Tips for an entry level C++ quant dev interview

[Daniel Duffy]

I see. That's something insightful and new I learnt today.

Out of curiosity, the BOOST, Eigen, LAPACK libraries - under the hood, they may be using many numerical algorithms originally implemented in good old FORTRAN, that were memory efficient & super-fast. Is that accurate?

AFAIK, not Boost but certainly NAG, LAPACK etc.

 

[pingu]

I see. That's something insightful and new I learnt today.

Out of curiosity, the BOOST, Eigen, LAPACK libraries - under the hood, they may be using many numerical algorithms originally implemented in good old FORTRAN, that were memory efficient & super-fast. Is that accurate?

LAPACK does. Neither Boost nor Eigen rely on FORTRAN.

 

[Daniel Duffy]

Boost UBLAS (matrix stuff) is not fast.

Probably nothing in C++ can beat NAG (it's been around forever, i.e. TESTED ad nauseum).

 

[Quasar Chunawala]

Hi all,

I am solving a few representative examples on LeetCode. For many problems e.g. longest palindrome in a string, while I can build a naive brute-force algorithm(mine is \(O(n^{2})\), I can't think of a faster algorithm.

I feel, this requires tonnes of practice & also training yourself to apply algorithm design techniques like divide & conquer. But, specifically pertaining to C++ interviews - do interviewers grill further and ask for a faster, efficient algorithm, once I talk about a brute-force algorithm?

Thanks so much for all the valuable inputs to you guys!

 

[Quasar Chunawala]

I had four interview rounds yesterday with people from credit, FX, rates and technology team interviewing me.

They asked mostly math puzzles e.g. which is bigger \(e^{\pi}\) or \(\pi^e\), riddles, probability puzzles, algorithms(with occasional source code).

I did quite well in three of those rounds, fourth was 50-50. I think intuition, presence of mind and talking my thoughts loud were the key.

 

[Quasar Chunawala]

One probability puzzle was as follows - a coin is flipped repeated until you or I win. If the sequence HHT appears, I win. If the sequence THH appears, you win. Which sequence is more likely?

I cracked this on the fly. I drew a probability graph(like a finite state automaton from my CS course). Perhaps, there is an easier way to solve this. Got a big plus for answering this correct.

 

[Daniel Duffy]

I had four interview rounds yesterday with people from credit, FX, rates and technology team interviewing me.

They asked mostly math puzzles e.g. which is bigger e^pi or pi^e, riddles, probability puzzles, algorithms(with occasional source code).

I did quite well in three of those rounds, fourth was 50-50. I think intuition, presence of mind and talking my thoughts loud were the key.

Prove that the max of pow(x, 1/x) for x > 0 is Euler's constant e.

 

[Quasar Chunawala]

Ah, got it! I found out the max of x/(ln x) and worked from there.

 

[Quasar Chunawala]

I had one final round of interviews today. I was asked three questions - (1) Pricing an option using a 1-period binomial model (2) Checking if arbitrage is possible (3) a probability puzzle. The second question was as follows :

A call option with strike \( K=110 \) has price \( c=55 \), maturity \( 1 \) year.
A put option with strike \( K=120 \) has price \( p=30 \), maturity \( 1 \) year.
The underlying is at \( S_0=100 \)

Is arbitrage possible? I wasn't able to answer this pretty basic question. After a couple of hints/prompts about applying put-call parity, I got to the solution. I answered the 1-period binomial model & used brute-force to solve the probability puzzle.

Does failing to answer the above basic question mean that I am absolute toast in a quant interview? I would like to take it constructively and prepare better now.

 

[Berenger]

I had one final round of interviews today. I was asked three questions - (1) Pricing an option using a 1-period binomial model (2) Checking if arbitrage is possible (3) a probability puzzle. The second question was as follows :

A call option with strike \( K=110 \) has price \( c=55 \), maturity \( 1 \) year.
A put option with strike \( K=120 \) has price \( p=30 \), maturity \( 1 \) year.
The underlying is at \( S_0=100 \)

Is arbitrage possible? I wasn't able to answer this pretty basic question. After a couple of hints/prompts about applying put-call parity, I got to the solution. I answered the 1-period binomial model & used brute-force to solve the probability puzzle.

Does failing to answer the above basic question mean that I am absolute toast in a quant interview? I would like to take it constructively and prepare better now.

Sorry for resurrecting this thread: doesn't put-call parity apply only to calls and puts having the same strike (and underlying and maturity, of course)?

 

[RowYourBoat]

Sorry for resurrecting this thread: doesn't put-call parity apply only to calls and puts having the same strike (and underlying and maturity, of course)?

The ITM put is worth a decent amount less than the OTM call.

Put-call parity tells us a \( K = 110 \) put should be worth \( 110e^{-r} - 45 \). But the \( K = 120 \) put is worth 30 and put price must be increasing in strike. So we need \( 110e^{-r} - 45 < 30 \Rightarrow r > -\ln \frac{75}{110} \). So unless the continuously compounded interest rate is above 38% there is an arbitrage

 

[Berenger]

The ITM put is worth a decent amount less than the OTM call.

Put-call parity tells us a \( K = 110 \) put should be worth \( 110e^{-r} - 45 \). But the \( K = 120 \) put is worth 30 and put price must be increasing in strike. So we need \( 110e^{-r} - 45 < 30 \Rightarrow r > -\ln \frac{75}{110} \). So unless the continuously compounded interest rate is above 38% there is an arbitrage

Very well, the wrong guess can always be justified by stating that it would be true in Weimar Germany, or early 2000s Argentina, or the stellar 7600% inflation of Zimbabwe under Mugabe.