MFE vs. MS Computer Science?

[Lyosha]

You don't have to go too far for verification.

In probability theory, a stochastic process (

/stəʊˈkæstɪk/), or sometimes random process, is the counterpart to a deterministic process (or deterministic system). Instead of dealing with only one possible reality of how the process might evolve under time (as is the case, for example, for solutions of an ordinary differential equation), in a stochastic or random process there is some indeterminacy in its future evolution described by probability distributions.

http://en.wikipedia.org/wiki/Stochastic_process

I was talking about high frequency trading amongst other things.

 

[alain]

We are talking about high frequency trading, right???

I don't think so. The OP change the tune to general trading.

 

[Alexander Krosky]

Obviously it's not entirely deterministic or else everyone would know where the price is going. I was conditioning everything I said on how someone with a trader's mindset would view it... in expectation. Stochastic processes all get phased out in expectation. In the model you mentioned, they don't even really offer an improvement over just using a simple z-score. So there's your pudding right there.

When you are looking for a loaded coin, you are looking for the win rate. Nothing else matters. It really is that simple.

But back on topic, in summary of the thread. Do you need a phd to do strats/modeling in high frequency? Yes. Is it wise to go get a phd to obtain this role? No. What topics should you study for high frequency trading? Signal processing, machine learning, information theory. Are stochastics relevant? In general, not really, (except for fixed income as Alain mentioned) especially in high frequency. As long as you realize that all of your parameters have distributions, you're fine.

 

[Lyosha]

Z scores are stochastic in basis Truth be told it's all a proxy to the "real" distribution of whatever you are looking at (if you want to look at it, the closest you can get is with historical simulation... I'm sure everyone has heard of 'fat tails' and 'tail events' before...).

I would agree that a rigorous treatment of stochastic calculus is not necessary for honestly most quant type positions. I would be surprised if anyone derived closed form formulas for their exotic options prices.

However, I do feel that a functional understanding of stochastic processes and the operating with them is highly advantageous, and in many/most cases of quant careers necessary. If for nothing else to understand the literature out there and more effectively draw inspiration from it.

That having been said, it is clearly not sufficient, and regardless of career path there is supplemental knowledge that may or may not be more valuable (I, personally, think the value of knowledge is relative and individual) but is required.

And that having been said, I completely agree that signal processing, machine learning, information theory, as well as optimization theory, parallel processing, dealing with excessively large data sets etc. is all very highly relevant to algorithmic trading, and probably more so than knowing how to solve the HJB equation.

And I would have to say, if you are a HFT trader, you probably have to understand stochastic processes a lot less than, oh, a flow trader trading exotic derivatives (a position that I would imagine it would be advantageous to understand the derivation/creation, underlying logic for and pitfalls in the models used).

Back to the OPs questions though - for more "flow" type trading, Finance/Math would be advantageous (be sure to take a programming course or two anyway though - it's useful). For more "algo" type trading - probably Finance/CS is the better way to go there (take an econometrics class if you can... assuming you're just an undergraduate).

 

[Aerial]

Back to the OPs questions though - for more "flow" type trading, Finance/Math would be advantageous (be sure to take a programming course or two anyway though - it's useful). For more "algo" type trading - probably Finance/CS is the better way to go there (take an econometrics class if you can... assuming you're just an undergraduate).

Alright, seems like a clear cut answer (which I like). I know for a fact that Stochastic processes and PDEs are especially useful when trading exotic derivatives but just wondering if it is necessary for any other financial instruments (for flow trading at BB S&T)? A lot of upperclassmen told me that when BB S&T recruiters look for "quantitative ability", they are looking more for quick mental math (ex. brainteasers) more so than stochastic calc and all that since ultimately a flow trader merely uses the models that the QUANTS create. Is this true, or is a full understanding of stochastic processes/calc/PDEs necessary to TRADE in flow trading? If so, which specific desks would this quant background be beneficial/necessary?

 

[Aerial]

And just to get back on topic - in terms of landing a job in algo trading, it seems that the general consensus is that a MS Comp Sci is better than an MFE? (assuming they're both from the top programs in their respective fields)

For anyone with experience in algo trading, could you highlight which specific courses (or concentration) should be taken in Comp Sci? (Ex. machine learning... and what else? Data Mining maybe?)

*sorry for so many questions

 

[Alejandro Canete]

Hi Aerial,

Machine Learning (kernel methods approach, don't get near the ANN/GP stuff), a big data course (map reduce, grid computing, etc), a computer architecture course (to understand the different factors that make up latency), concurrency programming, some course on design patterns focusing on enterprise integration (message brokering, ESBs, JMSs, AMQP, etc), scientific computing (a must) and an HPC course (GPU/FPGA oriented rather than MPI).

Hope it helps.

And just to get back on topic - in terms of landing a job in algo trading, it seems that the general consensus is that a MS Comp Sci is better than an MFE? (assuming they're both from the top programs in their respective fields)

For anyone with experience in algo trading, could you highlight which specific courses (or concentration) should be taken in Comp Sci? (Ex. machine learning... and what else? Data Mining maybe?)

*sorry for so many questions

 

[Yike Lu]

For buy-side style trading:

Theoretical stochastic is not very useful for linear instruments (i.e., no optionality). Time series analysis is far more useful.

One could argue that since time series is based upon stochastic, that stochastic is necessary to understand time series. This is tantamount to saying that one must study quantum physics to understand chemistry. This is faulty. Yes there is a subdiscipline of chem (physical chem) which makes extensive use of the underlying physics, but it is in the end a subdiscipline and not nearly the whole.

So in finance terms, you may gain an advantage by taking a full course on stochastic, but it will be relatively small and certainly would not be justifiable for BB S&T (non option). A cursory understanding of stochastic would be sufficient. Probability theory and statistics on the other hand... are a different story.

For sell-side (execution):

If you are going to write an algo, yes stochastic and stochastic optimal control will help you IF it's your job to come up with an execution model. On the other hand, I for example am adding bells and whistles to said model, so my job involves more Time Series.

Personally I feel statistics and intuition to be far more powerful in trading than stochal (again with the exception of options).

 

[pruse]

Yike makes great points and puts this all in a pretty clear light.

Somewhere else on this forum someone advised another member to look into Taleb's take on overconfidence. When people come looking for help on this forum, they certainly don't hope to get a bunch of replies from wanna-be experts with big egos spreading misinformation. They expect genuine responses.

With that said, let's return to regularly scheduled programming...

 

[Mr Doe]

HFT is probably the worst area of finance where young and ambitious professional should be looking to tap in, as discretely mentioned by amanda.jayne (though I disagree on the reason).

 

[Andy Nguyen]

Based on what I've read on this forum and WSO, it seems that programming is a very important aspect, if not, the most important skill of someone planning to enter Quant Finance. With this in mind, could someone explain compare/contrast which program is more advantageous - MFE or MS Comp Sci?

Get a CS degree and become very good with at least one programming language. You can get into finance by being coder genius and get a finance internship.
Finance jobs may come and go with every downturn/presidential election/congress bill/ but you can always find a programming job everywhere.

 

[amanda.jayne]

HFT is probably the worst area of finance where young and ambitious professional should be looking to tap in, as discretely mentioned by amanda.jayne (though I disagree on the reason).

?

 

[Mr Doe]

?

Regulation is going to kill HFT.

 

[amanda.jayne]

Regulation is going to kill HFT.

Quite the contrary - Swaps and other products are slated to go exchange

The regulations will bring changes to HFT, sure. But build a better mousetrap and you will build a better mouse.

I think screen trading would be a shortsighted destination for young ambitious professionals to tap.

 

[Andy Nguyen]

Quite the contrary - Swaps and other products are slated to go exchange

When you think it's going to happen
I heard the talk of CDS going to exchange in 2008 and I'm still waiting for it.

 

[Joy Pathak]

When you think it's going to happen
I heard the talk of CDS going to exchange in 2008 and I'm still waiting for it.

We will find out by July 16th.

 

[Joy Pathak]

Regulation is going to kill HFT.

I doubt that will happen. More and more of secondary market is turning fully automated.

 

[Mr Doe]

Swaps and other products are slated to go exchange

Unfortunately, I don't know much about this issues, so I cant give a sound comment. But I am sure that there's going to be a fierce battle not to bring it to a central exchange/clearing.

I think screen trading would be a shortsighted destination for young ambitious professionals to tap.

Was this ever a good idea ?

I doubt that will happen. More and more of secondary market is turning fully automated.

While it is easy to ping-pong 100 shares back and forth, I doubt one can do the same with, say, bonds.

As we are discussing about it, I am bold enough to say what will end it. My firm belief is that we are going to have another Flash Crash (hey, they can blame Greece again :D) , but this time all of the (smaller) HFT shops are going to stay in the market (remember, May 6th was a great day for HFT). They are either going to take a direct loss, or the circuit breakers/exchange shutdown will lock them with a wrong position. All lot of them will be wiped out, but unlike a year ago, SEC, FINRA, or who ever feels like is in charge, will actually change something. This wont matter too much to the guys who sit on top of the exchanges, but at least we might get a market where a bid doesn't disappear after 50 milliseconds.

 

[Alexander Krosky]

As we are discussing about it, I am bold enough to say what will end it. My firm belief is that we are going to have another Flash Crash (hey, they can blame Greece again :D) , but this time all of the (smaller) HFT shops are going to stay in the market (remember, May 6th was a great day for HFT). They are either going to take a direct loss, or the circuit breakers/exchange shutdown will lock them with a wrong position. All lot of them will be wiped out, but unlike a year ago, SEC, FINRA, or who ever feels like is in charge, will actually change something. This wont matter too much to the guys who sit on top of the exchanges, but at least we might get a market where a bid doesn't disappear after 50 milliseconds.

Do you have evidence or is this pure speculation?

 

[amanda.jayne]

Unfortunately, I don't know much about this issues, so I cant give a sound comment. But I am sure that there's going to be a fierce battle not to bring it to a central exchange/clearing.

Speak of the devil..... An 11th hour reprieve (on some stuff, anyhow)!

http://www.bloomberg.com/news/2011-...aps-measures-as-regulators-take-comments.html

You can see the full act here:
http://www.sec.gov/about/laws/wallstreetreform-cpa.pdf

I would skip to Title IV, Section II: Regulation of the Swap Markets

Was this ever a good idea ?

Prior to August 28th, 2000, sure.

While it is easy to ping-pong 100 shares back and forth, I doubt one can do the same with, say, bonds.

Bonds, probably not. Swaps are liquid enough.

As we are discussing about it, I am bold enough to say what will end it. My firm belief is that we are going to have another Flash Crash (hey, they can blame Greece again :D) , but this time all of the (smaller) HFT shops are going to stay in the market (remember, May 6th was a great day for HFT). They are either going to take a direct loss, or the circuit breakers/exchange shutdown will lock them with a wrong position. All lot of them will be wiped out, but unlike a year ago, SEC, FINRA, or who ever feels like is in charge, will actually change something. This wont matter too much to the guys who sit on top of the exchanges, but at least we might get a market where a bid doesn't disappear after 50 milliseconds.

Or they could just start enforcing FINRA trading rules like backing away and doing away with pegged orders and flash quotes....

There are also those pesky dark pools....