Need help for college project

[gammassas]

I am an CS undergrad and I love finance too.(MBA+CFA Aspirant)
I am planning on doing a Project in finance sector. Once I saw a post on Quora where one of the commenter said that most of the softwares used in hedge funds or trading firms are crap and are not good so I am thinking about developing a new or optimizing existing software.
So here is my question,
Do the software work like crap? If yes then what type of improvements are expected?
Note: I am a newbie in finance and trying to learn on my own.

 

[Daniel Duffy]

I am an CS undergrad and I love finance too.(MBA+CFA Aspirant)
I am planning on doing a Project in finance sector. Once I saw a post on Quora where one of the commenter said that most of the softwares used in hedge funds or trading firms are crap and are not good so I am thinking about developing a new or optimizing existing software.
So here is my question,
Do the software work like crap? If yes then what type of improvements are expected?
Note: I am a newbie in finance and trying to learn on my own.

You are a CS student and you do not know the status in the software industry?
The best improvement is that you learn to write structured, reliable and readable code during your stay in the ivory towers.

You can't put an old head on young shoulders. Seriously, check out these MSc theses in C++.

http://www.datasimfinancial.com/forum/viewtopic.php?f=198&t=1198

 

[Keith Tan]

for traders, there is no such thing as the 'best' software. what works for one may not work for another, hence 'crap' software to them. Bear in mind strategies are proprietary.

so you want to make the 'best' software? i hope you don't spend much time writing one only to realize there are hundred of vendors to compete with you for the 'best' title. And good luck finding end users.

 

[mhy]

I am an CS undergrad and I love finance too.(MBA+CFA Aspirant)
I am planning on doing a Project in finance sector. Once I saw a post on Quora where one of the commenter said that most of the softwares used in hedge funds or trading firms are crap and are not good so I am thinking about developing a new or optimizing existing software.
So here is my question,
Do the software work like crap? If yes then what type of improvements are expected?
Note: I am a newbie in finance and trying to learn on my own.

Well, think about this. You build a derivatives pricing system and wire it up to get LIBOR rates from the bank's market data repository and price interest rate derivatives like swaps and caps/floors. You spend weeks testing it and make sure it won't have issues such as missing market data, edge cases, slowness, etc. Right after you release it, the trading desk (for whatever reason) decides to use OIS rates instead of LIBOR and you have to change the code very quickly to accommodate this. There is no time to test because everyone needs everything ASAP. Next week you might have to unwind some of that code and support both LIBOR and OIS. Then the week after that, management gets reshuffled and your whole project is thrown out the window and replaced with another "strategic" thing that the new management has dreamed up.

That is not a real example, but it illustrates this: things change so quickly that you really have to choose between robustness and speed to market. The desk could be mispricing client trades and doesn't have a clear view of their risk - do you really want to wait another month to test everything again (at which point things have probably changed again)? In most cases, the answer is no; even if it's buggy, if it works 95% of the time it's better than not having anything.

The vendors in this space usually focus on slow-changing things like reporting/accounting software, reconciliation and matching engines, using well-established models and theories that are unlikely to change for a long time. You won't find a vendor out there that can supply a bank with a full suite of real-time pricing / trading applications because there simply is no way a vendor could get the product out and update it at the speed that these banks demand.

 

[C S]

I am an CS undergrad and I love finance too.(MBA+CFA Aspirant)
I am planning on doing a Project in finance sector. Once I saw a post on Quora where one of the commenter said that most of the softwares used in hedge funds or trading firms are crap and are not good so I am thinking about developing a new or optimizing existing software.
So here is my question,
Do the software work like crap? If yes then what type of improvements are expected?
Note: I am a newbie in finance and trying to learn on my own.

Hey, I saw a post somewhere where someone said most of the posts on Quora are written by people who don't actually know stuff.

Jokes aside, that kind of post is hard to take seriously. It's very difficult to get jobs developing at very good trading firms, and they are highly paid, often at a sizable-enough premium to cause big defections from Silicon Valley. So it's a given these people are very capable. If indeed the software they produce is crap, it's not for lack of smarts or education. It's organizational. Kind of like how Apple Maps will send you off a cliff. It's not because Apple isn't able to hire smart engineers.

In my experience, claims about the crappiness of Wall Street software arises from lack of understanding of the business environment (other businesses that have zero to little tolerance for errors have similar issues with legacy code, convoluted designs, etc.), complexity of finance (most people think financial instruments and transactions are simple), and severe selection bias (to combat this, please note there are major failures by tech giants, and if you pay attention, you'll find they have smaller failures like this all the time: http://www.ibtimes.com/google-gmail...d-failure-update-security-certificate-1870416 . If trading were halted for two hours because of some tech failure like this, heads would be rolling...nay, kicked across the tech floor like soccer balls. )

 

[mhy]

In my experience, claims about the crappiness of Wall Street software arises from lack of understanding of the business environment (other businesses that have zero to little tolerance for errors have similar issues with legacy code, convoluted designs, etc.), complexity of finance (most people think financial instruments and transactions are simple), and severe selection bias

This is very true. If I interviewed every developer at JPM/BoAML/Goldman/MS/Citi who claims to have developed "derivatives pricing systems" and asked them the definition of a bond, how to hedge a swap, or show me what a vol smile looks like and why it exists, I doubt 10% of them would be able to give me a coherent answer. The lack of basic financial knowledge, even among those responsible for writing arguably the most important software in a bank - pricing engines - is astounding.