IDSB Banner

VARisk


"With Quantrix, we can support the complexity that our customers face."


        - Kyle G. Lundstedt, Co-Owner

 
Kyle G. Lundstedt is co-owner of VaRisk, Inc., an Oakland, CA-based firm that builds valuation and risk management tools for large banks and other players in the mortgage market. The company formerly used Excel as a base for its software tools, but searched for a new platform to increase reliability and transparency while reducing other constraints. He recently switched the company’s models to Quantrix Modeler, a multidimensional financial and quantitative software product developed by the Portland, ME-based Quantrix. Offered on both PC and the Mac, Quantrix fit Lundstedt’s needs. The following interview with Lundstedt explains the complexities involved in VaRisk, Inc.’s solutions, why the company outgrew Excel, the beauty of Quantrix on the Mac platform, in addition to other topics.

Q. What problems does VaRisk help financial institutions such as banks and thrifts solve?

VaRisk, Inc.: Valuation and risk management are two functions. When an institution – a bank or cash fund or other investor in mortgages – is trying to understand what’s going on in its portfolio of loans, mortgage-backed securities, etc., it needs to address two major questions: How much is my portfolio worth? And what is the likelihood that the valuation will change?

Q. How do you help companies solve those problems?

VaRisk, Inc.: We build analytical software tools for analysts and people who make investment decisions within these institutions. The software tries to realistically capture the way borrowers behave. We help people figure out how to evaluate their portfolio – i.e., what fraction of their mortgage portfolio is likely to default, how that will affect the portfolio, and other similar issues. We help them project realistic future scenarios of the economy, and provide tools for calculating and analyzing the cash flow of their portfolio in these scenarios.

Q. Why do you work on a Mac rather than the PC?

VaRisk, Inc.: There’s no question that Macs have been less prevalent in the business context, but I think that is changing over time. A lot of people have jobs where the computer is kind of a “necessary evil” – they only use it for basic things like email, web browsing, basic reporting, etc. But in the kind of work that I do, a Mac saves you time messing around with drivers, control panels, viruses, and other things. You get to spend a lot more time being productive. Macs are used heavily in life sciences and biotech, and valuation and risk modeling software is very similar in terms of computational needs. We need lots of processing power, lots of RAM, etc., to do Monte Carlo simulation, statistical re-sampling, and other numerically intensive work.

Q. How did you find Quantrix Modeler?

VaRisk, Inc.: Back in the day, I owned a NeXT computer, and one of the critical applications I used was Lotus Improv. That app was in many ways the forebear of Quantrix.

Q. What tools did you use before you chose Quantrix Modeler?

VaRisk, Inc.: We used Excel with a number of DLL’s or libraries underneath it, and a Visual Basic front end as a “wrapper” around Excel.

Q. Why the switch to Quantrix Modeler?

VaRisk, Inc.: There were a number of reasons why Excel was problematic. We’re trying to tackle multidimensional problems. For example, we want to calculate a number of different metrics (dimension 1) such as principal and interest for a large number of loans (dimension 2) month-by-month (dimension 3) across many different economic scenarios (dimension 4). We’re trying to solve a four-dimensional problem using the two-plus dimensions (rows, columns, and worksheets) of Excel. We were stuck trying to cram four or more dimensions into two dimensions, and therefore were constantly making design compromises. Moreover, we also found that Excel, with its formulas hidden in the cells, was a lot harder for users to understand than we expected.

Q. Compare Quantrix with Excel.

VaRisk, Inc.: Excel is not designed to be a component in someone else’s application – Excel wants to be the boss. Our VB app provided the user interface and controlled the libraries, but Excel was a separate process -- there were always two separate programs running. Any time there was a problem with Excel, our application would shut down gracefully, but Excel wouldn’t. It also was difficult to test and debug this kind of “two headed monster”. There were stability issues – if a client was evaluating 10,000 loans and one had bad data, it would kill the whole job. Our clients are not big fans of letting a job run for three or four hours, and then having the thing bail out.

We currently are rewriting our application to use Quantrix at its core. Quantrix has been designed as a layer cake – if you just want to use the bottom layer as a component rather than a standalone app, it works well. Quantrix plays nicely with others – it can be a part of your application. Our new version is a single, robust application with communication between front end, libraries, and matrices.

Another critical advantage is that, unlike Excel, Quantrix is multi-threaded. It will use however many processors you have on your machine. That is a big deal because the tasks we are tackling are computationally intensive.

Q. How important is the multidimensional nature of Quantrix for your application?

VaRisk, Inc.: There is no arbitrary limit on rows, columns, and most importantly, dimensions – RAM is the only limitation to Quantrix. If we want to organize a model in terms of time, Quantrix lets you focus on each element, and you do not have to worry about mapping the data and formulas already created when you add a new scenario. If you write a formula that applies to only two dimensions and then add a third dimension later, Quantrix will automatically apply the formula across the third dimension without you having to re-do a lot of existing work. That is dramatic – in Excel, you have to copy and paste a formula down a row or down a column, even if you have to do it for 1,000 columns. If your problem is multidimensional, then you have to copy and paste those 1,000 columns into another worksheet to view another scenario. You always worry whether you have correctly applied all of the formulas, many of which you can’t see. From a transparency standpoint, it’s night and day.

Q. Give me an example.

VaRisk, Inc.: Take the problem of amortizing a mortgage – you can build a spreadsheet in Excel very easily with three simple columns -- beginning balance, payment, and ending balance -- and make all future months be the rows. In Quantrix that would take, say, four formulas – a payment formula, one that calculates beginning balance, one for ending balance, and one that says what is month zero. In Excel, you’d have to put a formula in every single cell. For a 30-year loan over 360 months, you need more than 1,000 formulas for a simple amortization table. Do that table for 10 different kinds of loans – for a different kind of mortgage, different prices – where do you put that in Excel? You end up with 10 worksheets – that’s just ugly. What if you want to do it for three different interest rates? How do you add the dimension of rate to what is already a three-dimensional problem in Excel? You have to add on, and it gets messy. Once you have chosen your structure, you’re stuck – if your first four columns amortize for different rates, you have 40 columns, with four for each rate. If someone wants to see payments side-by-side for all 10 rates, you’re in trouble.

Quantrix separates presentation from logic – its dimensional nature is such that if you want to rearrange, and flip it so that rate is on the top, and balance is on the bottom, you can do it – it’s dragging and dropping. In Excel, it’s a multiple hour redesign every time.

Q: How do your clients like Quantrix?

VaRisk, Inc.: We’re betting our business on it. While we had a nice tool that was built on Excel, we were getting negative feedback from customers – they needed more transparency and flexibility, and our tool was too buggy. They were doing very sophisticated analyses, and invariably something went wrong. With Quantrix, we can support the complexity that our customers face.




idbs_banner_960x41
About Quantrix - Contact - Site Map - Refer A Friend - Support
Solutions - Products - Services - Community - Partners - Webinars - Learning - Store
© Copyright 2014 Subx Inc.-All rights reserved -+1.207.775.0808 -