Ahead of the curve podcast

Implementing Open Source Risk Engine: Accessing online resources

Overview

Our host Devin Cook is joined by Alexis David, both Senior Consultants within our Quantitative Services team. They discuss the merits of Open Source Risk Engine (ORE) and explain the myriad of resources available to assist in implementing the software. They share nuggets of information which will be useful to anybody wanting to use ORE for the first time, or looking to expand their knowledge to understand the latest capabilities.

Listen to the podcast

Welcome to Ahead of the Curve podcast by Acadia.I'm Devin Cook, and I'm here with my colleague Alexi.Today, we're going to talk a little bit about ORE Academyand the growing library of resources within Acadiaand our Quant Services group. With that, Alexi, why don'tyou talk a little bit about ORE and just introduce a little bitfor some of our listeners who might not be familiarwith our open source risk engine? -Thanks, Devin.It's great to be in this beautiful, classic podcast room.ORE stands for Open Source Risk Engine, and it's a softwareproviding risk analytics for derivatives and structured productsfor the risk industry. It's based on QuantLib,which is another, very good open source library.Both of them are open source, and it means they're completely free.It was created in 2010.The founder of a company that was called Quaternion Risk Managementat the time, which was later bought by Acadia in 2021,and then, also in 2022. The founder of the companywas an avid user of QuantLib. During the financial crisisof 2007, they decided that they needed to increase the coverageof what QuantLib could do, and they decided to create ORE,Open Source Risk Engine. They did that for multiple reasons.The first one was to improve the coverage as a class of products.The time was needed, like credit derivativesand inflation products such as CDO, CDO Squared, ABS.Not squared yet. -All this complicated stuff.They also needed to increase the coverage of analytics,so they could cover CVA, value at risk, sensitivity calculation,backtesting, stress testing, et cetera. As you can imagine,all these features along the years have been greatly improved.Now we can cover hundreds, if not thousands,of different types of instruments in six different asset classes,multiple analytics, and we release our code every quarter now.For a couple of years, I think we've been we managedto do that every quarter. The last release last yearwas pretty successful. We released the American Monte Carlo,we released scripted trades, which is promising.We'll talk about it a bit later. Then there was a third reason.Probably one of the most important, why they decided to createQuaternion Risk Management and ORE was for a regulatory purpose,because they wanted to create a transparency frameworkwhere all the regulators could easily access all the code.I guess that's something we're going to talk about in a moment.It's amazing to me just to see how far ORE has come.Originally, it was this internal pricing tool used by that small company,and now it's grown into this extremely robust pricingand risk engine. What's so unique about thatis that the pricing and risk engines across the industry are not free.They're third-party black box vendors. That our product,which does all of the same things, and we consistently expand and maintain,is completely free open source. You can get it, absolutely.Whenever you want. We continue to expand and broaden our product scopeand coverage, although now with our scripted trade framework,which is a little bit less meaningful, but there arealways different risk metrics and things to expand.Part of the team has shifted the focus of ORE to usability,making it much easier to use because when it's an internal tool,it's not as important, but we would like to make it as easyto use externally as possible. Our goal is to makeit as easy to use ORE from PIP installing it or cloning the GitHub repo,to running one of our 50-plus examples, counting and our ever-growinglibrary of examples on both the standard ORE side and then the Python SWIG, whichwe'll talk more about later. With that increased focuson expanding our educational resources, expanding usability,you've played a huge role in that, so why don't you talk a little bit firstabout open source and how you see open sourcein the risk space? -Short answer,accessibility and transparency as well. Long answer,one of the first questions that the client asked when we demo ORE is.How come you're releasing all these features for free?It's not sustainable. Other companies have made their entire businessout of selling the same things exactly, butit's operating services. -Somehow, there's almostan adverse reaction when you hear it's free.It must be bad. It must not be that great.It must have all of these faults, or you must be tricking us in some way.Exactly. You're the product. -Genuinely, that's not the case.Oh, it's bad. It's neither.It's a great product that is perfectfor the users. You just have to download it.You can use it, you can break it, you can identify errors,and you can improve it. That's part of the community.We want people to do this. From a business point of view,any client that is using ORE as the source of calculating risk,for example, is greatly advantageous because they knowthat they have a software which has been usedand reused and broken and reviewed by the regulators.For the regulators, it's something that they already know.They know that they could have reviewed it in another company.They don't have to face a black box, which, a lot of the third-partyproprietary software is ORE. There are a lot of advantagesto being transparent. -Even aside from the user base of ORE,in addition to the user base of ORE, Acadia's SIMM engine is powered via ORE.You have, outside of our large user base, all of these tradesevery day, are calculated with SIMM calculator.It's also validated by large institutions across the globe,and this gets to the point that I think our mindsetwithin the Quant Services group is more aligned with the tech companyas opposed to a traditional old financial firm.This is how you see yourself. -Exactly. One of the coretenets of Quant Services is that open source is a mindset,not just our software or not just a software license.Pushing out our high-quality ORE code to get GitHubis only part of the job. Now we're seeing that the other part of the jobis democratising its usage and putting out resourcesfor others to learn it. Who aren't internal at Acadia,and don't use it every day and maybe in another bank.We'd like to grow this user base, whether that's with academicsor whether that's with institutions. Alexi, this is somethingthat you've been working on in the sense of growing a library of contentto help enable this. Would you want to speak a little bitabout your ORE Academy and the amazing workyou've done with that? -Yes. Your academyit's probably one year old now. We released that last year.It aims to be a library of content, library of mainly tutorial videos,explaining what ORE is, and how to use ORE.For now, we have a YouTube channel. Hopefully, we'll havemore medium of democratising this knowledge about ORE?ORE, for a very long time-- the way we developed OREwas really from a technological point of view.We didn't spend much time or much effort on marketing it or democratising it.That shifted when we were bought by Acadia and we hada little bit man resources. We had a very supportive marketing team.That was absolutely great. It started from there.We had a couple of videos, and all we needed was to add a few moreso that we have enough basic knowledge to provide to the viewers,so that they know how to install or know what ORE is.Install ORE, how to configure all the different filesrequired, how to use it, and then after we can use thatas a stepping stone to create moresophisticated subjects, that means likemore sophisticated analytics, et cetera. -Because usage of ORE can be as simpleas pricing an equity option to doing a more complex XVA or PFE.There's a whole breadth of potential content there,as far as learning in our current content is already extremely robust.We have anything from, just like you said, installing ORE,which now is relatively simple, to a whole enterprise-wide infrastructureof ORE, and how one might implement that. Then, of course, we have pricingand risk content in between those two.The libraries are already pretty robust, but would you want to talkabout what your big plans are, what the next stepsof the ORE Academy are? -There's a lot of stuff to do.We want to increase the coverage both internally,so people in the company as well know about how to use ORE.It's not always obvious, especially for potential userswho might find ORE interesting. We have a lot of ideas.I don't know if we're going to succeed in providing them.Some of the ideas, some of the subjects that we want to tackle,we're talking about more sophisticated subjects where,for example, bootstrapping or CVA value at risk.XVA, MVA scripted trades, which is probably oneof the most promising features that we released in the last year,that greatly increased our coverage of the product almost independently.That's pretty great. I'll give you a concrete exampleof what we're going to propose very soon. That's a work in progress at the moment,but we want to do videos that explain the subjectof the derivatives world to people and not necessarily showcase ORE,but use ORE as a medium to explain those concepts,which are sometimes very difficult, and you need numbersand graphs to explain plane better. We want to do a videoabout parametric value at risk. We are going to split it into four videos.It was supposed to be one video at the beginning,and then the more you dig, it's like a rabbit hole, the more you do.The first video will be entirely dedicated to OS bootstrapping.That's one of the core knowledge that our students learn,it's very important. That involves knowing howto price in OS instrument that involves thatto generate cash flow schedules, especially if there's a payment lag.That involves a log-linear interpolation, boots and root routefinding algorithm, et cetera. The whole thing.Then the second party will be dedicated to equity volatilitycalibration. The third one on sensitivity calculation,and the fourth one finally on value risk and how to use those sensitivitiesto calculate risk. That's a work in progress.Hopefully, we'll be able to provide that very soon.I think that's extremely interesting. That's a really interesting projectthat showcases every step of ORE. -That's what we want to do here.You'll be able to validate these numbers in many different ways.There are also a few more ideas that I can throw you.One is throwing some tipsabout things that gravitate around ORE, it's not directly related to that,gravitating around ORE, there's one about how todebug ORE in Visual Studio. Potentially, we should do oneon explaining how to generate the documentation,which is in LaTeX and how to generate that into a PDF.Sometimes, I remember trying to do that myself.It can be very difficult. -I would say it's difficult justbecause our documentation is so robust and we have a structured approachto our documentation. That's why we have many different productsand all the different risk metrics that we're able to run,and that's part of the reason it can be complex at timeswhen dealing with the documentation. It's not necessarily a bad thing.It's a good thing. -One thing that we could potentiallylaunch as well, like another, we could launch a new playliston the YouTube channel about the derivatives glossary,like definitions explaining some of the regulatory termsthat are quite Sacher or like CVA,all this stuff, it might get ORE. I think it will be very interestingto explain those in very short and focused videos.I think that'd be perfect. I would personally find somethinglike that very useful. -I would learn a lot as well.A lot of those regulatory terms you can't find in a glossary like that.Having a short snippet, it's just like a quick videowhere you learn a little bit and maybe learn its relation to OREand how to-- -It was a formulaand some examples that were great. -That would be perfect.I look forward to a future. -Let's see.There's a lot of work to do. Hopefully, that will manage all that.I think everyone's looking forward to further development in the ORE Academy.Thanks a lot for your work and for leading that initiative.Another thing I wanted to talk a little bit more about,and I think is almost equally as important as the recent Python implementationof ORE. Especially, this is important because I cameout of college knowing Python, not C++. In this case,it would have been a little bit difficult for me to upstart without ORE.Now, many graduates, much like myself, come out of the box knowing Python,and they can get ORE very easily now and start working with itwithin a matter of seconds. Would you want to talk a little bitabout the Python Swig and a little more detailon that than I gave? -Yes. The Python extensionis along with the script, which is probably oneof the most fundamental features that we released.Like you said, everybody uses Python nowadays.Now with the PIP install, there's one line to install itthat is compared to installing all the stuff in C++and in Visual Studio that it is amazing. The way it worksis we use something called Swig, which is a fairly common wrapper.Which is also based on Quant. Like I said.It's using Swig. There are multiple advantages of OREusing Swig to export its functionalities. First of all,in comparison to a library that would have been built completely100 per cent in Python, it's way faster, robust and powerful.Because the back end is C++. -Which is faster.That's one reason why it's great. The second is the way Swig works.It works in two steps. The first step is it takes your function,your class in C++, and export it into an interfacewhich is language independent. The second step is transformingthis interface into, for example, one of the languages available like whichis for example Python. That means that all the functionalitiesthat we exported already into Python. It will be very easy to expand itinto other languages like Perl or C#, et cetera.It's pretty powerful. Like I said,we listen to the community. For now, it's not extensive.It's enough to work, to launch ORE and run some of the analytics.I think we want to export way more of it. That's probablywhat people are expecting of us. We need to release those functionalities,and we need to cover them through the ORE Academy.That's one of the next steps. We listen to what people want.If some functionality takes priority, I think we will listen to you.That's exactly right. If any listeners have any ideasto reach out, and Alexa will make a video for you,and hopefully we'll get-- -Somebody else.One of the other many members of our team. We've talked a lot about education,usability, largely in relation to the open source risk engine,and how that's evolved over time. A lot of these ideascoalesce in a recent partnership between Acadiaand Northeastern University. Northeastern University,famous for its co-op approach to education, hands-on learning.We've partnered to do a certificate program wheresome of the best professors in their quantitative finance programand some of our top partners and original builders of OREcan teach students about quantitative finance,the financial regulatory environment, and, importantly, usingORE, all three components of the quant finance space.I think that's important, and I'm excited for that internship,or for that partnership. If anybody is interested in that,feel free to take a look. It's exciting. -I met some of the team membersof the Northeastern University, and they're very competent people.They're very nice, so pretty excited to see where this could go.It's a good mix between university and Acadia.I think so, too. Would you care to share, Alexi,where you can find the ORE Academy, and our listenerscan look at these resources? -Yes, it's on YouTube.You just go on YouTube and you search for ORE Academy. That's it.Thank you. Alexi. -Thank you, Devin.Thank you for listening to Ahead of the Curve podcast by Acadia.If you need more information,go to acadia.inc.

Also available on