The Python Podcast.__init__

Share on social media:

[00:00:13] Unknown:

Hello, and welcome to Podcast Dot in It, the podcast about Python and the people who make it great.   When you're ready to launch your next app or want to try a project you hear about on the show, you'll need somewhere to deploy it. So take a look at our friends over at Linode.   With the launch of their managed Kubernetes platform, it's easy to get started with the next generation of deployment and scaling powered by the battle tested Linode platform,   including simple pricing, node balancers,   40 gigabit networking,   dedicated CPU and GPU instances, and worldwide data centers.  

Go to python podcast.com/linode,   that's l I n o d e, today and get a $100 credit to try out a Kubernetes cluster of your own. And don't forget to thank them for their continued support of this show.   Your host, as usual, is Tobias Macy. And today, I'm interviewing Paul Stafford about building risk models to guard against financial exchange rate volatility. So, Paul, can you start by introducing yourself? Yeah. Paul Stafford. I'm the CTO of Diagla Partners LLC,  

[00:01:11] Unknown:

and we specialize in helping   corporates and funds manage their FX risk. And do you remember how you first got introduced to Python? It was probably about 3 or 4 years ago when I   was developing a number of algorithms for some corporates I was working with. And   back when I was at HP, I was working in c and c plus plus and when I started to look around, it was those seem to be on the decline and Python's kinda popped up as, you know, 1 of the more popular things. And what got me excited was that it was like all these free and open IDEs and tons of exciting libraries. So, yeah, it seemed like a obvious choice.  

[00:01:47] Unknown:

And so digging into the idea of risk management,   and I know that you're working in particular in the financial exchange space. I'm wondering if you can just give some of the overall principles involved  

[00:01:59] Unknown:

in risk management and some of the statistical methods and algorithms that are typically used for that space. Yeah. Sure. So I'm gonna start off by kinda giving an example of where FX risk pops up just to kind of people get their minds around it. So let's say you're an investment fund based in Brazil and you're looking for US investors.   So they're gonna send their dollars to you, they're gonna get converted to reais, they'll be deployed in country.   2 or 3 years later, those reais need to get converted back to dollars. In the meantime, if reais has depreciated against the dollar,   healthy returns can get chewed away or even completely eliminated by, you know, spot rates changing.  

So that's where the risk arises.   With corporates, it's kind of similar if you've got sales in another country and their local currency,   the revenues in that country, you know, when translated back to your reporting currency   can easily   not meet your metrics and expectations. And so managing that risk becomes very important for those groups.   In terms of   principles, I mean, the first thing basically is to look at quantifying risk, and and this is where things already kinda get funky because, you know, finance is full of ex   physicists and   math guys that tend to apply   real world rules to finance, you know, which is not driven by real world, it's driven   that sort of thing. And so, basically,   when you create a model where, like, spot rate can use model with a normal and a variance, then, you know, you can use things like far value at risk, which basically   looks at the worst case of how much you could lose with a certain amount of confidence.  

And another 1 which kinda compliments that is something called expected shortfall, which is   all of the tail risk netted together.  

[00:03:51] Unknown:

I know that there are also a number of other domains for risk management. So the most obvious 1 would be in the sort of insurance and underwriting case or, you know, where you have actuaries who are determining, you know, what is the probability   of some certain, you know, event happening that is going to require us to actually pay out this insurance rate, but then there are also other elements of risk. And I'm wondering how much of the kind of modeling and mathematical aspects of those risk calculations are   universal across the space and how much of it is specific to a given domain that you might be working in? I would say insurance is a completely different domain, you know, because basically,  

[00:04:31] Unknown:

you're looking at rare occurrences.   Whereas if you look at,   you know, say,   selling or buying a stock or exchanging currencies,   the risk is evident every single day because the rates are changing in FX and, you know, stocks are going up and down every single day. And so the ways that you model those risks are quite different than than an actuary would. In terms of your involvement in the FX market,  

[00:04:56] Unknown:

I know that you started your career as a scientist working with NASA and then moving on to HP. I'm wondering   what your   motivation was and your path from the more sort of hard science and engineering space into the financial markets and some of the ways that that background in engineering   has benefited you in this new domain.  

[00:05:20] Unknown:

So how did I get over to the dark side? Is that the question? Can't say   More or less. Sure. Okay. Yeah. So, you know, it's funny about 12 years ago, a friend of mine, somebody we, you know, we call a high net worth individual   was speculating in currencies and he actually asked me about   how he might manage his risk when doing so. And that was the rabbit hole I plunged down, you know, it was and it was infinite deep and interesting and I never climbed back out. You know, looking at the tools, the derivatives that are available, how those derivatives are created and modeled and priced. It was actually there's this whole new world, but it was mathematically   based. Right? I mean, the black shole's equation and for options and   differential interest rates and such. And so it was fairly straightforward to get a grip on how the whole system was put together.  

On that risk management note, I basically decided to look at what I could do for firms here in Montana. Look at the Montana World Trade Center, reached out and led me to create my own advisory, which I cleverly named Currency Risk Management. Everybody makes fun of it, but it it did help with SEO. Yeah.   Yeah. You know, something so obvious. Right? Anyway, so and then you asked about background. So, you know, I think that, you know, having started off in, you know, in science and engineering,   it all kinda comes down to maths, basically, in particular linear algebra. You know, I mean,   there are concepts that you see in finance called PCA, right, principal component analysis. But when you dig underneath the hood, that's really just eigenvector, eigenvalue   compositions, you know. So they give funny names to stuff that we're all familiar with in linear algebra.  

You know, the other thing that I get to use a lot is vector calculus because I'm working with   volatility surfaces and exploring gradients and such to optimize option strategies. And so I would say that, you know, mass is probably the most important  

[00:07:10] Unknown:

tool to bring into finance and it prepares you pretty well for for managing all that. And then, of course, programming and modeling systems on top of that. Yeah. Just briefly keying off of your note about being good for SEO. That's essentially where I ended up with with my other podcast, the data engineering podcast, you know, that does what it says on the tin.  

[00:07:29] Unknown:

Exactly. Right? Everybody laughs at it, but, you know, I came top of the list in a lot of searches, so it worked out well.  

[00:07:35] Unknown:

Absolutely.   And so in terms of the specifics   of the sort of tools and mathematical principles, you mentioned a few things already with sort of eigenvectors, eigenvalues, and vectors working in linear algebra. But what are some of the   methods and techniques and some of the specifics as far as the the programming aspect that has been particularly useful in your career and risk management? I would say that, you know, the most  

[00:08:01] Unknown:

useful thing has been Markov chain Monte Carlo simulations.   You know, basically,   1 of the big things that we bring to our clients that   many other players in the space do not, you know, is   a a grip on all of the potential outcomes.   So for example, if a bank is gonna be   presenting an option strategy to a client, they'll show them a payoff diagram. It's like, if the spot rate is this, you'll get this much. If it's this much, you lose this much. Blah blah blah. But they never once talk about what the likelihood is, if any 1 of those spots actually happening. Right? And so you have no idea whether you're most likely to get a payoff or most likely to get a kick in the butt. Where a simulation   and you plot, you know, the distributions   of the outcomes, you know, in either a box plot or a violin plot or something like that, and you plot that against, say, 3 or 4 different strategies that banks have given you,   now you have a visual   tool to compare the efficacy of those against the unhedged exposure. Right? I mean, that's always a choice is to run unhedged, to not manage the risk. And depending upon the cost of the derivatives you're working with, that's always an alternative.  

So I would say that that's probably the biggest 1, and I've spent quite a while   building simulation engines that allow me to vary the volatility,   the skew to give it fat tails, you you know, kurtosis, for example. So we've got a lot of levers to pull   and   then able to simulate a lot of different types of markets, different types of dislocations   so that we can see how all of these strategies hold up under some real tests. Kind of the other tools   use a bit of classification tools like SGD and SVC.   Those are probably the biggest ones. I use, obviously,   all of the Python libraries.  

No scikit learn and scipy are probably the biggest ones that, you know, outside of the normal NumPy and Panda stuff that I use.  

[00:09:52] Unknown:

You mentioned the markup chain Monte Carlo. I'm wondering if you have used the pymc3   library at all for being able to build out some of those simulations.  

[00:10:01] Unknown:

No. I built my own before I knew that existed.   You know, the Python libraries are so extensive. I'm sure there's a library for everything. In fact, I just discovered 1 for astronomy the other day, which was pretty exciting, but did not know about that. And I thought actually after if you mentioned that that pi n c 3 is mainly about using Bayesian statistics. Yeah. So that is not part of what I'm talking about here.  

[00:10:23] Unknown:

In terms of the sort of mathematical approaches, a lot of what we're discussing   is sort of well known statistical methods that are sort of classically applied and, you know, theoretically can be calculated by hand, although it's going to be very tedious and time consuming.   But I'm wondering how recent trends in terms of the more widespread adoption of deep learning and deep neural networks have impacted the sort of capabilities   and   strategies around risk calculation   and being able to apply those to the foreign exchange market? Yeah. So that's a great question and kind of,  

[00:11:02] Unknown:

multiple bits of answers there. So to be honest, I've played with neural networks and deep learning only a little, and the reason is is that   their application in this   space would be in a predictive manner. So for example, you might   take a bunch of different macro factors, you know, like GDP and unemployment   and, you know, inflation and stuff, and see what kind of predictive power they have in terms of where spot rates might be going.   If you're an investment firm, that's a fine thing to do. But if you're a risk manager, that's a dangerous as hell thing to do.   Especially because   as we all know, you know, neural networks are impenetrable. Right? You can't explain how it got to its answer. Right? It just did. And so I've done very little of that. The main tools that I've been using are Bayesian regressions and constrained optimization because   typical things that I'm doing   are to, for example, identify a proxy currency.  

If you have a a currency like, say, the Argentinian peso,   which,   you know, is if you follow any kind of world events, you know, Argentina is kind of a mess economically and has been for, like, 50 years, and there's no derivatives, no spot rate, no, it's a disastrous market, but people still need to do business there. And so if you can identify   a basket of currencies that behave similarly, but have liquid markets and derivatives,   then you can substitute those and protect yourself.   And so those are the   approaches that I've been taking to this sort of thing. But, yeah, the predictive stuff is not in the risk managers toolbox.  



[00:12:34] Unknown:

In terms of being able to   accurately calculate and understand the risk that you're working with, obviously, you're going to need to have high quality and trustworthy data to be able to inform your decisions and inform the way that you are thinking about how to   mitigate the risks that you and your clients are taking on. And I'm wondering how you handle   sort of validating the inputs that you're using for these different statistical methods and being able to   sort of identify what are the sort of most important attributes in the data that you're working with. You know, obviously, you have the concrete data from the different financial exchange markets, but I'm sure that there are also   other external factors and secondary data sources that you're dealing with and just how you're able to synthesize that into a meaningful  

[00:13:21] Unknown:

risk assessment. Spot rates, derivative, you know, premiums,   historical spot rates, derivative,   you know, premiums,   forward points, all those sorts of things. Like I said, the   other data like macro data is   extremely hard to come by in a timely manner. Well, I actually looked at this a few years ago to see if I could do some sort of predictive stuff. You know, basically,   I was pulling down the kind of macro data that I mentioned, you know, a minute ago. And what I found was that it was very problematic, you know, it's delayed, number 1. I mean, you can't get   this month's GDP, you know, on any country. Right? You got to wait 3 months. And so it's highly delayed, and there's a lot of missing data, and they don't seem to give a shit about it. And   there's no way   I mean, I suppose that you could, you know, fill, you know, using some sort of an interpolation algorithm, but, you know, that is also fraught with its own issues.  

So from the standpoint of what I'm doing about the only thing that is   variable and then difficult to predict or find good data on is   the   price that you pay for various derivatives is opaque. It's what's called an over the counter market OTC.   And so, you know, the banks can charge whatever kinds of spreads they want, and that can be quite variable. And so if you   create   like, a rolling hedge strategy, like, say, you've got   a events that you're hedging against this 5 years from now, there are no derivatives that go out 5 years. So you have to hedge out 1 year and then a year from now roll it again and a year from then roll it again. And   knowing the costs of those roles is impossible. Right? You you have to do kind of a   statistical distribution on what bank spreads are gonna be and how liquid the market is gonna be at that time.  

So that's the data that's hard to find or hard to come by. But I don't deal with the macro data in our business.  

[00:15:17] Unknown:

Obviously, 1 of the sort of most unpredictable elements of financial markets and world markets in general is the recent pandemic, and I'm wondering how that has   influenced your overall sort of strategies or if you've had to sort of tune or update your algorithmic approach to understanding the risks because of the volatility and uncertainty that the pandemic has brought about on a global scale? Yeah. Sure. That's a great question. So number 1, you know, like, we don't try and predict, which is the good thing. But what we do when we are, for example, using our tools and assisting clients,  

[00:15:52] Unknown:

we   go beyond just the current levels of volatility   and neutral skew. So typically what I will do is take like a high low and mid volatility and a negative 0 and positive skew, and I'll evaluate   the strategies against all 9 different scenario. And that really points out weaknesses and strategies that work well in some market scenarios and not at others.   And especially in things like the heightened volatility,   which we certainly saw back in 2020,   that's always been, you know, 1 of the things that we look at and   that will affect   the strikes and such that we do when we are actually setting up the strategy. So we might tweak   based on, you know, initial simulations   under high balls. We might move around the strikes or use different options strategies   to be   more robust against high vol.  



[00:16:42] Unknown:

As far as the sort of development of those strategies, so you were able to use the spot markets and understand sort of what is the price of this currency at this particular point in time. And I'm wondering if you can talk through some more of actually   the sort of next steps after you say, you know, this is the potential,   you know, upside or potential downside of this particular strategy   and then being able to guide your clients in making those decisions. And then, you know, once they have made that decision, going back to understand what was the actual output and being able   to, you know, use that to   inform future decisions for risk mitigation.  



[00:17:21] Unknown:

We've gone through several cycles where, you know, we've been able to put put a strategy into place, watch it play out, close it out, you know, and get the p and l, you know, for both the exposure and the strategies.   And, you know, in general, what we've done is held up very well. You know, the only thing is is that conditions are always changing. So I'll give you an example. Last year during the pandemic,   interest rates dropped worldwide. Right? And so the   cost of various hedging strategies like forwards   were almost negligible. And so in a high vol   environment where the cost of hedging is low, you know, it was pretty easy to do no wrong. But then we close those out and now we're in 2021   and interest rates are rising. And so these strategies   for their rolling hedges or new hedges are changing based on the new market. So it's a constant evolution   based on,   you know, all of the different financial factors that come into play on on the derivatives market.  



[00:18:18] Unknown:

And, again, speaking in terms of current events, there's the question of rising inflation at least in the United States. And I'm curious how things like inflation within   different countries impacts the overall risk, the financial exchange   layer where, you know, you might have a certain amount of currency where the exchanges between currencies are at a certain point, but the actual purchase power of that currency within the bounds of that country is impacted because of the inflation rate. Right. Yeah. So inflation is a small piece of why spot rates move. It really is truly a supply demand question. So, you know, why does the dollar rise against yen or or euro or why does sterling rise against  

[00:18:58] Unknown:

South African Rand? You know, if a country's economy is very strong,   people are investing in it and buying their products, and so I need to exchange my dollars for, you know, name the country, you know, Rand.   And conversely, you know, if I'm selling my dollars,   then that depresses the market for dollars,   increases the market for Rand. So those are some of the forces.   The other forces are the interest rates, and this is where the inflation bit comes in. So   central banks around the world, you know, they've got several mandates usually, but 1 of them is to keep inflation, you know, in some certain band, you know, say 1 to 2%. And when it moves out of that, then they need to start pulling levers.   And so they will pull the interest rate lever as this happening in Brazil. Right? They've got pretty bad inflation and so the central bank is raising interest rates.  

What that does is generate   what's called the carry trade. So   investors   will sell   low yielding currencies and buy high yielding currencies   and make money off of the differential interest rate. And so now that now I'm buying reais that increases the value of reais, the spot rate, you know, moves in reais favor. So that's how   inflation   influences interest rates, which influences   spot rates.  

[00:20:17] Unknown:

Yeah. Lots of lots going on. Yeah. Absolutely. They're interesting and complicated beasts, and I'm wondering sort of what your   educational path has been moving from the engineering into the financial markets. You know, once you started going down that rabbit hole, what were some of the   approaches that you took to being able to understand more fully the market that you were working in and the constraints of the problem domain that you were trying to solve for. Thank god for doctor Google. You know, there was years of lots of study on just what drives these things, reading,  

[00:20:49] Unknown:

you know, everything I could. I'm an avid consumer of The Economist Magazine.   But, yeah, it was a self driven, you know, not a formal education at all, but working with the banks, you know, in   getting derivatives set up for clients and,   you know, also just   Bloomberg, you know, is kind of the basic big source of   raw data market data, you know, like what spot rates, what, you know, option premiums and all that kind of stuff are. And so kind of a deep study of how those things are put together and how they relate to each other. It's continual education, you can't ever stop, basically.   Same thing with being an engineer. So Yeah. And that's the fun of it. Right? I mean, if there was no more learning, I'd be pretty bored. Exactly.  



[00:21:32] Unknown:

And speaking of newer developments, particularly in the financial industry,   sort of potential strategies for risk mitigation and currency exchange   and how that might contribute to various elements of volatility   or even potentially act as a stabilizing factor for certain currencies?  

[00:21:57] Unknown:

I will divide that question into 2 parts. You know, 1 is blockchain and the other is cryptocurrency.   Blockchain, I see is quite a   valuable tool, especially   with Ethereum and smart contracts.   There's definitely a role for smart contracts to play in this space, in hedging and in cross border transactions of of all kinds, of self executing contracts.   For example, contingent   hedging, something I've been working on where, you know, if you've got,   let's say, you've got a cross border m and a. Right? You're about to buy this company, but you gotta do 3 or 4 months of due diligence. You agree on a price upfront in your currency and theirs, and of course, to the next day it's different. Right?   So 3 months from now, what will the price be in each currency that needs to be hedged?  

But if the deal doesn't go through,   that eliminates a number of different derivatives that you might have used. And so   developing an alternative method, which is what I've been working on is   badly needed in this area. And connecting that to smart contracts so that in the beginning at the inception of the m and a, both parties could enter into that smart contract.   And based on the outcomes of the due diligence, then it could either the deal could go forward or not and currencies exchanged or not. So   blockchain definitely, I believe so.   Cryptocurrencies   themselves,   I personally don't think they're ready for prime time for a lot of reasons.  

Number 1, you know, from an environmental standpoint, they're a disaster, you know, from their power consumption.   Number 2,   as a store of value, they're a disaster because they're probably the most volatile thing out there. I mean, they make the Argentinian basil look stable.   And so as a store of value, they're a complete failure.   And, you know, for countries what was it? El Salvador? Was that the country that decided they'd accept crypto? I think so. I think it might have been 1 of them. I think Colombia might be another. Yeah. Very, very bad idea. And here's why. I mean, when a country issues its own currency,   it its central bank has a lot of leverage to pull in terms   of devaluing or, you know, making more valuable their currency. And so   their inflation, managing their exports and imports, because they can control their own currency, then they have those levers. With crypto, they've got no levers at all. And so, you know, they basically give up a lot of really important control for a country's economy.  

So from those 3 contexts,   I don't   see it entering into the big plus, lastly,   because of the computational load,   was it 7 or 10 transactions per second? I mean, some pathetic number that wouldn't even handle the smallest country in the world. Right? I mean, to scale that up so that you can handle the trillions of exchanges that happen every day worldwide,   there needs to be a different way. So main answer is no.  

[00:24:46] Unknown:

It's a good answer.   The answer to cryptocurrency is no. It doesn't have a question. You can quote me on that. I'm prepared to marry up.   And so in terms of   the tools that you're using and the sort of mathematical principles that you're building on top of,   what are some of the areas that you're working in where the existing tools or the established practices are insufficient and you have had to either build out your own capabilities or develop new algorithmic approaches or   decide that, you know, a potential risk mitigation strategy is not worth exploring because it's too sort of computationally complex or, you know, you have insufficient confidence in the results that you're generating? Yeah. There's  

[00:25:31] Unknown:

parts of that I can answer and parts that I really don't have an answer for. I would say   the biggest problem with   bringing   math to finance is,   you know, it's not physics. You know, there's humans in the loop. And, you know, there's all of the elements of behavioral finance. Right, biases and stuff that creep in. The efficient market hypothesis is the   foundation of a lot of math and finance, and it's proven completely bogus. Right? There's no efficient market. I mean, if that was the case, then, you know, you wouldn't have market crashes, you wouldn't have stock market circuit breakers, all that sort of stuff, you know, kinda makes a mockery of the efficient market hypothesis.   And so, you know, when you make   assumptions like we do, you know, that things are normally distributed and that, you know, the the kurtosis is right around 3 and there's no fat tails and that kind of stuff.  

It simplifies the problem enough so that we can talk about it and work with it, but it also does not provide,   you know, for the things that, you know, that happen. I mean, you look at, you know, Taleb's, you know, black swans, you know, writings, and it's completely true. Right? I mean, you know, it it all works fine until it doesn't.   And that's where a lot of tools that we have break down and I frankly don't have a lot of,   answers for those because when they really, really break like they did in, you know, 09 or 10,   you get into the area of counterparty risk. So, you know, I can enter into a contract with a bank to exchange a certain amount of currency in the future. But if that bank goes under, they're not gonna honor that and I'm stuck. And so that's where   the tools   don't handle everything that the market can throw at them. I frankly don't have an answer for those situations.  



[00:27:10] Unknown:

Yeah. Your response about the being able to properly model the market, you know, the humans in the loop. There's the famous quote of all models are wrong, but some of them are useful.  

[00:27:21] Unknown:

That was that's exactly right. Who was what was that? I forget. I forget who But, yeah, that's exactly right. You know, when things are running along smoothly, all of these things work really well, and then they don't.  

[00:27:32] Unknown:

So for people who are interested   in understanding more about this space or potentially getting involved, I'm wondering what are some of  

[00:27:40] Unknown:

the useful resources or references that you would recommend or specific sort of background knowledge or capabilities that people should have? Yeah. So that's a great question because I'm yes. I'm sure all engineers are all the fans of lifelong learning. So some of the resources that I've used around Python   is that if the SciPy convention, the annual SciPy convention, you know, they have all of their stuff on YouTube. And so I've watched   bazillions of hours of SciPy convention presentations, and they're exceptionally valuable.   There's a number of   kind of continuing resources like Real Python, which I subscribe to as well, and they give you kind of a weekly   newslevery thing about the the latest releases or here's a new function you might have thought about.  

The 1 I think that's nearest and dearest to my art is there is a YouTube channel called 3 blue 1 brown. I don't know if you're familiar with it or not, but, of course, my wife runs from the room when I put it on the TV. But it's   Grant Sanderson's   channel and he makes math   spiritual,   beautiful,   instantly graspable.   He's got sessions on,   you know, linear algebra, he's got stuff on calculus, he's got stuff on   crypto,   and he makes it all accessible   with the most incredible,   not simulations, you know, animated graphics that bring all these concepts to life. This is like you immediately get into your gut. Oh, shit. That's what an eigenvector does. Okay. Now I got it. You know what? Never got it when the professor told me about it, but now I get it.  

[00:29:10] Unknown:

In terms of your experience of working in the space   and developing your own methods and discovering other people's methods, I'm wondering what are some of the most interesting or innovative or unexpected strategies or techniques that you've seen applied to this area of risk management? I guess,  

[00:29:26] Unknown:

something that I don't use or don't need to use, but I find fascinating is sentiment analysis.   So there's risk managers and there's traders. Right? Risk managers   try and take, you know, all of the beta out of everything. Right? So if nothing moved, we would be successful.   Whereas, you know, people that are trying to extract alpha out of the market, they're not like us. And   I'm sure that, you know, the sentiment analysis is 1 of their huge tools. And being able to scrape the web across all news and social media channels   to kinda get an idea about   whether to be bullish the euro or not, to me, that's really really interesting. And if I was in that space, I would kinda be all over that.  



[00:30:04] Unknown:

In your experience of working in this space, what are some of the most interesting or unexpected or challenging lessons that you've learned? Explaining  

[00:30:11] Unknown:

what I've done to fund managers. When I first started working with fund managers,   I was full of trepidation because I was like, oh, shit, these guys are really smart and they're making 100 of 1, 000, 000 of dollars and, you know, probably talking,   you know, kindergarten to them. It was the other way around. And,   you know, statistics   and, you know, violin plots or box plots, man, you'd have to hold their hand through the thing. I mean, once they get it, they definitely grok it, but that's been   1 of my biggest surprises is their   unpreparedness   given   the amounts of money they're moving around and the things they're doing.  



[00:30:49] Unknown:

Absolutely.   And so   in terms of   the work that you're doing going forward, what are some of the economic and industry trends that you're keeping a close eye on, either for your own personal projects or for the work that you're doing with Diaglo?  

[00:31:03] Unknown:

Yeah. So 1 of the things actually that I've just recently started digging more deeply into is metrics and benchmarks. So,   you know, when a fund is, you know, pitching itself to investors, they, you know, talk about their past   performance and they use, you know, terms like an internal rate of return   or MOIC, you know, which is basically a capital multiplier.   There are now a number of new measures, something called PME, public market equivalent,   which is basically,   if I had invested in the S and P 500,   would I have done better than investing in your fund? Right? So there's some   new metrics and getting the data to calculate those and some of its variants like direct alpha is something I've recently been digging into. Because 1 of the things that we need   to do when we're   running back testing is to use the metrics they're interested in so that we can show how our strategies would have performed   against the metrics that they're actually using.  



[00:32:00] Unknown:

Are there any other aspects of the work that you're doing at Diablo or the overall space of FX risk management   or your experience   of, you know, building these systems and algorithms to be able to understand financial markets that we didn't discuss yet that you'd like to cover before we close out the show? I can't really think of anything right now other than I really love what I get to do because I get paid to think, and I don't think there's a better job. So that's pretty much it. Alright. Well, for anybody who wants to get in touch with you and follow along with the work that you're doing, I'll have you add your preferred contact information to the show notes. And so with that, I'll move us into the picks. And this week, I'm going to choose a movie that I watched recently called The Vault. It's an interesting heist movie. You know, 1 of the main protagonists   was a an engineer, and he was hired on to this crew to be able to figure out how to   counteract this very sophisticated vault to be able to, you know, get the contents from inside of it. So it's a fun movie, you know, not quite your typical heist movies. There's a little bit more sort of physics and thinking involved, so it's it's a fun film to watch. So definitely recommend that if you're looking for something to stay entertained with. And so with that, I'll pass it to you, Paul. Do you have any picks this week? I'm taking my motorcycle to the Grand Canyon  

[00:33:11] Unknown:

for a week long tour of the North Rim, and I'm gonna be trying to avoid every paved road and and stay on the trails alone and be by myself. So that's what I got coming up, and I've been planning all the mapping and stuff for that. It sounds like a fun trip. We hope you enjoy it and definitely stay safe out there. Alright. Thanks.  

[00:33:26] Unknown:

Alright. Well, thank you very much for taking the time today to join me and share the work that you're doing at Diaglo and your experience of doing FX risk management. It's definitely very interesting problem space and appreciate all the time and effort you've put into your work there and the time that you've taken to share your experience with us. So I hope you enjoy the rest of your day. Thanks very much, Tobias.   Thank you for listening. Don't forget to check out our other show, the Data Engineering Podcast at data engineering podcast dot com for the latest on modern data management.   And visit the site of python podcast.com   to subscribe to the show, sign up for the mailing list, and read the show notes.  

And if you've learned something or tried out a project from the show, then tell us about it. Email host at podcastinit.com   with your story.   To help other people find the show, please leave a review view   on   Itunes   and  

[00:34:17] Unknown:

tell   your   friends   and   coworkers.