Does anyone know of an online course focusing on big data analysis for investing/trading purposes?
The only ones I have seen are for investment banking, equity research, or other specific company analysis. I’m looking for a good overview of how to utilize different economic indicators. ex: when US started QE, how to calculate the effect of the QE on the US stock market. Or if consumer expectations misses by .2%, what is the expected effect on the S&P?
What you want isn’t “big data”, it’s just normal statistics. Just find a good statistics class first. Learn about time series econometrics too.
I would define “big data” as so big that it needs to be distributed across multiple servers. So we’re talkin’ mapreduce/hadoop/etc. I’ve never had this issue. Tick data might be one case where you could use those techniques, but other than that, I don’t know. The most I’ve ever had is data big enough that I need to make SQL queries instead of loading it into memory.
Thanks jmh. I think you’re right, I’m basically looking for a good econometrics course/overview. The course/site/book should have examples of real economic indicators and the correlation to the market. Basically a factor model…i would think large hedge funds choose macro allocations with one extremely complex factor model and I would like to attempt to build something similar.
over time it could turn into big data but obviously managing that amount of data would cost some money to handle the server costs
I hate data driven factor models. Yuck.
What are you going to do with these econometric models?
^ Create a fund that generates 50%/year returns then blows up spectacularly causing systemic damage everywhere? I’d only assume.
I have a feeling 1997 wants its LTCM strategy back.
yes i do know a few courses. i studied econometrics back in college. prolly my most favorite class besides money and banking. i actually did a lil bit of sas at a consulting firm, getting the output stuff is easy. the hardest and most time consuming part. figuring out what data to analyze. finding a way to get that data. inputting that data. then cleaning up that data (quality check). so it can be analyzed. btw sas coding is the easiest of the bunch.
having said that there are a few investor that have done this with big data similar to fama french. i juss saw a stock-specific factor analysis from jim oshaughnessy yesterday. he basically used basic value metrics that he found in stocks ranging from 1930s? to now and it showed pretty convincing alpha. As for macro factors i’ve seen a few about PE/ multiples and inflation saying they are invesely related. shiller pe and long term performance inversely as well and of course no correlation for short term. etc. etc.
anyways, i wanna do this quant stuff in the future. i’ve seen enuff research showing that it works. its def easier than qualitative analysis in the long run. Short run it’ll prolly require a lot of work to get the data in. its less reactionary. its more rule based. and prolly less reading. not to say that reading is bad. but htere is so much things to read nowadays that it is either garabge, or repetitive stuff. when you have a large set of data though, you can fact check ppl’s bs that they say on cnbc or you read in a report.
I am enjoying the discussion here but still no courses are introduced for time series /econometrics
LTCM fell apart due to treasury bets…basically leveraging up something like 100x (I forget the multiple) and trading the spread on on the run vs off the run Treasuries. When Russia crashed they fell apart.
I’m not sure how a factor model along with bottom up company analysis would be worse than just bottom up analysis without a clear understanding of the economic cycle/current economy.
I’d like to know the effect of certain indicators on the stock market/bond market. An example would be: why does a 2% surprise in the nonfarm payrolls number lead to a .01% increase in the stock market?
I suppose in this instance I can look at all nonfarm payroll numbers, expected and actual, vs before and after market cap of the market. This seems like the best way to go about creating a model…I was just wondering if anyone had any other tips/websites that would help find useful indicators (st louis fed, shiller p/e, s&p, etc.), or a class/book that shows you specific websites that have data and then describe how to relate it to the market.
I’ve never used sas but it sounds very useful–I know excel extremely well along with some programming languages. I may check out sas next.
There are a few econometrics courses on Coursera.
You could also review L2 quantitative methods. I’m pretty sure that everything you have described is in the texts. A class might be a better way to approach however.
Yeah…I’ve been hesitant to re-open those books…!
thanks raw raw…I’ll take a look at the SAS course and the econometric course too.
@Stockton_Malone, I’d recommend learning R before SAS.
@Gabelogan Econometrics is basically just statistics for economists. There are a few things that economists deal with more often than more general statisticians do (instrumental variables for one). A first course in econometrics is very similar to a first course in statistics. Lots of overlap. I’d recommend getting a solid foundation in regression (using matrices). In particular, you should understand the assumptions of OLS and what to do when they are not met. Only after getting the foundation should you move on to time series.
Any reasons why r over sas? I haven’t looked at any videos but it seems like sas is a simple version of SQL while r is used for complex math calculations of large data sets.
also anyone have any other good sources for economic data? I named a few above and was curious if anyone had others.
^This has probably been discussed to death. I’d say it depends on your needs. I would guess that most people on AF would be better off learning R first then SAS only if they need it (and someone’s paying for it). There are database and big data options for both, though you might get better support with SAS. If your company is paying for SAS, then by all means learn that first. The benefit of R is it’s free and it has a big community.
SAS® University Edition Free SAS® software. An interactive, online community. Superior training and documentation. And the analytical skills you need to secure your future.
http://www.sas.com/en_us/software/university-edition.html
There are a few quirks since it runs in a Virtual Machine in terms of referencing files on the computer, but it is a good free way to learn.
Your model won’t tell you anything different than what the general market consensus is and the odds are you aren’t going to do it as well as someone who does it for a living- no offense intended. So yes, it could absolutely be worse.