In this episode of Financial Modelers Corner, host Paul Barnhurst welcomes Ruby Liu, Director at KPMG Australia’s Valuation Team, to dive into intricacies of financial modeling. Having led KPMG Australia’s Sydney modeling team and worked as the head of corporate finance in the energy sector, Ruby’s vast experience spans from transactional modeling for infrastructure assets to advising clients on valuation and capital raising. Ruby Liu brings over 16 years of professional expertise in corporate finance, financial modeling, and valuations.

This episode offers practical insights on managing complex financial models, optimizing processes, and the challenges of transitioning models from transactional to operational use. Ruby’s guidance provides a valuable learning opportunity on how to build efficient and user-friendly financial models. Her deep knowledge and hands-on experience make her a key voice in the field of financial modeling.

Key takeaways from this week's episode include:

• The key differences between transactional and operational models and how to adapt them for long-term use.

• Why keeping the model user in mind is crucial for building effective financial models.

• Ruby’s top strategies for creating models that are simple, visually appealing, and efficient in calculation speed.

• Insights into the challenges of working with large, complex models and how to manage issues.

• How Ruby’s experiences at KPMG shaped her understanding of the financial modeling landscape.

Here are a few quotes from Ruby Liu:

• "The most efficient model isn’t the one with the most complex formulas, but the one that’s easy to understand and use." - Ruby Liu

• "In financial modeling, the goal is to fit the user’s purpose, not dazzle them with complexity." - Ruby Liu

• "I think AI can help build models, but we’ll still need human oversight to interpret the results." - Ruby Liu

In this insightful episode, Ruby Liu offers a masterclass in financial modeling. From her candid reflections on the pitfalls of overly complex models to the importance of keeping the end user in mind, Ruby’s practical advice is invaluable for both novice and seasoned modelers alike.

Follow Ruby Liu:

• LinkedIn - https://www.linkedin.com/in/liuruby

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 www.fminstitute.com/podcast and use the code “Podcast” to save 15% when you register. 

In today’s episode:
[01:13] - Introduction to the episode and guest
[03:01] - Worst financial model and lessons learned
[08:28] - Ruby reflects on her evolving perspective on the industry
[12:47] - Ruby’s leadership experience at KPMG’s Sydney team
[16:47] - Insights into Ruby’s current role in valuations
[22:14] - Different valuation methods; income and market approaches
[25:57] - Converting transactional models into operational ones
[41:16] - Rapid fire questions segment with Ruby
[47:49] - Advice for modelers and contact information

Full Show Transcript

[00:01:13] Host: Paul Barnhurst: Welcome to Financial Modelers Corner. I am your host, Paul Barnhurst. This is a podcast where 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, and that is why I completed the Advanced Financial Modeler earlier this year. I'm thrilled to welcome to the show, Ruby Liu. Ruby, welcome to the show.

[00:01:50] Guest: Ruby Liu: Hi, Paul. My pleasure to join you today.

[00:01:53] Host: Paul Barnhurst: Yeah, I'm really excited. I know we've took a little while for us to connect. We've both been busy, so I'm excited to have you on here. So let me give a little bit of background about Ruby. Ruby has over 16 years of professional experience in corporate finance, valuations, and financial modeling. She previously led the Sydney modeling team at KPMG Australia. She also served as the head of corporate finance for an electricity transmission business. Today, she is the director in the valuations team at KPMG Australia. She has successfully managed and delivered valuation and transactional modeling projects for large infrastructure assets, both in Australia as well as internationally, receiving positive feedback for the work she's done. She currently applies her unique expertise in corporate finance valuation modeling to advise KPMG clients on capital raising. Investment management, financial analysis and modeling. So that's a little bit about Ruby. Ruby. First question: we're going to jump into it. What's the worst financial model you've ever seen?

[00:03:07] Guest: Ruby Liu: Okay. It's actually we want to receive a model for our valuation analysis. And the model itself the size. It's over 100MB. And a lot of the calculations the model is using is actually via macro. So you can guessing it's a big black box. And because of the size and the macro driven and you found from normal user perspective it's extremely painful. So the size make the model. Every time we were joking about you can hit open the model and we go grab a coffee and then come back. The model will open. And then with some analysis we are doing because it's very macro driven. So people need to spend a lot of time to understand the code first. Then we can understand how to run certain analysis. We want to do. So in addition of those two things, and there's a lot of like math formulas in the model as well, which is definitely not helpful. So basically we are using a very high spec laptop, particularly for this model analysis. And every time using the model, it's not a pleasant experience.

[00:04:22] Host: Paul Barnhurst: That sounds awful. A lot of user defined functions are just. Yeah, no thanks. I'm not a big VBA guy. I think I would have just closed it and been like, all right, somebody else's problem.

[00:04:34] Guest: Ruby Liu: Yeah, we all wish that was the case as well when we're using the model. Yeah

[00:04:38] Host: Paul Barnhurst: I bet I inherited one and I won't. I call it a model loosely because it really wasn't a financial model. It was a data dump. The fed had churn analysis and the person had kept years of history and they saved the file is in XLS, not an xlsb. And it was one gigabyte. And they're like it opens pretty quick on my gaming machine. We had a 64 gigabyte gaming machine and I'm like, okay, if you have to have a 64 gigabyte gaming machine just to open this model, you got a bigger problem. Yes, it was over one gigabyte.

[00:05:13] Guest: Ruby Liu: Wow. Okay. Well, clearly I'll tell this story to my team to make them feel a little bit better because 100MB compared to one gigabyte, it's not too bad.

[00:05:25] Host: Paul Barnhurst: Yeah, but you had a lot of macros. I didn't have any macros. It was mostly just a ton of data and some basic formulas. So it wasn't complex. It was just they were saving way too much history. They were trying to use it like a database was the biggest problem.

[00:05:38] Guest: Ruby Liu: Got you.

[00:05:39] Host: Paul Barnhurst: But yeah, what I saw that I was just like, are you kidding me? What have I got myself into? So with this model that had a lot of user defined functions, 100MB, a lot of VBA. What was the learning experience? Or maybe the takeaway from seeing somebody build a model that way?

[00:05:56] Guest: Ruby Liu: The biggest one I would say, is a lot of times people thinking we build a model to make, what are we trying to do? Like basically meet the purpose. Like we get a calculator, right? We get the analysis we want to do in the model. That's a good model. But actually I found the most important thing is a good modeler should always keep the model user in mind. You just need to remember just because you get all the calculation correct in the model and in the way you're thinking, it's most efficient. Best. It really doesn't mean it's a good model because you need to considering who are the user. So that's probably the biggest experience I actually learned from this model is being on the other side. And now as a model builder and a modeler, then on the different side. As a model user, I really appreciated how important when the model. Build a model and thinking about. Okay, so that's probably the biggest experience I. Learn from this experience. You know, you just need to be thinking about the other side. When people use your model, will they find easy to use. And they will find you know. It's efficient and meet their purpose instead of the modelers purpose.

[00:07:12] Host: Paul Barnhurst: And that's what I'm constantly having to remind myself. I have a tendency to be like, oh, I can. Build that formula, no big deal. And then you're like, yeah, but does somebody want to have to figure out what I just did?

[00:07:22] Guest: Ruby Liu: Exactly.

[00:07:23] Host: Paul Barnhurst: Yeah, I think that's something that you start to learn more and more with experience early on. It's kind of like, oh, it's kind of cool to build this complex and it works. Look at me. And then you realize no simple is the way to go.

[00:07:36] Guest: Ruby Liu: Absolutely. And you found, like, I found some people, you know, especially from the mathematics degree sometimes exactly what you're saying they found using dynamic formula or some, you know, unique, normally used formula is actually quite cool. But honestly, as a good modeler, the hard part is how to use a simple formula to make your job done instead of use some really funky. And you know, people don't know formula to do the job. So that's absolutely it's very important for good modeler to thinking about that.

[00:08:12] Host: Paul Barnhurst: 100% agree. The more simple you can make it the better. And as I like to say, keep it simple because your business partners and your users will complicate it no matter how simple you keep it. So let them do all the complication for you.

[00:08:26] Guest: Ruby Liu: Exactly.

[00:08:28] Host: Paul Barnhurst: So I'm curious what interested you in finance as a career? What made you choose this kind of career path?

[00:08:33] Guest: Ruby Liu: I always like number when I was a kid, and we've always been told, you know, you probably want to do the things you are good at. When I grew up, you know, I constantly get recognition about my math, good scores, etc. and then when I started going to the uni, I chose the major. I was thinking, well, if I'm relatively good in math and at the same time, you know, I think finance is quite cool industry because we can use money to make money. So I think let's go with finance. But that's the young Ruby. Now I was thinking about actually, finance industry is nothing like what I initially imagined.

[00:09:17] Host: Paul Barnhurst: Not surprised.

[00:09:19] Guest: Ruby Liu: But yeah, but that's where at the beginning I was brought me to finance because I always thinking finance is quite amazing industry. You know, it's not like electricity or, you know, commodity industry. They actually have a tangible asset. Finance is really people, you know, using money to make money. So how to do that well and still maintain a good process and governance in the industry. I find it's actually quite intriguing and very interesting. And as we can see already, you know, because of the finance industry like global financial crisis, you know, all these things, you actually have quite big impact to the whole world. And the, you know, 19th, the.com bubble. So I still think in this whole industry is fascinating. And I probably just learned the scratch of it. But again, it's not as simple as oh, we just make money by simply use money.

[00:10:20] Host: Paul Barnhurst: Be nice if it was. But no, definitely a lot more complex than that. And when you mentioned the global financial crisis, I was in grad school at the time. So I remember, you know, Lehman Brothers and Bear Stearns collapsing and thinking, am I going to be able to get a job when this is all done? I was there at the 2006 to 2008 and we didn't. We had an investment club where somebody had put up an endowment, and we were actually investing this money when it hit a certain part. It would be set aside for scholarships. And I remember we did our presentation in the investment committee and we were like, this is the best presentation we've ever had. It's clear you guys learned a lot, but we never want to go through a year like this again because we did not perform well. You know, we, I don't know what we lost, you know, 20% or whatever the value of the fund was. And the funny thing is, the undergrad students made money, and it was all because they picked a stock that got acquired and went up like 200% in the acquisition. And, you know, they didn't have any other winners. They just kind of got that one high risk stock. And it was really interesting. It was a good learning experience. But yeah, my stocks did not perform well.

[00:11:26] Guest: Ruby Liu: Yes, I actually have quite, probably a little bit skeptical view about share market. I found like, you know, unless I have a lot of time, like a massive time to monitor the market on a daily basis, or I have a lot of money, I can potentially make the market. Otherwise, even for myself. Interestingly, as a finance person, I have very limited investment in share market because firstly, I don't have a lot of time. And secondly, unfortunately I also don't have a lot of money to make the market.

[00:11:58] Host: Paul Barnhurst: You know, it says I think Warren Buffett has said it really well. He's like, unless you're, you know, going to really study the market and you think you can beat the market, just invest in an index fund, invest in the whole market and let it grow and move on. You're going to get good returns. You're not going to hit home runs, but you're also not going to lose everything.

[00:12:18] Guest: Ruby Liu: That's definitely a good advice, especially for people. Just want to get a relatively stable return instead of, you know, make a big money overnight 100.

[00:12:28] Host: Paul Barnhurst: Yeah, we'd all love to have invested in Microsoft or Apple or whoever when they started. But the reality is, you know, there were ten people that invested in the other companies that we don't even know the names of anymore.

[00:12:40] Guest: Ruby Liu: Exactly. And you may lose your whole investment. You know already by now. Yeah. 100% agree.

[00:12:47] Host: Paul Barnhurst: Well, I want to ask you a little bit. You worked as the head of the Sydney modeling practice. I'd love for you to talk a little bit about what that experience was like. Maybe what that entailed.

[00:12:58] Guest: Ruby Liu: Yes, that one is actually, I really appreciate KPMG gave me that opportunity because when the first time they talked with me saying, do you want to leave the Sydney modeling team? I was actually quite flattered because I was like, wow, okay. I didn't expect that. I was thinking I just joined the team, you know, as a senior, more senior member. So I found it quite interesting experience. So firstly, I have a lot of freedom to shake the team culture, and I really enjoy that. We actually have around nine people, so it's not a big team, nine people, including myself and our average age, probably around 27. And I constantly joke very young team. And at that time I was 32 and I constantly being told I was the one dragging the team's average age down. I'm like, thanks very much, guys.

[00:13:56] Host: Paul Barnhurst: Nothing like people letting you know your age. They'll do that.

[00:14:00] Guest: Ruby Liu: Yes. But I had a really good experience with them because they are very eager to learn and we have quite a healthy pipeline. So we do both model audit and model build. So, I actually gave them regular training I think once per quarter training. And we definitely played hard and definitely work hard. So you can imagine the big four culture. We always work hard, but we also play quite hard like I think we did team building event every quarter. So we went to do some laser tag, you know, escape room, and we ran the boat, you know, do a one hour trip around Sydney Harbor. But at the same time, you know, after 10 p.m., you will see our team still working in the office. That's before Covid. So overall, it's a quite good experience for me. And I made some good friends as well during that experience time. And also it really helped me to basically build some rapport with the client as well because KPMG have quite good network. So we able to build in a lot of different models for different clients. That's where I realized because before I joined KPMG, I was more focused on transactional model. But then during this experience, I also, you know, being able to expose to a lot of different type of models, like operational model, consulting model, and sometimes I will even quote a model. But it's. They say, hey, I want you to help build a model and understand what the requirement I was thinking. This is just one of analysis tab, you know. So overall it's a very, very pleasant experience for me. I learned a lot. And I also get a lot of friendship and knowledge from that.

[00:15:54] Host: Paul Barnhurst: And there's a few things in there I love that you really said one is being willing to play and work hard. You got to have fun, especially if you're in the office after ten at night. You know, valuation, investment banking, some of those fields can be sometimes feel like they're 24 over seven when you have a deadline coming. Right. Really long hours. And if you don't build a good culture, it's the burnout is going to be even quicker. There's still a high risk of burnout. Anytime we work a ton of hours, there's always that risk. But the more you enjoy the people you work with, the more you have a healthy culture, the easier it is to thrive or enjoy an environment like that?

[00:16:32] Guest: Ruby Liu: Exactly. Yes.

[00:16:34] Host: Paul Barnhurst: So I can understand that. And I was going to say your activities sound fun. I want to come to the escape room and laser tag. And although I think it would be a little bit of an expensive trip for me to get out there today, you support valuations. You work in the valuations team.

[00:16:49] Guest: Ruby Liu: Correct.

[00:16:50] Host: Paul Barnhurst: What is your role? What is your primarily doing there? Are you building the models today? Working with the clients. Talk a little bit of what all that entails.

[00:16:58] Guest: Ruby Liu: Yeah, it's actually slightly different as what I used to work in the modeling team. So in the modeling team I was the lead in Sydney and it's a relatively smaller team and we are mainly building models and doing the model audit. But right now in valuation it's a larger team. And in Sydney we have almost 25, 25, 26 people, I think, and we also is a national team, same as the modeling team. Anyway. We are also national team and what I'm doing now is we still managing projects team on each job. But mainly I will be more like, I have to say, more executing instead of, you know, lead the team. But I still quite enjoy because it's very different experience and the evaluation. We don't really view the model anymore and definitely no model audit because that need a special credential. But we actually use model on a daily basis because most of the evaluation has a model underlying to support our evaluation. And I will lot of analysis. So we still have model. We basically still live with model on a daily basis. And using the model like we more like a reading the client's model and based on client's model to run analysis. That's where it really helped me to understand the other side, the model user side, you know, to thinking about, oh my God, this is how we feel when we have a good model or bad model.

[00:18:33] Guest: Ruby Liu: And so that's the modeling involvement. And then for valuation we doing mainly three types of valuation in the high, high level. For the umbrella saying one is for the listed company. If there's a transaction triggered the list company, go private or there's a merge or split. We're doing a valuation as an independent expert. And two is the infrastructure valuation. So basically for large fund they invest into unlisted infra fund or infra asset. Then you engage us as independent valuers to value those unlisted assets. And this is where we see a lot of like model from the underlying the infrastructure asset. We just come from the transaction at the beginning then the transition to the operational model. And then the last one is purchase price allocation which is more accounting valuation. So that's probably the key three valuations we are doing at the moment. But as I mentioned you know it is quite fun and quite different experience. But at the same time we still use model on a daily basis.

[00:19:48] Host: Paul Barnhurst: Yeah. It sounds like you're seeing a lot of different things and probably gives you an appreciation, especially for what's good. And those little disciplines that sometimes, especially if you don't have to have a model audited or you're not in a formal area, it's easy to be like, oh, I'll just quickly do this because I'll remember it's there, or I'll build this and you know, the next person's like, why did they do that? As you pick up the model, like driving me crazy.

[00:20:15] Guest: Ruby Liu: Tell me about it. That's the one. First thing we see, as I mentioned earlier, once I received the model from the large file transfer because you can't send your email. Before I even open the model, I look at that 100 megabyte size. I was like, is this for real? You know.

[00:20:33] Host: Paul Barnhurst: I totally believe you. That's how I was when I saw the one I did that was, you know, a gigabyte. I was like, are you kidding me? This can't be real.

[00:20:41] Guest: Ruby Liu: Yes. And when I was in the modeling team, I'm very proud to say my team never built any model reach that size. Not even close, you know.

[00:20:52] Host: Paul Barnhurst: Yeah. No, if you're getting up, I would say over probably 20 megs. You need to be rethinking what you're doing. Ideally it should be under ten. There are times where you have a lot of data, okay. Manage it via a database or Power Query or, you know, pull it in separately because you get a big enough data with enough columns. It's going to take up a lot of space. Outside of that, it's being efficient in your formulas a lot of times, and your formatting, which is just good hygiene in my mind.

[00:21:21] Guest: Ruby Liu: 100% agree, and I think 20MB is really the best practice. Sometimes the model, if they structured really well with the proper formula use, it can still running smoothly once you exceed 20MB, but I honestly never seen any model can still running well once they reach 50MB. That's my personal experience. Yeah, but 20MB 100% agree with you, Paul. I think that's more like the best practice from the modeling side. Yeah.

[00:21:51] Host: Paul Barnhurst: Yeah, I mean, definitely the smaller you can keep it, the quicker it's going to open and run, especially if you have people on older versions of Excel or anyone using 32 bit versus 64 bit. I mean, obviously machines are getting much quicker and as we go forward, they'll be able to handle more and more. That doesn't necessarily mean we should do more and more.

[00:22:11] Guest: Ruby Liu: Exactly.

[00:22:14] Host: Paul Barnhurst: I'm curious, kind of stepping back to the transaction models and doing valuations. You mentioned purchase price allocation. Different things. Are they mostly DCFS or do you do a lot of, you know, comparables because it sounds like you got splits, acquisitions, you know, going private, a lot of different type of transactions. Is there a most, most common method you're using for valuation or is it typically multiple different approaches in these models?

[00:22:42] Guest: Ruby Liu: Valuation. Always time to as long as we can with the available information sometimes which is very unique situation. You really lack of information. But normally we tend to use one primary method. And then we cross-check the value with the second method. So in valuation mainly have three approach. So the first one is the everyone knows the income approach. So you will see the method like you mentioned the DCF. And sometimes you know the earning capitalization like you know and a dividend models. Those are all in the income approach. The another one is the asset approach. Also they say the cost approach. So it's basically about, you know, from the net asset perspective to value the business. And the third one is the market approach. So like the one you mentioned that use the comparable companies or comparable transactions, just basically trying to get some evidence from the market as a reference to help us to get the value. So I have to say, based on my personal experience, the most frequently used approach is income approach and then market approach. Second, we use asset approach from time to time, but it's not that commonly used. But definitely it's also very standard valuation approach as well.

[00:24:10] Host: Paul Barnhurst: I would imagine asset is a little bit more for unique cases. Most of the time it's income and Into multiples or using the multiples to cross-check the income. Exactly that type of thing. That. That's what I've seen on the few. I worked on a couple models for M&A, some very small ones in my career, and that was usually the case is building some kind of DCF and then checking some multiples against it to see what we thought the valuation would be.

[00:24:38] Guest: Ruby Liu: Exactly. Yeah, that's quite common. We used to approach DCF, and then we use the market approach and the comparable method to like you mentioned, you know the market multiple to do the cross-check.

[00:24:50] Host: Paul Barnhurst: FP&A guy here and as you know I am very passionate about financial modeling and the Financial Modeling Institute's mission. I have been a huge fan of the FMI for years, and I was super excited when they decided to sponsor the Financial Modelers Corner. I recently completed the Advanced Financial Modeler certification and love the entire experience. It was top notch from start to finish. I am a better modeler today for having completed the certification. I strongly believe every modeler needs to demonstrate they are a qualified financial modeler, and one of the best ways to do that is through the FMI program. Earning the accreditation will demonstrate to your current and future employers that you are serious about financial modeling. What are you waiting for? Visit www.fminstitute.com/podcast and use code Podcast to save 15% when you enroll in an accreditation today.

[00:25:57] Host: Paul Barnhurst: So I'd love to talk about one thing, right? You build a transactional model, and I think anyone who's built both transactional and operational models knows they're quite different, right? Transactional are usually a one time event for the model. Operational. I come from FP&A. That's where I spent my career Budgeting and forecasting. You're always updating those models. And so there is a lot of thinking about how do I roll things forward every single month, you know, for years versus one time, you know, how do you think about that? Because I imagine a lot of the transactional models and the customer comes back and okay, now how do we make this operational so we can see how we're performing against whatever the transaction was, whatever that value was, they assumed. So maybe talk a little bit about that.

[00:26:42] Guest: Ruby Liu: Yeah that's actually quite interesting question. Cool. And again this is only my personal opinion. Sure. And I actually found this very interesting because as I mentioned before, I also building transactional models in my earlier careers for investment banking side, you know, when you worked ten hours at 2 a.m. and the model, the need to finish all the updates for the next day. You really don't. Thinking about much about how can I make the model running fast? How can I chose the best formula? You really just want to thinking about, let me finish this one, get it done and then I can go to sleep. So transactional model a lot of times at the beginning it's always building in a nice structure, well built. And then towards the end of the deal with time pressure, with last minute change you will see transactional model. Most of the time it's not built for ongoing usage. It's more like for the transaction and locked up the financial close version. And then for the bank and the people to referring to for the future payment schedule. But unfortunately, a lot of times I find people actually directly use transactional model for the operation purpose because it's handy. You know, we have a transactional model with order, forecast, order and tax arrangement order situations.

[00:28:18] Guest: Ruby Liu: We will discuss during the deal and all the assumptions already in the model. So it's very handy. We can use the model and normally it's quite complicated model already, and we can use the model easily for our ongoing valuation and ongoing operational, you know, management. So a lot of times the deals model directly being used for the fund managers for the operation in the future. And then you realize, okay, the transactional model situation changed the business. After one year, two years, things changed, the arrangement changed. And you will see some different model. Obviously not going to be the IB modeler, coming in based on their understanding of the model. Make some change or amendment from here or there. And potentially you know this person may resign from this company and then a new modular comes in and then thinking, oh, the business changed again. We need to add something or edit something again. And then they come in and then they do another tranche. So once we receive some operational model firstly you can tell oh this is definitely a transactional model because it's very standard build up structure for the transaction. But then here and there you will see some amendment clearly can tell from different modular and some of the section, the link has been dropped and been told, don't worry about that part.

[00:29:50] Guest: Ruby Liu: You know, we never use that part. And the output you will have a lot of like random editing sheet with the output because they need to build a report for the board meeting and the board reporting, which has nothing to do with the initial investment Decimal output, and then you will have some random shit added in for their output building. So overall it's very messy. No, very well structured and with a lot of different modulus signatures on the model. And trust me, there's always linkage dropped you know. And then if you go to the name manager you will see there's so many ref error in all in half of the names etc.. So interestingly, the model still working, but honestly, we all run our high level analysis to make sure direction wise the model does not go crazy, but I will be not be surprised if the model auditor comes in. They can still identify hundreds of things. The model can be either correct or, you know, doing well. So that's probably the key things. You know, I think I feel once I start to see the model transferred from transaction to the daily operation.

[00:31:11] Host: Paul Barnhurst: Yeah, the worst models are the ones that you pull data from a lot of sources, and they keep all those named ranges from all the different sources, and you end up with thousands of hashref and you're just like, okay, where do I start? Can I just delete all these? What's going on? I've seen some I had one that I had to start over. I tried running a macro, I tried opening. Wouldn't even open. It wouldn't run the macro against all of them. I had so many in it. I finally just had to give up and start the model over because I had so many. I'm like, I'm done, let's just rebuild this thing. 

[00:31:43] Guest: Ruby Liu: Tell me about it. And another thing is the format cell format. A lot of times, you know, the finance team or accounting team, they have some underlying data source and people just copy and paste data directly. And sometimes you open the cell format. You will see it like this. Once I open a model which is actually known in KPMG, in some other employees I used to work with. I see two languages like, you know, like in addition to English, there's another two languages in the, basically in the cell style, you know, and the cell style. I have to scroll down for a while to see all the different styles. And I was like, do you guys know the cell style? Also, it's memory of Excel as well, so it can go quite crazy. And I'm sorry. You know, if I can talk this, I can really talk about this nonstop.

[00:32:40] Host: Paul Barnhurst: I'd like to get your thoughts. Do you think, you know if you have a transactional model, you know there's going to be operational? Is your advice to just rebuild a separate model for operations? Or do you think it makes sense to try to take that transactional and make the changes, knowing all the challenges, the broken links, the issues you see? What would you recommend? In most cases, do you think it's better to just, hey, let's build an operational model after the transaction is done. So it's really designed and the thought goes into, hey, updating it all those monthly things. Or do you think most of the times it's better to just, you know, modify that transactional model? Do you have a preference.

[00:33:22] Guest: Ruby Liu: As a modeler? You know that, Paul. You know my answer probably already. As a modeler, I always like to build a model from scratch because it's much easier. Like, you know, sometimes client will say, hey, Ruby, you know, you have a model. Let's just do some quick amendment. It will be easier for you. I'm like, no. Trust me, it'll be much easier for me to rebuild one from scratch instead of understand the current model or linkage and make sure everything is still correct after my amendment. So that is definitely my preference. Although, having said that, I also understand from commercial perspective timing and money and effort is always an issue. So sometimes for because even when we're building an operational model, you need to fully understand the transaction model before you build an operational model, because the transaction model is support the M&A and hopefully have the best logic at the time for the targeted business. So if you finish a transaction within a year, you need to rebuild an operational model. You need to install, you know, use transactional model as the reference point and then building your operational model tailored for the finance team's ongoing usage purpose. Basically, I would say if you have the time and effort, it's definitely well worth to rebuild that one from scratch using transactional model as a reference. Otherwise, you can use the transactional model for operation. Now, if your business, like the business you acquired, doesn't change much from the acquisition date. But normally after two years I almost always suggest people to rebuild one from scratch. Yeah, because normally the business will change after the transaction.

[00:35:19] Host: Paul Barnhurst: Makes total sense. There's enough changes that after a certain amount of time, you're just better off. But I get it. There's also the economic, the commercial, the time. I just need something quick and dirty. Customer doesn't want to pay a lot, whatever it may be, that you're like, all right, we'll just figure it out.

[00:35:35] Guest: Ruby Liu: Exactly. Yeah. But I would say I see some transactional model on a surprisingly quite good and quite well you can probably use for the first year or even second year. But honestly, after a while, the target company changed a lot as well. It's really much easier. And actually also in clients Klein's benefit review because you need to make an informed decision, and if you can't trust your model, then how can you make an informed decision to help yourself? Right. So fundamentally, it still will be the benefit to the client. Yeah.

[00:36:11] Host: Paul Barnhurst: We're supposed to trust our models. I'm getting you know, you've obviously built a lot of different models, cover different industries. Do you have a favorite industry that you like, model or maybe a favorite deal type that you like to work on?

[00:36:27] Guest: Ruby Liu: Wow, I actually will say that's a trick question. And it's actually a difficult question. I actually found one thing I really enjoy in my work at the moment. It's the variety. So I can get exposed to a lot of different industries. And as you said, seeing a lot of different models, doing a lot of different valuations. I probably don't have a favorite Road industry. But my most experienced industry probably will be in electricity industry. And the more popular one is renewable. At the moment, because of my work experience and because of that one. So I have a lot of knowledge of the industry. So when I'm building a model or reading a model from this industry, I found it's quite easy for me and quite quick. Having said that, for the deal type, I personally don't like PGP, the public private partnership deal. I quite enjoy infra because it's my background. I always find the deals type. It's more like, the pleasant experience is not really about which type, but about the people you are working with. Sometimes I work with really experienced investment bankers or very, very good lawyers. I think a lawyer is very important in the transaction. And I found, you know, just to make things so much easier compared to, you know, sometimes, you know, you find like a man, you know, you need to meet in the middle ground with us. You know, you can't just say, this can't be done, you know, as a lawyer or as the legal area. So you need to help us to thinking about how can we make the deal work. Middle ground. Both parties feel comfortable. So this type of things I found is the key things to make my deals experience instead of the particular type. Yeah. I'm not sure if I answer your question.

[00:38:36] Host: Paul Barnhurst: No, you did that. That's helpful. And it sounds like you really enjoy the variety. You have some deals that you don't like. I get it, it's it's a difficult question to answer. It's always kind of fun. I get all kinds of answers when I ask that. I ask it from time to time. Yeah. I had one person goes in Bennett. He's actually from Australia. I don't know if you know him, but his comment was hotels. He just really liked, like, modeling hotels. He thought they were very straightforward. And I thought that was interesting. Yeah. You get all kinds of different answers. We're going to move on to a little bit more, what I'll call kind of the standard questions in a minute here. We're going to go to rapid fire. But I have just two more quick questions I want to ask before that. First, what's your favorite Excel shortcut. If you have to pick one, what's that one you like or maybe use most you can define favorite however you want. 

[00:39:24] Guest: Ruby Liu: I will say two. One is the shortcut. It's a shortcut. It's easy. It's the I don't even know how to say the shortcut. I have to use the shortcut on my keyboard to remember. This is so natural now. So it's when you hit the shift control and to the right and the shift control to the down. So that's where you carry all your formulas to the control key and the Ctrl R. So that's where you basically can carry your formula very quickly from the left to right and top to down. Those ones are my favorite and I probably use the most. Yeah.

[00:40:00] Host: Paul Barnhurst: So yeah, I've heard a few people say that I'm not surprised and I laugh. And you said you have to do it. I worked with a guy who spent a long time in investment banking, and he told me at one point he'd be sitting in meetings after he'd done a ton of model building, and he would just be kind of automatically, without even realizing what his finger is doing. Different shortcuts. Yes. Kind of be tapping on the desk because it was such just built in memory.

[00:40:22] Guest: Ruby Liu: Yes. And the shortcuts actually are very useful. I would recommend anyone want to do modeling. Learn that in the modeling team I used to train my team, it's I take over their mouth just like they don't have mouths. And I just like if I see you guys using mouse, I'll hit your head. Hit your head. But then later I found they are using shortcuts much better than me. Then I'm like thinking, oh dammit, you know, you get so much better than me and much quicker now.

[00:40:51] Host: Paul Barnhurst: I'd probably get my, not probably, I'd get my hand hit a few times. I've got better but I still occasionally will be playing with that mouse.

[00:40:58] Guest: Ruby Liu: Oh no, I'm using mouse this time as well now because I don't build model that frequently. So yeah. 

[00:41:05] Host: Paul Barnhurst: That's my thing too. I'm the same way. There's definitely times I'm like, I'm just using the mouse, but there's times when I'm trying to hurry or doing something. I'll be like, all right, let's use more of these shortcuts. So I'm a little bit of a hybrid.

[00:41:14] Guest: Ruby Liu: Yes.

[00:41:16] Host: Paul Barnhurst: All right. I think we're gonna move in here to our rapid fire question. So let me set how this works. I don't know if you've heard this before, but you can't say it depends, because you can answer that to all these. You have to pick one side or the other. Okay. So kind of a yes or no. And then at the end you can elaborate on 1 or 2. Because the reality for most of these is you could answer it depends. The idea is to give a quick answer. So I'll give an example here and then we'll go through them. If I was to ask you circular or no circular references if you had to pick one, which one would you pick?

[00:41:48] Guest: Ruby Liu: No. Circular. Right.

[00:41:50] Host: Paul Barnhurst: And so that's it. And then at the end, if you want to elaborate on a couple of them that you're most passionate about, we can do that because I realize we can always find an exception to almost everything. Right? It's kind of how it works.

[00:42:02] Guest: Ruby Liu: Absolutely.

[00:42:03] Host: Paul Barnhurst: All right, so you gave me this circular or no? We know: No circular. VBA or no VBA in models?

[00:42:10] Guest: Ruby Liu: No VBA.

[00:42:11] Host: Paul Barnhurst: I figured after your example of worst model, you would say that. Do you prefer a horizontal or vertical layout to your models?

[00:42:18] Guest: Ruby Liu: Horizontal.

[00:42:19] Host: Paul Barnhurst: All right. Dynamic arrays in models. Yes or no?

[00:42:23] Guest: Ruby Liu: No.

[00:42:24] Host: Paul Barnhurst: All right. External workbook links. Yes or no?

[00:42:27] Guest: Ruby Liu: No.

[00:42:28] Host: Paul Barnhurst: You said that with authority. Most people do. They're like, please just don't do it. Named ranges. Yes or no?

[00:42:34] Guest: Ruby Liu: Yes.

[00:42:35] Host: Paul Barnhurst: All right. Do you follow a formal standards like fast or Smart or some of those standards that are out there when you're building your models?

[00:42:44] Guest: Ruby Liu: Yes.

[00:42:45] Host: Paul Barnhurst: That's what I figured. If you work for one of the big, big four there. Will excel ever die?

[00:42:51] Guest: Ruby Liu: Oh, maybe. I know you said I can only say yes or no. I really this is a tricky one because I was thinking in my lifetime. Let me put it this way. In my lifetime. No.

[00:43:04] Host: Paul Barnhurst: I've had more than one person that says that. Or yes, but just don't do it in my lifetime. I've got that one as well. Do you think AI will build the models for us in the future?

[00:43:15] Guest: Ruby Liu: Yes.

[00:43:16] Host: Paul Barnhurst: Okay. Where do you stand on sheet cell protection? In your models. Should you use it? Yes or no?

[00:43:22] Guest: Ruby Liu: No.

[00:43:22] Host: Paul Barnhurst: All right. Next one. Do you believe financial models are the number one corporate decision making tool?

[00:43:30] Guest: Ruby Liu: Yes.

[00:43:31] Host: Paul Barnhurst: You hesitated a little bit on that one.

[00:43:35] Guest: Ruby Liu: The reason for that is because model is a tool. So it's garbage in, garbage out. So I would definitely think your model is a number one tool making decision, but it's a tricky one to say because sometimes people can make a right decision with a model. Some people can make a wrong decision with a model.

[00:43:56] Host: Paul Barnhurst: 100% agree. Yeah. It's not a simple one, I get it. And then the last one. What's your favorite lookup function? Do you like choose Vlookup, index match, Xlookup or something else?

[00:44:09] Guest: Ruby Liu: Index match for sure. But these days Microsoft have the new formula lookup I found it also, I use some time and index match is the most one xlookup yes, sometimes as well, but the index match is the most common one used.

[00:44:26] Host: Paul Barnhurst: That's what I find with most modelers. Index match is probably number one, followed by xlookup and then you get a few others choose. I've had people say offset, others like I don't care, just use whatever works. I get a little bit of everything.

[00:44:41] Guest: Ruby Liu: If people use offset for lookup okay. Offset self is a dynamic formula. Interesting.

[00:44:46] Host: Paul Barnhurst: It's volatile. So if you're using it all over your spreadsheet it's going to take a while.

[00:44:50] Guest: Ruby Liu: Exactly. Yeah. So that's the way I was thinking. That's interesting to hear.

[00:44:55] Host: Paul Barnhurst: Yeah, it was interesting. I mean you can do it, but it's definitely not my first choice.

[00:45:00] Guest: Ruby Liu: Yes, it won't be my as well.

[00:45:03] Host: Paul Barnhurst: I figured as much. Is there any of those you want to elaborate on a little further? You know, kind of what you're thinking was.

[00:45:09] Guest: Ruby Liu: I honestly thinking it's probably already mentioned, like AI built a model for us in the future? That's an interesting question. I didn't say yes because I seen how powerful, you know, the generative AI working these days. And the GAI is mainly working on the unstructured information at the moment. And we both know modeling is a very structured data processing. I think it's because in the past people have trouble to process unstructured well. So the generative AI is more focused on processing the unstructured data at this stage. But given how powerful I think the AI these days, I actually have no doubt AI eventually can build a model for us. But I still have the strong belief. Doesn't matter at which stage. I really hope you know the scientific movie situation will not happen. We're not going to be dominated by AI, so as long as human beings still dominate the animal, I think, you know, AI can build model for us, but we still will require human beings to overseeing the model building and the analyzing the model output to get comfort. That's probably my view. And also I'm thinking for AI to build a model for us at this stage, based on my reading, seeing and experience, I think still will be at least 10 to 15 years away time to get the really good, let's say, transactional model AI built for us, man. You know, even human being, it's it's very painful process, you know. So it's newer large data processing. It's more like the quick changing and also a lot of like finance, tax and commercial elements in there and changing all the time. And each case are different. So I found those are all the challenges to AI. But definitely it's a very interesting topic for AI and modeling in the future.

[00:47:12] Host: Paul Barnhurst: I agree, I mean very basic models that can build some stuff today. The more complex you get, the more you need the human involvement and you're always going to need the human. In my mind around the assumptions, the validation, the part that needs that judgment.

[00:47:29] Guest: Ruby Liu: Yes.

[00:47:31] Host: Paul Barnhurst: The more task based your model is, the more it's do step A, B, C, the easier it is going to be for AI to do it. And it's pretty amazing what it can do today. But yeah, the more complex the model, the more you're going to need the human involved for sure.

[00:47:46] Guest: Ruby Liu: I agree, 100% agree. 

[00:47:49] Host: Paul Barnhurst: All right. So just have two last questions and I'll let you go here. First one is if you could offer one piece of advice to our audience to be a better financial modeler. What's that one piece of advice you'd give them?

[00:48:03] Guest: Ruby Liu: Well, you probably know the answer already, Paul. It's keep the model user in mind.

[00:48:08] Host: Paul Barnhurst: I love that. And how do we do that? How do you really keep the user in mind? What do you found helps you do that?

[00:48:13] Guest: Ruby Liu: I used to say this three things you should keep in mind, which is a good model, needs to fit the purpose and know. Feed your purpose, feed the model user's purpose and then. Simple. It's very important. Don't thinking you want to dazzle the people for the model build. That's not the right stage for you. Simple. And I'm a very big visual person, so please make your model as pretty as possible, because sometimes I see people make the color look like a Christmas tree in there. One spreadsheet, one sheet. And like, let's try to make it a little bit more visually appealing, please. And the last one is calculation speed. I always think if you can make the model meet your user purpose by providing them a simple as possible model, visually appealing and then calculating fast, that will be good already. So that way you're building a how to structure the model, how to chose formula, how to formatting them. You keep those in mind, you will just help you to make the decision correctly.

[00:49:19] Host: Paul Barnhurst: Well said and I agree. Keep the calculation speed. Make sure it can calculate quickly. Keep it simple where you can visually appealing. We don't need a Christmas tree. We don't need a million colors. Even if you think it looks pretty, it's usually just distracting. Yeah. And then just keep your user in mind. Really think about what they're trying to solve, not what you think they're trying to solve.

[00:49:42] Guest: Ruby Liu: Exactly. 100%. Just listen to your user. And if they really want, they can try to be a user on one time and they will feel a lot of pain from the user, you know?

[00:49:54] Host: Paul Barnhurst: Very true. So last question. If our audience wants to learn more about you or get in touch, what's the best way for them to do that?

[00:50:02] Guest: Ruby Liu: Okay. They definitely can go through your channel. And also, you know, through my LinkedIn profile I would say.

[00:50:09] Host: Paul Barnhurst: Okay, perfect. We'll put your LinkedIn in the show notes. So if anyone wants to contact you there they can. But thank you for taking some time to meet with me. I really enjoyed chatting with you today and just talking about financial modeling. Appreciate the experience you shared and hope you have a great rest of the day.

[00:50:26] Guest: Ruby Liu: Thank you so much, Paul. Thanks for having me here. Definitely a pleasure and I also enjoyed our chat. If you ever come to Australia and see me, you know where to find me so we can grab a lunch and talk about modeling for sure.

[00:50:40] Host: Paul Barnhurst: No, I definitely will. So it's on my bucket list. I hope to get over there one of these days, so I'll look you up when I get there. Thank you again, Ruby. Really enjoyed.

[00:50:49] Guest: Ruby Liu: Thank you. Paul.

[00:50:50] Host: Paul Barnhurst: Financial Modelers 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 modeling. 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.