Der heck is this Site?

I once came across a meme on Twitter that said if your job can’t be described in three words, it is not a real job. It got me thinking whether my profession constituted a job. Is financial engineering a real job? Is it even a science? Why exactly does my degree doc say B.Sc. even? These doubts surfaced in my mind for two main reasons: one, I have never met an actual practicing quant. I know they are out there, and I know what they are trained to do, but I have never met one in real life so I don’t know if theory and practice match. Most quant jobs seem to be taken up by people from other fields, especially physics. Secondly, I have always had to explain to people what financial engineering is. From 2013 to 2018, I would often get asked by my relatives or acquaintances what I did at Uni. Those conversations usually went something like this:

Them: “Btw Brian, what are you doing in campus?”

Me: “I’m doing Financial Engineering.”

Them, intrigued and confused: “What is that?”

Me: Proceeds to try to explain about pricing financial assets, valuing companies, applying risk management, backtesting trading strategies et cetera.

Them: Looking increasingly confused the more I explained.

Me, panicking: “It’s like actuarial science but for banks!”

Them, still not quite understanding what I meant: “Ah okey, it sounds very intriguing.”

…

Sometimes, when a trade is kicking my ass, and my heart is getting increasingly heavy as the price etches perilously close to my stop loss, I put on a trading movie at the office to remind myself why I do this in the first place. I usually search through the small catalog on my computer, and choose one that fits my prevailing mood. Say Margin Call, but sometimes Wallstreet-the original one. But, instead of soothing me, these movies often leave me feeling frustrated. There is a dearth of proper trading movies. Maybe another test of whether you have a real job is if they make movies and TV series about it. As Jared Dillian once noted, there are TV series about doctors, lawyers, cops, teachers, firefighters, even friggin’ lifeguards, but close to none about traders. What’s more, the ones we do have don’t quite get it right. Trading movies feel to me as a trader like a porno, except instead of seducing each other and having sex, the actors talk about their childhood traumas and play Scrabble. Industry comes close, but the scenes where actual trading happens are a bit overly dramatic and sometimes farcical. I suppose that’s the point, otherwise why make a TV series? Maybe even doctors, lawyers, etc. watch series about their professions and think: “That’s not what it’s like! That would never happen! Not like that anyway!” 

If there are few movies about trading, there are virtually none about quants. The only one I can think of that comes close is The Bank. If you haven’t watched it, don’t worry I won’t spoil it for you, but I will say this: it gets a few things right about the profession. Firstly, TV (and perhaps real-life) quants are expected to be working on something big with the potential to make people a lot of money. Same way TV doctors are always saving a life, or fighting an unheard-of disease. People have this perception that quants have some special knowledge and/or skills that translate to a lot of money. Secondly, quants often have people stepping on their necks, mostly fueled by greed. As a quant, even if you do find something big that could make people a lot of money, the chances of it ( and you) being abused thereafter are high. This is probably what happened to David Li. David did what quants are expected to do: find something big that could make people a lot of money. He used a Gaussian copula to price CDOs. But his invention was taken up by greedy people who abused it. 

I once had a dream that I had been invited to a private viewing of a new Red Bull F1 car. It was in what looked like a small castle in rural England, surrounded by green moors and dense trees. I remember walking around a corner and into a plain, arched, light-brown, hardwood double door that was slightly ajar, and into a carpeted room that looked like a professor’s study. The humongous car sat across the room in the middle, its front and tail nearly touching the walls. I started to marvel at every detail and harbored a childish desire to touch it, and even though no one had said not to, I knew not to. Suddenly, I realized Adrian Newey was on the other side of the car, also looking at it, and I said to him: “Even sat here in this room, it looks like it is going past us at 300km/h.” He said nothing and merely looked at me nonchalantly. Then I asked him: “It looks fast, but on a track, do you think it can it beat Leclerc, Hamilton, and Sainz?” Suddenly he smiled and said excitedly: “We’ll have to find out, won’t we?”

…

One of my most profitable trades of 2023 went down as follows. I was scrolling through Twitter in the days leading up to the NFPs, and saw some posts about just how many times the NFPs had beaten expectations. I felt that these posts implied that the upcoming figure would be the one that didn’t beat expectations, but I was frustrated by that kind of logic. There had to be a way to at least analyze the data and see if there was any evidence supporting such a claim. The simplest model for a time series is an autoregressive model, so I collected some employment data and fitted one. Using the historical data, I was trying to estimate what the next figure would be. The model spat out 175k which was below the expected figure, so I went long EURUSD about 40 min to the release. The figure came in at 187k, the EURUSD rallied higher and I took profit about an hour later. Winner-winner-chicken-dinner! I high-fived myself.

Then, after the high of making a few quid had subsided, It occurred to me just how bad and dangerous that trade had been. I had trusted a model blindly. I was no different from anyone who used Li’s model to price a CDO. I had fooled myself into thinking that by doing some analysis, I was being rational, or at least more rational than my cohorts on Twitter. My model had looked dependable to me in the same way the Red Bull car in my dream had looked fast, sitting in a carpeted study. But unlike dream Adrian Newey, I lacked the humility to understand that even if it looked good in theory, it might not get on a track and beat Leclerc, Hamilton, and Sainz. I shouldn’t have made money on that trade. Dream Adrian Newey probably understood that even if the car was a winner in theory, only a fast racing driver could bring out its best. This was also a key theme in the movie Ford vs Ferrari which I will also not spoil for you. 

I am the kind of person to try to add a persistent-homology-based loss function to a CNN to see if it will be more robust to images of price charts with specific price patterns, like head-and-shoulders or wedges, for example. Persistent homology works as follows: imagine you have a dark and barely visible image of something, but you don’t know what it is and you need to figure it out. A method similar to persistent homology would be one where you make many copies of the image, tweak the brightness, contrast, rotation, etc., of each image, and then check for the ‘persistent’ features in the images. The advantage of using this approach is that it allows you to do all sorts of distortions, and as long as the main subject isn’t substantially distorted, your model will be able to identify it. So for example, you could give such a model similar but slightly different iterations of the same thing, and it should know what the thing is. Think different-looking head-and-shoulders images for example, it would get the gist. The technical aspects of such a project sound really fun to work on, and even though I haven’t done it before, my toxic trait is that I think I can do it. I always think I can, until proven otherwise. 

…

Just as traders always have a view, quants always have a hypothesis. I have a hypothesis about the readership of this and other quant stacks. First of all, my hypothesis is that quant subs have the lowest readership among finance stacks, the reason being that the information in such stacks is not directly applicable and sometimes requires some technical know-how. Still, there are subscribers. My other hypothesis is that readers of my stack will come here for two main reasons: either the market is kicking their ass, and they are looking for new ways to make money, or there is nothing going on in the market so they are looking for entertainment and/or ideas. Either way, the driving force will be a desire to find a new way to make money. That is also one of the driving forces behind my work, so there is an alignment there, but like dream Adrian Newey, I just don’t know if my cool quant blog can get on a track and beat Leclerc, Hamilton and Sainz. That is to say, I’m not sure if what I write about will boost your bottom line or not. What I can guarantee is good analysis and eloquent presentations about them. My goal is for my posts to be useful to you somehow, even if just by giving you an idea for your analysis. 

Apropos movies, you know that part in a movie when the guy won’t talk. Maybe it’s a good guy who’s finally been caught by the bad guys in the big black SUVs. They’ve tied him up, knocked him about, perhaps threatened his family, nothing. Or maybe it’s a bad guy down at the precinct and they’ve just done the good-cop-bad-cop routine. Either way, when they’ve tried everything and he still won’t talk, there is always that one person who is their final recourse. If anyone is going to make the guy talk, it’s going to be him. Often he looks the part too. Big and burly, perhaps muscular and with a mean mug. Something about him freaks you out, but you can’t say what exactly, or why. Besides looking the part, he also acts the part. He might come in cracking his knuckles, or with a special toolbox. Maybe he pulls down the blinders to show that whatever he is about to do is unsightly.

That’s the sort of person I like to think of myself as when I analyze data. I will torture1it if I have to, but it will talk, and reveal any important information that could be of use to our cause. This is where I will be bringing my findings.

Welcome to my Sub.

The motivation behind it is that most traders and investors don’t have enough time, and sometimes know-how, to do the quantitative analysis needed to answer some of their questions about markets. Institutional traders and their ilk probably do, but that research is often private or paywalled. So I believe there is room for some Subs that crunch the data for everyone else. My edge in this venture is that I went to actual Quant school, have taught myself a bunch of useful stuff, have 4 years experience as a freelance quant and data analyst, and I thoroughly enjoy the process of analyzing data. I’m in a great position to use various models and techniques to uncover helpful information for traders and investors. That’s my value proposition.

My focus is on usefulness. I have a lot of interesting ideas that would be fun to try out. But I would like to focus on the ones that satisfy everyone’s curiosity instead of just my own, and more importantly, the ones that can make people money, or prevent them from losing a lot of it2. In every single post, the question I will be looking to answer is simply: “Can this make people money or not?” Based on my observations and studies3, the best way to make money in the markets is to avoid losing a lot of it, and then bid your time. Thus hedging efforts can sharpen one’s edge, ergo the name Quantitative (h)Edge.

As always, remember:

kthxlet’sgo!