Ahead of the curve podcast

Risk Reimagined: The Rise of Open Source in Finance

July 15, 2025

19 min

Overview
More episodes

This episode dives into the evolving regulatory and modelling landscape in the UK for traded risk. Scott Sobolewski speaks with Xabier Anduaga and Joey O’Brien about trends in counterparty credit risk, model validation and the growing adoption of the Open Source Risk Engine (ORE). They discuss how firms are replacing legacy vendor models with ORE to gain transparency, reduce costs, and improve governance. The team also introduces the Risk Analytics Lab—a hosted ORE environment designed to accelerate adoption and support benchmarking, stress testing, and validation use cases. A must-listen for anyone navigating today’s risk and regulatory demands.

Hello and welcome to Ahead of the Curve,the podcast from LSEG Post Trade Solutions.My name is Scott Sobolewski,Co-Head of the Quant Services group within PTSand today we're here to discuss modellingtrends and regulatory trends within the UK market.Joining me today, I have two local colleagueshere in London.Our newest hire here,Xabier Anduaga, Senior Partnerwithin our expert services consulting team,and Joey O'Brien, a Senior Consultanthere, also based in the UK.So to kick things off off, Xabier is a new faceon the podcast and, Xabier joins us from PWC UK,where he's spent the last six yearsdeveloping their traded risk and modelling capabilities.He has over 12 years of consulting experiencein the financial institution and quant spaceand before he got into quant finance he completed his PhD in physics at CERNand worked on the Large Hadron Collider.So probably a topic for a different podcast.But Xabier, let's talk to you first.Can you tell us a bit about your journey to LSEG?What led you here, and kind of whatsparked your interest in quant finance?Yeah, sure.Thanks for the kind introduction.As you said, I've been in the consultingspace for over 12 years.That gave me the opportunity to work on a wide rangeof engagements.I would say probably covering the whole model lifecycle.I started my career after that timeas a physicist, I started my careerworking alongside the front office quants for a large USbank developing derivative pricing models.Back in the day, SR 1107 was coming out,and the bank was working onkind of a multi-year engagement to basically buildpricing models across the set of asset classes.We're still to this day,feeling the effects of thosepost-crisis regulations in the form of stress testing.We'll talk a bit about that here today in terms of theBank of England stress tests and Basel III, continuationof the immediate post-crisis Basel II regulationsthat came out, around the time you startedyour career and mine as well.So it's been just about three monthsthat you've been with LSEG nowleading the local market here in the UK.Could you share your initial impressionsof the quant services teamand your impression of,the quality of the folks that you've had a chanceto meet and work with here within LSEG?Yeah, I've been having a great time so far,and I'm actually if I need to point out,maybe one or two things about the team and the firm,I'm really quite impressed abouthow knowledgeable our team is, how strong our quant team is,and I think not just interms of technical expertise, but also the teamculture, how we all collaborate with each otherto kind of deliverthe best outcomes possible.So I think that's quite powerful.So in terms of the firm,I think,the type of engagements that we are doing at the moment,they are challenging.They are kind of quite complex,and I found that quite exciting.So yeah, so far so good. Here at LSEG, those of youwho are previous listeners to Ahead of the Curvemay have some familiarity with theOpen Source Risk engine,the modelling frameworkthat we use to support clients,on a commercial basis, both for our hosted riskand pricing calculation services,but also facilitating the client's use of,that open source software locally,being part of LSEG has if anything,turbocharged the client adoption that we see,from the Open Source Risk Engine.Probably a coincidence,but one of LSEG's slogans 'open to possibilities'has open right there in the nameand we've been a big beneficiary of thatas a small quant services team,having the weight of LSEGand and not only the weight behind, awareness,of the Open Source Risk Engine, but also usage and adoptionof ORE internally here within LSEG,we'll talk a bit about that, a little later on.Joey O'Brien here also, local to London.Could you give usa bit of a flavour for what sorts of,client projects and experienceshave been keeping you busy to start the year?Sure. Thanks, Scott.So our main focus really this year so farhave been in two directions.The first one is in replacing legacy vendorsoftware in groups,and that is the case where they have realised thatthey perhaps don't have as much ownershipas they'd likein these kind of black box vendor models,and they realise that they're doing thingslike pricing, stress testing, sensitivity,CVA calculations, and they havea different library for each one of thoseand each of those libraries come with heftypremiums of time, but also just the factthat they don't fully own them has become a problem,and they've started to realisethat they can usethe Open Source Risk Engine or OREto actually replace all of those things.So we've had a number of projects at the momentwhere we've been actively doing that hand-holding clients through implementingORE locally and making it the risk library of choice.So that's been a big focus at start of the year.The second big focus, as usual over the pastnumber of yearshas been in assisting in model validationand I think particularly here in the UK,there's been a renewed focus on that.So you guys mentioned SR 1107.That's been what, 15 years now.So there's been a push in the US to do these things.The PRA actually have also put out the SS 123which is a similar kind of document,and now the banks here in the UKare feeling similar pressuresto form full blown model governance and validationon a frequent basisand we're definitely feeling the benefit of that.Yeah and the Open Source Risk Engine is very versatilein the way that clients can deploy it,either as the primary modelling infrastructureor within a model validation team,being able to use production-like software,but for benchmarking purposes. Xabier, I think that ties in very well to your previous experienceat PWC, where you were doingquite a bit of validation style projects.Could you just talk a bit aboutyour view of the state ofthe current traded risk environmentaround the UK and what sort of, clientfeedback you've gotten nowtalking to them about ORE in particular?I would say there is,there is a big focus in the UK,probably also globally around counterparty credit risk.I don't think thatcomes as a surprise,considering we saw at the end of last year,the publication from the Basel Committeeof a new set of guidelines forrobust management of, of counterparty credit risk.This comes on the back of the lessons learnedfrom all the events that took placethe last few years, including Archegos,with what happened with Credit Suisse after that,SVB, the gilt crisis in the UKand also the PRA listed CCR as one of the key priorities for 2025.So that's quite consistent.I think the industry,although there's a lot of stuffthat has been done post financial crisis,I think now there is, after what happened,clearly the institutions realisedthat they need to do more,and so we see institutions working toenhance their overall CCR frameworks,including better systems, better infrastructure, better models.We see, for example things like intraday ratesthat some institutions are seeing now that more as amust-have rather than a nice-to-have.Stress testing.You mentioned it in the introduction.Stress testing I think is now becoming a clear componentof the risk management framework,as opposed to a purely regulatory driven exercise.Yeah and I think, the Open Source Risk Engine providesa unique ability to, assist on the quantitativebenchmarking side of a lot of these projects.So, Xabier, maybe you could speak a bit about kindof what differentiates us when we come intoa project, working with clients.I know the open source realm is unique in that, we like to bring our expertise to clients,but we also learn quite a bit when we workwith different clients in different types of firmsand in the open source realmthe majority of our open source users are very happyto contribute thatwork back into the open source realm and sort of enhancethe standards thatthe entire global modelling infrastructure,the tools that they useto support their local requirements.So in your experience, how can that helpclients on a typical validation project?Yeah. I mean, you mentioned it, right.The modern risk function has been under kind ofcost pressure for many years now.So it's quite difficult and expensive to maintaina proprietary library just to do independent testingand some institutions are actually not doing it because of thatand I think ORE can play a key role there.Where we see that effectively the model validation functionget access to a complete library for pricing,XBA and risk where they canvery cost efficient way to run their ownindependent testing,and I think that's something that is kind ofworked very well received by by the model riskpeople we've been talking to.And our goal when we work with clients,on consulting projects is to make them experts in OREand truly pass that ownership baton over to them.Most of our projects are short term, and they're one off,and it's really about assistingwith that initial setup and implementationand then over the course of a project,facilitating that knowledge transferso that they can own and operate OREas if it's their own internal model. A lot of times clients use the ORE codeand they'll call it, you know, their ownproprietary risk engine when it's, in effect,the open source code operating under the hoodand that helps providethose clients with a lot of thosecost synergies that that we mentioned.Joey, in a typical validation style project,maybe you could just frame what that looks likefor the average client, end-to-end andwhat we bring to projectsand then what clients bring to completion.So we start any validation project really with,first of all, introductions to what the model actually is.What products does it cover?What are the client's viewson its limitations and justificationfor using that model along with its business use caseSo getting that kind of introductionand that will really steerwhat the rest of the validation is going to do.From there, we in some sensetake over in leading what happens from that point.I view it really as two separate streams, right?There's a qualitative side and a quantitative side.The qualitative sidereally dives under the hood of why they chose that model,what that model's assumptions are,how does it line up withindustry best practice,and we're quite fortunate in our positionthat we do have a good sensefor what the industry is using different models for,as a result of our hosted service.So we can also opine on what we think best practices areand if they're missing out,and that qualitative component involvesthings like tiering,which both of those kind of regulatory documentsrequire firms to now doto justify how often models are revalidated.So that's the first big component.The second is the quantitative sideand that's really where I thinkwe're in a unique position because of theOpen Source Risk Engine.What we have to do is perform an independent validationor benchmark exercise versus things like valuationsand sensitivities.What a lot of firms would do in those cases is implementa spreadsheet of sorts,or a Jupyter notebook or something,and do it on the fly.What we can actually do is use this fully fledgedrisk system that we have that has been in some sensevalidated itself by the industry.So performing that exercise of doing a benchmarkof their internal valuationsand sensitivities is a big part of our documentationand by the end, what we tend to dois bring those two strands together,qualitative, quantitativeinto this kind of comprehensive document,along with all of the supporting codethat we use to produce the reportand then we perform a hand off, really give themthe comprehensive document at the end with both sides,and allow them to take ownershipof using ORE in the future,using those inputs for future revalidations.I mentioned LSEG turbocharging OREto some extent.We're two years into ORE and the quant services teambeing part of the LSEG family,coming to LSEG through an acquisition of AcadiaSoft in early 2023,and we are now subject to LSEG'smodel risk management policies here internallyas a large financial institution.So ORE itself, you mentioned it's been validated by dozens,if not hundreds of clients by now, by their own local teams.ORE is also in the process of being validated end to end.All of the pricing models, all the risk simulationmodels by LSEG'smodel risk management team,which provides users anadditional layer of assurance and governancethat those models are working to or above, market standards.Just as any other LSEG model that you're using across the firm,including within the data and analytics business.So Xabier, I understand you came to LSEGwith some previous ORE experience.So your three or four months in, but I thinkyour experience with ORE is a bit deeper than that.What do you find most exciting aboutORE at the minute in terms of addressing relevant eitherregulatory or modelling trends here in the UK market?Yeah, no that's true.I've been a big sponsoror support of Quantleaffrom probably over 15 years ago when I discoveredORE, maybe three or four years ago,as I was working on an FRTB research paper.Something that really stood out for meis that I rememberback in the day 15 years ago,how complex was to work with Quantleaf.It required a good amount of workto familiarise with the library,to kind of be able to do some kind of complex analysis andI must say, my recent experience with OREhas been completely different.I managed to do a full installation of OREfrom scratchwith all the dependencies, run the examplesin under 90 minutes on my own laptop.So how that journey is now so much easier in terms ofthe adoption of ORE is quite powerful.But I think the one functionality that I love about OREis the fact that within one platform, you can do so much.So once you spend a bit of workto configure the library to bring in the market datato set your trades, then you can pretty muchdo a lot of different kind of risk metricswithin the same kind of infrastructure,and that's I think is quite powerful.Yeah and I would say so far the feedback we've gottenon ORE, despite vast array of documentationand even a YouTube channel that walksthrough various examples and use cases,there's still quite a bit of legworkin terms of setting up those inputs and all of the ITand infrastructure plumbing to get ORE working locallyon clients own market data, clients own portfolio data.One of the things that LSEGPost Trade Solutions has developedto help, achieve a quicker adoption of OREis actually a new productthat we're unveilingin the second halfof this year called the Risk Analytics Lab,which is essentiallya hosted instance of ORE with the market data inputsand for clients of Acadiaand Post Trade Solutions hosted services, even portfolio inputs alreadypre-configured on the back end.Post Trade Solutions isprobably most well known for uncleared margin rulescompliance, where we're calculating initial marginon non-cleared derivatives daily for over 150 clients,those same clients can have access now for the first time,to all of the additional analyticsthat ORE supports extendingwell beyond justthe scope of initial margin calculation and backtesting.So, Joey, maybe you could talk a bit more aboutthe Risk Analytics Lab and how we think it will helpsolve for some of those initial implementation hurdlesor timelines that clients experiencewhen they use ORE currently.Yeah, so when we move to OREit takes a long time really to get an initial setup.It can take monthsto over a year to move from initial testing to productionand that is, it's a hurdle, really.It's a hard thingthat a group has to invest in and that's a blocker fortwo groups, really.There's this very small groups who just don'thave the resources to be able to do thatand maybe they don't need a fully fledgedrisk system, they just have to domaybe an annual CVA benchmark or something.They don't need full production environments.Then you also have these big banks who of coursehave the financing and resourcing to do it, butmaybe are a bit doubtful about whether ORE is up to scratchand there are two big hurdles,probably the two biggest hurdlesin terms of uptake and initial testing of OREand what I think is part of the mainhurdle is just this whole setting upof the configuration that you mentioned.So how do they want the curves built?What about the market data?Where does the plumbing come in?And they are all blockers that do stoppeople trying ORE.I think we've realised that and have also realised thatwe've actually solved that problemas part of our hosted service already,so why not extend thatcapability to other types of analyticsand see the full power of ORE?And once we do that,I think it will,you know, remove that hurdle of initially playing aroundwith ORE, already seeing how the numbers are,maybe doing an initial POC, benchmarkingand giving encouragement to groupsactually fully blown,replacing their own systems with ORE longer term.I think if we're looking for an analogy here,it's probably like, a fine dining tasting menu.Really, that's what Risk Analytics Lab could be.People mightjust use the tasting menu, but they might also come backand try more and more if they enjoy itand I think that's what Risk Analytics Lab can offer here. I think particularly valuable,just tying it back to our earlier discussionon model validationmodel risk management teams who don't want to gothrough all of those hurdles to implementa production risk system purely for the purposesof occasional benchmarking or annual validations.The way that we've priced the commercial risk analyticslab service is very flexiblebased on the number of users and essentially justcharging back for cloud compute, based on the specific analytics use cases.So we think it'll be a very cost efficientalternative to, those teams needing to buy secondary,third party vendorsor investing the timeto create those models from scratch,within their own infrastructure.So I think with that,Thank you very much for listeningand to learn more about the Open Source Risk Engine,you can visit opensourcerisk.organd to learn more about Xabier, Joeyand the commercial services that we offeron the consulting side, you can visit lseg.com/posttradesolutionsto learn more about the teamand get in touch with our sales team.Thank you very much.

Risk Reimagined: The Rise of Open Source in Finance

Overview

Listen to the podcast

Also available on

More episodes

Navigating Complexity: FMS, Risk Management & the Power of Open Source

2024 Year in Review – Open Source Risk Engine (ORE)

Navigating Counterparty Credit Risk & NBFI Oversight