1. I remember (not so fondly) about crashing out of the last round in Jump’s Quant Research team. I was the only candidate in the final; between nobody and me, they chose nobody. I was despondent. Someone, somewhere out there is struggling trading, and here’s a thread for you. ▼
2.Some background: I was pretty fortunate to attend one of the top ranking institutes in my university days, doing Computer Science and Stats. At that time, they were sending a pool of students to Silicon Valley & other places for internships, and me and my homies all applied.
3. So between the 5 of us, all of em except me got offered roles overseas, and I was turned down, despite my grades being echelons above the rest. All I can remember is, I took that L big, and I remember tearing up and feeling like I ate dogshit.
4. I probably got rejected because of my sucky interview skills, but I was blinded and miserable. You know how when your girlfriend dumps you, and you start gymming? Yep. Trading started off as “vengeful” motivation for me. Those interviewers, when I am rich, that will show em…
5. So that was the birth of my trading career, a “dorm-room” fund of sorts. Turns out, these “shocks” to my system often proved to be inflexion points in my life. Of course, the market being ruthless, the path from there was tumultuous.
6. I was solidly unprofitable, but learning scrappily wherever I could. It took me a year of trading/coding, and a lucky stipend-less internship at a quant fund before I started seeing some profits. I was ecstatic, and I was having great fun. Then the dark spell hit.
7. The unspoken law of FAANG is, you better get in the internship program in your penultimate/final year or you are ngmi. Also, the way to ace this interview is to pick up an Algorithms book and grind it out on leetcode.
8. I thought “wtf, I’d rather pick up a book on Stoch Calc than learn about greedily flipping pancakes”. So not only were my interviews lacking, my resume was sparse. Hell, I wasn’t going to make public tens of thousands of lines of code I worked on just to “showcase projects”.
9. I also applied to roles I was completely out of depth with. Job description clearly said “Masters/PhD” and I was just cold-emailing them. I even sat through half an interview with Virtu before I realised the interviewer did not read my email and thought I was going full-time.
10. So basically, I was not going anywhere with my internships, while my friends started getting roles at Google , Amazon, Visa etc…throw in Jump in the middle of my dark spell, and the all too familiar anxiety set in. To make matters worse, my trading strats went into drawdown.
11. I can lie tell you when I first started, I had no shred of doubt that I was going to be successful. But right there, right in the throes of pain, I started having doubts if I f-ed up. Maybe I should have just applied to SE roles instead.
12. I was screwing interviews, losing money and spent good number of days curled up in bed in pain as I kept bleeding out my savings. I felt “hard-stuck bronze”, like I was not moving forward in life.
13. Here’s the problem, a lot of times, when traders struggle, they feel like they are moving backwards or stuck. Most want to give up, some give up, and few strive on. At these points, the trader’s mental health are in vicarious positions.
14. Here’s what I want to say: why should the extent of your skill, to which is reliably and dependably improvable upon effort be measured by which we know to be largely arbitrary?
15. Using equity curve/PnL of a trader is a particularly poor measure of skill, and a harmful one at such. Don’t get me wrong, they are a perfectly fine measure of performance, but in the trading world, over short periods of time, performance and skill are not correlated.
16. My point is, if you are in a dark place right now, and you think you are “ngmi”, let’s take a step back and reevaluate the evaluation itself. It might just be not what you think. Dark spells end, and if you keep working on your craft, chances are, things will come around.
17. Eventually, for me, the dark spell proved yet again to be temporary and an inflection. That same year, instead of doing internships, I took my chances on a graduate level course, got to know the Prof who introduced me around.
18. Lo and behold, I ended up writing a paper with one of the finest quantum physicists of today. I’m talking the calibre of S. Lloyd (MIT), Nikitas (GS) ish. I can bet ZERO of my cohort got such an opportunity to work with legends like that in their Bachelor’s.
19. And my trading? That wasn’t the last of my struggles, but poco a poco, I kept improving my craft and the blood curdling and bed curling days are over. Probably will come back some time, but remember: the dark spells end. If you don’t capitulate, eventually the sun will shine.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
1. Couple of days ago, a paper from the great Bouchaud was brought to attention by Twitter-sphere’s Vol-Potter. One of the sections “My Kingdom for a Copula” introduced copula methods.
A thread; Portfolio Risk-Management, and What The Hell is a Copula?
2. The invariant of trading rule is to never go “bust”. Unless you are Hwang, chances are, getting re-financed is not an option; once you’re out, you are OUT. Unfortunately, correlation dynamics are not stable. To make matters worse, correlations increase in extreme conditions.
3. In Bouchaud’s words: tail correlations in equity markets are notoriously higher than bulk correlations. In order to deal with non-linear correlations, mathematics has afforded us with a seemingly powerful tool – “copulas”.
So, what is a copula?
1. Quant firm I worked for pitched as "computationally intensive,modern ML methods". Asked around and the “big guy” confirmed “Linear Regression”. Thoughts? Good Sharpe and attractive fees, no harm no foul. Also they weren’t wrong.😂
Thread; Deeper look into Regression Methods
2. Skip to 12 if you don’t want the ”college recap”, a lil something for the more unique. Btw, this is more of the sort of stuff you learn in an STAT101 class rather than ML class. Anyway, LR begins before college, doing y = mx + c. This we call Simple Linear Regression.
3. Ok that is bit too simple for most practical use cases in finance. So extend that to Multiple Linear Regression, we have multiple predictors. Now the mathematics need some college level algebra and calculus, but those should be fairly accessible and I don’t want to bore you.
(14) 1. Throwback: Remember interviewing for quant firm and when explaining projects I talked about AI models I built. We discussed some diagnostics about where the models failed. PM took a look and said, "well garbage in garbage out"😳 thread on application of AI in markets. 👇
2. This is a #travelthought, so bear with me if some points are incoherent. First, my advice on approaching AI
Now, AI sounds cool, so this warning obviously is not enough to put you off. If you insist on going down that path, then let's dive in.
3. First, let's take it down from its pedestal and demistify this buzzword. It is nothing more than a class of search algorithms.
Brute force? Let's improve. B/DFS? We can do better. A*? Now what if our task setting is partially observable?
There are different levels of trading in all kinds of trading techniques, and alot of people overestimate themselves/do not understand the intensity of competition in markets until it is too late.
The beginner level is for people who slap techniques from random sources
haphazardly, thinking that there is an "alpha leak" everywhere. Alpha leaks do exist, but at this stage it is difficult to tell legitimate alpha from marketing scams. These (not always but often) tend to be Youtube videos and Market Gurus, as well as amateur blogs written by
college students pursuing a side hobby. Don't get me wrong, some of them are awesome, but on aggregate finding reliable, legitimate sources of alpha/trading advice is almost equivalently difficult as finding the alpha itself. Trading attracts primarily 2 types at these stage,