Your thoughts - Columbia MSOR or MA in Stats?

K

kapo

New member
Joined
Jun 18, 2026
Messages
0
Reaction score
0
So my company has a generous tuition reimbursement scheme and i have been accepted to both MSOR and MA Stats at Columbia as part-time candidate. Anyone has any thoughts on either program? Personally i prefer msor but my boss told me to go for the stats program. But I think the Stats program looks a little light (8 courses) as oppose to the OR one (10 courses). On the other hand, Stats does seem more relevant to what I am doing and MSOR seems to have a bad rep because you can earn the degree online.
Substantively, i don’t think there are a lot of difference between the 2 programs since one can take courses from either department as electives.
Not considering a full-time degree now because 1) I’m up for promotion and 2) I’m in prop trading at a BB. Graduated from UChicago in econs and math.
 
Maybe I take that back - Why do you want to do either of them? Statistics is not especially useful in prop trading and the courses you get in this program are pretty light weight data analysis courses. You could pick up women psych students all day long and help with their dissertations but not much of that will have applications in finance (where stochastic processes and computational statistics seem to be much more valuable).
I’ve never been much of a buyer on OR, although some of it is certainly cool. Learning about scheduling, quality control, logistics, etc. just doesn’t help anything in finance.
And, of course, the problem with both of these is that the world is awash in people with Ph.D.’s in them.
Columbia has that great MFE program. Why not take that?
 
Joey,
statistics is actually very useful in prop trading, especially in algo (aka technical analysis) trading strategy, which is what i am doing. Time series analysis, and advanced regression (lasso, ridge etc) are very useful techniques to know. That’s why i considered the MA in Stats.
As for the OR, it is not only about scheduling, quality control and logistics. In fact, the columbia msfe that you mentioned is housed in the ieor department. If OR has no relevance to finance, then who’s the bigger fool housing the msfe program at ieor department here, you, or columbia? And in fact, schedulin does have some relevance to finance. Optimal trading strategies are, afterall, a schedule of buy, sell and hold.
i’m not too keen on columbia’s mfe program because i did earn my bachelor’s degree in math and econs, and the last time i checked, uchicago still has one of the best econs department and financial math program, and i took a lot of those courses in my junior and senior years (and yes i took a bunch of stochastics courses - processes, sde, methods in finance). Why would i want to do a “light-weight” msfe now? Plus it is full-time, which i explicitly mentioned that i am not interested in.
As for PhD, don’t see why i need it since i’m working in an area where most phds-who-hopes-to-turn-trader would kill to be in.
volkovv,
Thanks dude for the suggestion. I did consider the program you mentioned and NYU’s msmf. But as I mentioned before, I have actually taken most, if not all, of the financial math courses, so I’m not sure how useful those programs will be. In mathemtical finance, if one had taken courses in stochastic processes, pde to finance, sde to finance, martingale method to finance, numerical mtd to finance and some finance courses, then there is really very limited to what a master in financil math/engineering can offer, except for the brandname, which i really don’t need.
I’m actually looking for more useful and applied stuff like time series analysis (or any course that will go through RS Tsay’s book in good coverage; i had wanted to take his classes at chicago, but didn’t have the chance), advanced regression techniques that deals with autocorrelation and stuff, and some statistical learning techniques.
 
kapo Wrote:
——————————————————-
> Joey,
>
> statistics is actually very useful in prop
> trading, especially in algo (aka technical
> analysis) trading strategy, which is what i am
> doing. Time series analysis, and advanced
> regression (lasso, ridge etc) are very useful
> techniques to know. That’s why i considered the MA
> in Stats.
>
Yeah, well how come my Ph.D. in stats has never been particularly useful in trading? Maybe if you study it for awhile, you will see why time series analysis is nearly useless in trading (Lord knows, I wish it wasn’t).
>
> As for the OR, it is not only about scheduling,
> quality control and logistics. In fact, the
> columbia msfe that you mentioned is housed in the
> ieor department. If OR has no relevance to
> finance, then who’s the bigger fool housing the
> msfe program at ieor department here, you, or
> columbia? And in fact, schedulin does have some
> relevance to finance. Optimal trading strategies
> are, afterall, a schedule of buy, sell and hold.
>
Where they house the program probably doesn’t have much relevance to anything (except you can some of the courses). The core curriculum there is just standard issue OR. I think OR is a fine thing to study and there is some great stuff there; I just don’t think it’s very useful in finance.
As for optimal trading strategies being a schedule of buy, sell, and hold about the only application I have ever seen on Wall St for something that resembles OR is portfolio management (mean/variance stuff is quadratic programming) and some applications of dynamic programming in stat arb.
>
> i’m not too keen on columbia’s mfe program because
> i did earn my bachelor’s degree in math and econs,
> and the last time i checked, uchicago still has
> one of the best econs department and financial
> math program, and i took a lot of those courses in
> my junior and senior years (and yes i took a bunch
> of stochastics courses - processes, sde, methods
> in finance). Why would i want to do a
> “light-weight” msfe now? Plus it is full-time,
> which i explicitly mentioned that i am not
> interested in.
>
> As for PhD, don’t see why i need it since i’m
> working in an area where most
> phds-who-hopes-to-turn-trader would kill to be in.
>
>
> volkovv,
>
> Thanks dude for the suggestion. I did consider the
> program you mentioned and NYU’s msmf. But as I
> mentioned before, I have actually taken most, if
> not all, of the financial math courses, so I’m not
> sure how useful those programs will be. In
> mathemtical finance, if one had taken courses in
> stochastic processes, pde to finance, sde to
> finance, martingale method to finance, numerical
> mtd to finance and some finance courses, then
> there is really very limited to what a master in
> financil math/engineering can offer, except for
> the brandname, which i really don’t need.
>
> I’m actually looking for more useful and applied
> stuff like time series analysis (or any course
> that will go through RS Tsay’s book in good
> coverage; i had wanted to take his classes at
> chicago, but didn’t have the chance), advanced
> regression techniques that deals with
> autocorrelation and stuff, and some statistical
> learning techniques.
 
Joey,
I’m sorry that you cannot find any use for your phd work in stats. but i’m sure you appreciate that that is not my problem, nor am i interested to know about it.
it is really a pain reading your lazy reply. please, if you can, refrain from posting in my thread. I’m sure you have a lot to say, but i’m really not interested.
 
kapo Wrote:
——————————————————-
> Joey,
>
> I’m sorry that you cannot find any use for your
> phd work in stats.
Read more carefully. Very important in math/stat.
> but i’m sure you appreciate
> that that is not my problem, nor am i interested
> to know about it.
>
> it is really a pain reading your lazy reply.
I can teach you finance, math, stat, OR and even reading.
>
> please, if you can, refrain from posting in my
> thread. I’m sure you have a lot to say, but i’m
> really not interested.
 
whatever, just stop behaving like a leech and go away. Do you have no friends to talk to?
 
gents, let’s keep this civilized please.
- Joey, could you elaborate on why time series is not applicable to finance ? Since you have worked in industry for more than a decade, I would certainly like to hear your opinion.
- kapo, the same question as Joey.
JoeyDVivre Wrote:
——————————————————-
> kapo Wrote:
> ————————————————–
> —–
> > Joey,
> >
> > statistics is actually very useful in prop
> > trading, especially in algo (aka technical
> > analysis) trading strategy, which is what i am
> > doing. Time series analysis, and advanced
> > regression (lasso, ridge etc) are very useful
> > techniques to know. That’s why i considered the
> MA
> > in Stats.
> >
> Yeah, well how come my Ph.D. in stats has never
> been particularly useful in trading? Maybe if you
> study it for awhile, you will see why time series
> analysis is nearly useless in trading (Lord knows,
> I wish it wasn’t).
> >
 
I didn’t say that time series analysis was not useful in finance; only that it is not useful in trading. The reason that it is not useful is essentially that most data streams in trading (i.e., price streams) are either Markov (in which case there is no use for past data) or are so badly behaved in the sense that they are non-linear, non-stationary, and filled with jumps that they are intractable by most time series methods. In particular, when lots of students learn about ARIMA and spectral methods it occurs to them that the data sets they use in class look a lot like stock data and they just discovered this money machine. That’s a useful and good stage, but it doesn’t work.
You could probably use time series analysis in finance for things like predicting seasonal demand for a product, inventory levels of something, etc. but this is all second-order kind of info that is, e.g., very fundamental equity analysis (i.e., several steps removed from any trading).
 
JoeyDVivre Wrote:
——————————————————-
> I didn’t say that time series analysis was not
> useful in finance; only that it is not useful in
> trading. The reason that it is not useful is
> essentially that most data streams in trading
> (i.e., price streams) are either Markov (in which
> case there is no use for past data) or are so
> badly behaved in the sense that they are
> non-linear, non-stationary, and filled with jumps
> that they are intractable by most time series
> methods. In particular, when lots of students
> learn about ARIMA and spectral methods it occurs
> to them that the data sets they use in class look
> a lot like stock data and they just discovered
> this money machine. That’s a useful and good
> stage, but it doesn’t work.
>
> You could probably use time series analysis in
> finance for things like predicting seasonal demand
> for a product, inventory levels of something, etc.
> but this is all second-order kind of info that is,
> e.g., very fundamental equity analysis (i.e.,
> several steps removed from any trading).
uhh Jim Simon. The very rare interview i recall he indicated he was using time series & predictive models for trading. The problem was he couldn’t just set it and leave it. They always had to tweak the models as others discovered the same relationships. Thus, one must be ahead of the game, not late to the party… last I checkeded they were up 35%+ ROR net (after fees of 5&44)
 
I have enormous respect for John Simon but I’ll bet they aren’t using anything that resembles traditional time series models. Just because he said it in an interview doesn’t make it so. (I’ve certainly known what other hedge fund managers were doing and what they said in an interview and they weren’t even close).
 
You must log in or register to reply here.
Back
Top