[Thread of Threads]
This thread should refers all threads done on this account. These threads deals mostly with maths and computer science and are mostly focused on simulation
[Preparing Quant Interviews]
Disclaimer: I'm not an expert, I'm just a student who graduated some months ago and is now in the industry and I want to share my experience and what I learned so far. So the following thread might not be exhaustiv but it might be a good help for you
I have the better question - I say you N random variables with correlation p between each pairs; what is the maximum and minimum for the correlation p?
[Brainteaser 101 – 1 : rolling dices]
One very simple brainteaser you might get in an interview is the following one. You are playing a game where you roll a fair dice. If you roll n, you get n dollars.
@quantymacro
Yeah kinda hard
Tbh the only way I see is to :
- Follow usual courses (stats, linear algebra, programming)
- Read some books on the market (Hull for example is a good introduction)
When you have some technical skills, try to build stuff (Options pricing etc. )
I recommend you as well Abdul Bari's channel () for the explanations of how the algorithms work etc. and if you are more into books, you can read 'introduction to algorithms' which is a reference in the domain
Concerning Stochastic Calculus, you need to know your stuff on how to determine the law of a Wiener integral, apply the Ito lemma, solve usual SDE (geometric brownian, Ornstein-Uhlenbeck, ...), Girsanov, Feynman-Kac.
Just, a little difference is that if you apply for a trading position, something that appears a lot in those tests are 'mental calculation' For those I can recommend you to go on zetamac ()
Like when you overestimate your skills by saying that you are an expert in C++ by example. Well in many cases it's fairly easy to see that it's a lie and it's really bad for you if it's discovered (and it will be) as your interviewer can easily destroy you and worse
To train I can recommend you two websites :
- leetcode
- hackerrank
Those websites have a lot of problems to train and, fun fact, you can find most of your assement's problem on those websites
So for the books recommendation :
if you don't know anything about finance you can start with the Hull (Options Futures and Other Derivatives). I'm not a big fan of this book as it doesn't go very deep but it can still give the big picture so more like a book for 'context'
@Lptomov82
@MathMatize
The square root is only defined on positive real numbers
Thus the square root of 4 is 2
On the other hand, the square is defined on all real numbers and thus (+/- 2)^2 = 4
It's just a question of definition
Something that I personally find strange is that when you want to start your Machine Learning journey, there is a lot of "best ML courses" and similaire stuff but not that much about "good datasets"
I mean, I don't think that you can learn a lot of things on this kind of dataset:
@ubertortank
Wall street Oasis et un peu tous les autres sites du domaine
C'est le format qui est recommandé donc les gens ne se prennent pas la tête
(Et c'est mieux que les CV avec une photo qui prend les 3/4 du CV mdr)
For a focus on interview questions, there are special books :
- Heard on the street
- 150 Most Frequently asked questions on quant interviews
- A practical guide to quantitative finance
- Quant Job Interview Questions and Answers
[Simulation of gaussian processes]
Suppose that you work with a gaussian process. You may be in a case where the theoretical study cannot be further advanced. Therefore, you might want to simulate your process to get more information.
Specifically for brainteaser, or at least to work on the thinking, you can take a look at brainstellar (). The discrete/probability puzzles are kinda interesting for this. Furthermore you can take a look at '50 problems in probability with solutions'.
So first of all let's discuss about CV and fit part. What I mean by that is that you mostly have the 'right' CV format so that's good. but something that you might do is lying on it.
Now that you have your CV, if you pass the screening, the company you applied to might send you the link to a test.
So far I've seen three types of test :
- interview oriented test
- coding assessment test
- 'mini project' test
You need to prepare because if you have never seen how to solve a knapsack 0/1, it's very unlikely that under the stress you can come up with an answer.
For the coding assessment test, it consists in solving algorithms problem in a defined time (60 or 90/120 min in general for 2 to 3 problems). For this there is no shortcut, you have to PREPARE A LOT.
📣 C'est l'ouverture des inscriptions
#Parcoursup
!
➡️ Les vous pouvez citer ce tweet avec votre filière d'étude post-bac et votre établissement pour que les terminales/étudiants en réorientation puissent vous DM si ils ont des questions
#Parcoursup2024
📚
Now for finance you can have basic questions, difference between a forward and a future, payoff of a digital, of a straddle, how to reproduce a straddle with vanilla options. What are the greeks, their interpretations, the limits of the black-scholes model, ...
Started reading "Trades, Quotes and Prices" so far I appreciate it but reading it as a bedside book might not be my best idea
If you have any recommendations on the same subject I would gladly appreciate
Concerning numerical methods, we talked about Monte-Carlo and finite difference so be aware of the convergence order, the pros and cons of each methods, ...
@miniapeur
Feynman-Kac
I find it so astounding that somehow you can link PDE and stochastic processes and thus two different people with two different background can in fact be working on the same object
and for numerical methods it will mostly be around Monte-Carlo and finite difference.
On the other side if you are interviewing for a hedge fund, question might be more directed towards statistics and machine learning.
Now looking at the difficulty of the problems, if you are a 'simple' Quant, you will mostly have medium problems but if you apply for Quant Dev position or SWE, then the odds are on the hard difficulty.
Is it worth to learn rust ?
As a *fast* programming language it seems good so far but if the objective is the industry then C++ might be a better choice
But you can as well have stuff about optimization like the gradient descent or more 'finance' oriented stuff like the Longstaff-Schwartz algorithms and in which contexte it's usefull.
For maths oriented questions, the 'subjects' kinda depends on who's interviewing you. If you are interviewed by a bank, as you mostly work on derivatives, your 'pure maths' questions will be oriented towards probability, stochastic calculus, algebra, analysis
For SWE it's good for you to read as well 'cracking the coding interview'. And in a more general way, if you need a roadmap, you can take a look at neetcode ()
I might -in the following weeks- write a thread on MC simulations. Might be useless as every student has a fairly good understanding of it but if you struggle on one point of MC, feel free to reply here, I will try to answer it in the thread
For the interview oriented test, I mean that those tests are mostly oriented towards questions that could be asked in interviews and thus it generally convers everything from brainteasers to numerical methods but I will develop it more in the interview part of this thread.
Now for the 'mini project assessment', for what I have seen, those mostly appears for hedge fund and are mostly oriented towards :
- Data Analysis
- Machine Learning
- market making
It's a bit a pain in the ass but there is not an infinity of tricks therefore if you do a lot of them and understand the patterns, you should be good on it.
Concerning machine learning, most questions I have seen are around the basics. What is the linear regression, lasso, ridge (pros and cons), PCA, bias-variance tradeoff etc...
[Brainteaser 101 – 1 : rolling dices]
One very simple brainteaser you might get in an interview is the following one. You are playing a game where you roll a fair dice. If you roll n, you get n dollars.
The more I do quantitative finance, the more I become aware that in fact, nothing (or at least not that much) stuff comes from quantitative finance. It's mostly tricks taken from other math areas and developed for other problematics
@L_Econologiste
@JomauxJulien
Depending on the situation (meteo, congestion, issue with capacity with other countries, ...) prices go up and down and it can be quite volatile thus for the same delivery for the same zone you can have multiple prices and the difference can be huge
@BlackSwan_ptf
1) A Digital option (let's suppose it's a digital call option) is an option that pays 1 if the underlying is above the strike at the maturity and pays 0 otherwise
I think that's all. There is a lot more ressources but tbh I have not read all of them so I'm not that much into recommending it. If I read them and like them, I will add it to this thread. Good luck for your interviews guys.
@draculacomeback
T'as vu ça dans beaucoup de HF ?
Côté como c'est vraiment pas déconnant
En power ça joue énormément vu que le delta entre ton forecast et la réalité est une composante importante des prix
BUT don't forget that this is in the GENERAL CASE, when talking with some friends, depending on the exact position / subject of your internship, the question asked vary a lot.
For the brainteasers there are two kinds :
the first one are brainteasers on probability and the second are 'tricks based' brainteaser.
For the probability one, most of them use cards, dices, ...
To go deeper into finance and with a bit more mathematics, you can read the Bouzouba (Exotic Options and Hybrids). In this one you will see more stuff dealing with options, volatility surfaces, models, ...
@random_poisson
@oooooooorion
L'argent
Pas simple de bien vivre sur Paris
Honnêtement si j'avais mes parents en banlieue proche ou mieux sur Paris, je serais resté chez eux plutôt que de passer au CROUS (encore plus maintenant avec un *vrai* loyer à payer)
@L_Econologiste
@JomauxJulien
For power in the short term, you can either buy electricity on the day ahead (an auction) where there is an unique price per country. However you can as well buy power on the intraday market and in Europe it's a Limit Order Book therefore with bid/ask
So, to put in a nutshell:
The Cholesky method is a general method useful for simulation of gaussian processes because:
- low hypothesis on the covariance matrix
but it has two drawbacks:
- complexity in O(n^3)
- take care of "too bad" conditioning
@BlackSwan_ptf
Issues with short term maturity so a stochastic volatility model.I would say Heston but just because it's the only one I know for equity
You could try to improve by adding jumps (thus you would work with bates model) but from what I remember adding jumps make the market incomplet
@TonyTheLion2500
Depends if you are Buy or Sell Side but to put in a nutshell :
- Stochastic Calculus
- Probability
- Statistics
- PDE
- Optimisation
- Linear Algebra
So far my ML journey has been mostly around find *nice* datasets, EDA/Dataviz to understand what's going on and reading documentation/code. I'm far away from only building models
@usulmuabddib
Postuler != s'engager
Et passer des entretiens ça permet d'engranger de l'xp en choppant des questions et en étant en situation. Ça peut toujours te faire step up
@usulmuabddib
Side avec Ask/Bid
Vu que ce tu veux faire (un LBO de ce que je vois), il faut que tu puisses reconstruire facilement le LBO à partir des orders
J'imagine qu'après tu voudras aussi sauvegarder les trades donc rajouter une "execution_date"
@thecarefreebear
@BlackSwan_ptf
@kebabroyal_
@Almost_Sure
@CoeusCap
Isn't it the generator of the semi-group associated to the Geometric Brownian ?
It appears in the Feynman-Kac theorem in the PDE and you can link that to the pricing of options through the risk neutral pricing with the discounted products being martingales under Q