there. And sometimes when it's required okay, what's the cheapest I can get. Other times it's like okay this is the biggest investment in the world. It's my house. I want more. You know you're going to make different behaviors. I'm curious talking to I know you wrote an article about, you know, the current and future state of AI and legal finance. You mentioned, you know, there's definitely areas you can't use it. What areas have you most excited about AI? What are you most excited about in your field?
[00:29:11] Guest: David Perla: I think we're very excited about the ability to find cases that are amenable to legal finance solutions, and where the claimant would likely want to finance their claim if they knew about what we did. So we're still a nascent industry. And so we think there are still cases out there where claimants or their law firms don't necessarily know that they've got a claim that they can either monetize or where we could pay the fees and expenses. So from an origination standpoint, we think AI is very powerful, much more powerful than simply sending emails to lawyers or to companies. So we're using it to identify potential cases once filed. We're also seeing a lot of exciting use cases where technology companies are now finding behaviors and patterns in the public sphere that give rise to meritorious claims of damage. So, in other words, they're finding harm that wouldn't otherwise have surfaced, and they're able to say to the groups of people or the corporations, you've been harmed. And number one, you can have that harm addressed. Right. And you can be made whole for that. But number two, you can also turn that harm into a legal asset via the use of legal finance. So those are the two areas on what I'll call the front end of our business. Paul. The other area is just when I talk about our modeling, our own models, you know, are now sitting on technology platforms where our ability to model them faster is dramatically improved with the use of artificial intelligence tools or even just software tailored towards modeling.
[00:31:07] Guest: David Perla: Right. And that an AI is helping us sort of write through that code and write through those models. So that's a third area. The other reason I wrote about it was there's a lot of attention in the litigation space to this perceived holy grail of predicting the outcome. And I just don't think it's real. And I think the more people write about something that's not real, it does a disservice to the actual use of artificial intelligence in ways that are helpful and thoughtful for both sides. Whether you fund affirmative litigation as we do, or whether you're on the defense side or you're an insurance company, we want people to know there's a real value in use to AI. But when people are writing things that I think are just silly, you know, about, well, we can, you know, we can predict that this case is going to win. Sure. If it's a personal injury case, you know, to my earlier point. But when you start saying things like that in complex commercial litigation, it it calls into question the credibility. And then in our view, we needed to address that, to say there really are uses, but don't get distracted by this outcome that if it ever were to happen, it's decades away. This predictive modeling.
[00:32:27] Host: Paul Barnhurst: Yeah. Got it. No. That's helpful. Well, I want to ask a question. You mentioned software and different things. What are your teams that are building these models? Are they building them in Excel? Do you have a special software you use or how does that work?
[00:32:41] Guest: David Perla: So we know everyone uses Excel. It's the thing everyone loves to hate and loves to use. Um, we also use Python's, so most of our quants can code. That's sort of a widely known secret among Modeler's. I suspect most of the people listening to this are going to say, yeah, I can write code also. So they are just for the straightforward models in Python. And the reason we're doing that is it allows us to to do anything we want with the data and to manipulate the data. And then we use a series of financial software tools in our finance and quant department that are, you know, third party tools.
[00:33:15] Host: Paul Barnhurst: Yeah, I got it. I'm sure for the risk modeling, whatever else, Monte Carlo, all those types of things.
[00:33:20] Guest: David Perla: Correct? Correct. But you know, look, we are like most organizations in the finance world, incredibly conversant in Excel, and I expect us to remain incredibly conversant. Um, you know, my son's going to business school next year, and I joke with him, you know, my world is horizontal, not vertical. You know, if I'm not in Excel, I'm often in PowerPoint. And, you know, otherwise, I'm writing emails. But we're not really writing narrative documents.
[00:33:50] Host: Paul Barnhurst: Sure. And I'm curious, how much time do you spend kind of reviewing models in Excel doing things? I know you're in the law position here. You're a lawyer, but you're also helping run the business. You're not doing the modeling. How much time do you actually spend in Excel? I'm just kind of curious.
[00:34:05] Host: Paul Barnhurst: Well, you may know me as the FP&A guy. What you may not know is that I am also passionate about financial modeling. Like so many other financial Modelers, I was self-taught. I didn't know that there was an organization dedicated to advancing the discipline of financial modeling until I discovered the Financial Modeling Institute. FMI offers the Advanced Financial Modeler Accreditation, a program designed to strengthen and validate your financial modeling skills. I am a proud holder of the AFM accreditation. Preparing for the accreditation exam made me a quicker and more confident modeler. If you want to become a better modeler, I can't recommend the AFM program highly enough. Podcast listeners. Save 15% on the AFM program. Just use Code Podcast at www.fminstitute.com/podcast.
[00:35:09] Guest: David Perla: So on the model side, it's probably hours per week. Every potential matter is modeled and so you have to be able to review the model. When we look at what we call commitment memo or commitment committee memo, it's got the model built into a narrative overview of the case and all the scenarios. So you have to be able to read the model. I prefer to read it in Excel, you know, the underlying raw data. But in terms of how we use Excel every day, there's not a day at a place like Burford where I'm not looking at an Excel spreadsheet of one type or another. They're not all models. Sure, but Excel is the lingua franca of how we sort of communicate the numbers at the business.
[00:35:53] Host: Paul Barnhurst: Yeah. And so leads me to a question. When you're looking at a model, let's say you're evaluating a case. What lets you know that, hey, they've done a good job modeling. What type of things do you like to see to make it easy to read the model? Are there certain things that make it, you know, easy for you as someone who's a consumer in that sense, right. You haven't built the model, but you got to use it to evaluate things. What do you kind of look for?
[00:36:16] Guest: David Perla: So we've evolved when we've been modeling for many, many years already. I think number one, our models are dynamic. And so the presentation has to look somewhat different than the underlying data. So I love being in the spreadsheet, but when it presents, it has to present in a way that allows someone to change a variable and see how that ripples through. Which means it also has to render all the appropriate variables without looking at, let's say there were 70 nodes the average person can't look at 70. So one of the important things is, which are the models we need to present. So, you know, the settlement node is always going to be really, really important. If we're early on the case, what's the summary judgment node. What's the trial. You know, what's the node that says, you know we're going to go to trial. What's the node that says we can win a trial. What's the appellate node? Those things became very, very important over time. Beyond that we have some variability in the models depending on the type of case. You know, the subject matter of our case, what we think of as our pipelines are going to look different. And we also have certain expectations as to within a type of case, how the model needs to treat certain things differently. So let me give you an example. In the international arbitration world, it's a very unique world. It is very long.
[00:37:40] Guest: David Perla: So if a client or its law firm comes to us and says, you know, we think it's worth this amount of money, but we're confident it's going to settle in two years, unlikely that model is going to render an outcome we like, because it's a much longer duration type of of an industry. The international arbitration, when you're arbitrating against sovereigns or under treaties, just takes much longer. Another area where the models had to account for differences is in the patent world, the appellate rate. The percentage rate by which things are reversed on appeal is higher in patent than in other areas, and everyone knows that going in. And so you have to look careful and say, well, is this more or less than we normally see in patent? If you saw the model for a patent case, if you saw the same numbers come out of the model for a breach of contract case, we would be much less optimistic about that breach of contract case. We would say, why is the risk on appeal so high? Whereas in patent it's just a built in risk in the patent world that things get reversed and then they get then they have to get, you know, there's multiple lines of appeal. And so we're used to that. We would not like those numbers. We would not like that outcome in a different field of law.
[00:39:00] Host: Paul Barnhurst: Fascinating. I think I could probably go on for quite a while asking questions, but I know we're nearing the end of our time. Like I said, you're the first person we've had in legal finance on the show, and I found it really fascinating just to learn a little bit about it, because obviously this isn't something I've ever modeled or even ever really even thought about until you reached out and, you know, mentioned coming on the show. So this has been great to just learn a little bit about it, a little bit of I how you think about things. And I have a couple, you know, kind of questions I just want to get your thoughts on. Fortunately, you did not get the long list of typical questions I ask in rapid fire for a modeler like circular references, yes or no VBA, I figured I'd spare you all.
[00:39:40] Guest: David Perla: Thank you.
[00:39:41] Host: Paul Barnhurst: You're like yes, thank you, but I will ask a couple. So I want to get your thoughts. Do you think Excel is ever going to die?
[00:39:46] Guest: David Perla: No, I have this whole long answer plan. But no, I think I think Excel is here to stay. I think people actually like it. Um, they know how to work with it. I think they know how to write code and macros around it, and AI knows how to help them. So I think I think we're with Excel for, you know, I'm 55, so at least my lifetime.
[00:40:06] Host: Paul Barnhurst: I had one person respond with, yes, it just can't be in my lifetime, you know, like, don't make me learn a new tool.
[00:40:14] Guest: David Perla: I won't live long enough to see it.
[00:40:16] Host: Paul Barnhurst: So and just in general, all you know. Will I build the models for us in the future? Do you think you get to that point? In general.
[00:40:25] Guest: David Perla: I think so. I think we're there, by the way. I think it's building models today. What it's not doing is building the models in a way that lets us see what it did to make sure they're built the way a modeler would build them. Yeah. So there's not that confidence because AI is by definition a black box, right? An LLM is trying to predict what it thinks you want. It's great for a speech because you don't care how the speech got written. Even today, most people don't care if it's plagiarized. But for a model where you do need to know what's underneath it, there's not that confidence level yet, but I think it's building plenty of models today and it's going to get better and better. And I think you'll see some hyper specialties where for particular areas, once people have that confidence, they're going to layer on top of the model. I mean, we used it. We built a web scraper for something and we hired a student to do it, and she used AI to help write the code. But we could see the code. We weren't looking for an original work of authorship, so why wouldn't we let it write complex models as long as we have confidence in them?
[00:41:32] Host: Paul Barnhurst: Yeah, I think you make a great point. It's confidence, the black box. It's, you know, getting the point where we can see all the pieces that go into it. There's definitely areas where it can take something and predict out or do you know, some modeling and things and it will get better with time. But typically what we get most people say, yes, last one. And then we'll wrap up here. Do you believe financial models are the number one decision making tool for corporations?
[00:41:57] Guest: David Perla: So my short answer is no. I think financial models are an indispensable tool in decision making for corporations, without which the corporation is going to be at a massive competitive disadvantage and therefore ultimately be less successful than its competitors. But at the end of the day, I don't think it's the number one tool. My suspicion is that the number one tool is probably still the human brain, at least for a few more decades. While businesses require strategic decision making and they require judgment in a way that the model can't yet get there. Do I think it's indispensable? Yes. I mean, we just wrote a job description that three years ago, prior to GPT, would have taken hours and hours of numerous people's time. It took me eight prompts and then a little bit of editing, and it's largely done and it's really good. And I'm sure someone, not me, but I'm sure someone can use it to actually look through resumes and pick the ones that facially meet the criteria. Yeah. Will it be broad enough? Probably not. But I don't want in any way diminish my hope and my optimism about models and about artificial intelligence, merely to say, I don't think it's the single tool that we can rely on.
[00:43:22] Host: Paul Barnhurst: Sure, you're not the first to said that we get differing opinions. I love to hear what people think, so I appreciate you giving that view and some context on it. That was the last question I had. So just kind of wrap up. If someone wants to learn more about, you know, your company, maybe get in touch with you, what would be the best way for them to do that?
[00:43:41] Guest: David Perla: Number one, you can always write to me at https://www.burfordcapital.com/ You can just go to Burford Capital. That's our website name of the company, and there are also email addresses on there for who to write about specific things. But the website is a wealth of information also to sort of help understand how it works and how we think through a claim or set of claims. Most people who call us have gone on our website to learn first, but by all means they can reach out directly as well.
[00:44:12] Host: Paul Barnhurst: Great. And next time we have you on, I'm going to give you all the modeling questions. Expect, you know, study up and we'll get you all geared up on modeling. You can come to Vegas with me for the financial modeling world championship.
[00:44:24] Guest: David Perla: I'll come to Vegas with you, and I'll bring one of my quants to give me the answers to the questions.
[00:44:31] Host: Paul Barnhurst: Yeah, it's your quants that should come to Vegas. I watch, I don't participate, but thank you so much for joining me. I enjoyed having you on the podcast.
[00:44:40] Guest: David Perla: Thank you. This was a pleasure.
[00:44:41] Host: Paul Barnhurst: 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 www.fminstitute.com/podcast and use Code Podcast to save 15% when you enroll in one of the accreditations today.