Computational Investing 1 --- Georgia Tech

S

slacker

I sat through this class when it was first offered last year but did not take it seriously. Taking it again-this time I plan to work.

If you want an intro to programming and investment analysis this would be a good start. The course combines Excel and Python to download a portfolio of stocks and then measure the implied risk of each combination of stocks/ETFs. My background is in Java and C++ so learning a little about Python was fun. I have also been wanted to do longer term trading instead of the short term stuff I have focused on for the past few years.....

Take a look at the course ---- share your experience here.... Learn something new!!!

It is all free and part of the coursera.org collection of courses.


https://www.coursera.org/course/compinvesting1

Overview
Why do the prices of some companies’ stocks seem to move up and down together while others move separately? What does portfolio “diversification” really mean and how important is it? What should the price of a stock be? How can we discover and exploit the relationships between equity prices automatically? We’ll examine these questions, and others, from a computational point of view. You will learn many of the principles and algorithms hedge funds and investment professionals use to maximize return and reduce risk in equity portfolios.

Topics
We start with a tour of the mathematics and statistics that underlie equity price changes, and the relationships between different groups of equities. We’ll review the most important economic theories of investing and how to create programs that take advantage of them. We’ll look at the data needed to do this, and how to manipulate it effectively. Take a look at the course syllabus here.
 
It doesn't look like anything you could not pick up off the internet if you were suitably motivated...CAPM & Bollinger Bands?
 
All right, Slacker. Thank you for the cogent reply. I'm sorry you feel that way. I'm really not a malicious or vindictive person. I also apprieciate the link to the course . The most cogent thing I have learned is that the market is full of deception. Things are not as they seem. Think about this. Regards, surf
 
Quote from Craig66:

It doesn't look like anything you could not pick up off the internet if you were suitably motivated...CAPM & Bollinger Bands?
More...

John Bollinger himself states the bands should only be used in connection with other analysis. He doesn't even call what he does technical analysis but rather rational analysis.

Peace. Surf
 
The class first week should be getting the software needed up and running. All software is open source and free.

This is the third time the class has been presented so it is much improved over a year ago. In the first class offering over 4000 students enrolled from all over the world. The discussion forums were amazing with comments about alternatives to the approach Tucker took in class. Some of the students where managing large accounts and had some real good comments and suggestions. For me, the student forums were as informative as the formal videos.

The discussion forums are also a good place to get help with Python or the QSTK package. Unless you are managing a real fund it is hard for me to imagine needing more than this package. You could spend thousands and not do better IMHO.....

From Week1 class notes:

QSToolKit (QSTK) is a Python-based open source software framework designed to support portfolio construction and management. We are building the QSToolKit primarily for finance students, computing students, and quantitative analysts with programming experience. You should not expect to use it as a desktop app trading platform. Instead, think of it as a software infrastructure to support a workflow of modeling, testing and trading.
Portfolio optimization: Using the CVXOPT library.
Event studies: An efficient event analyzer, Event_Profiler.
Simulation: A simple backtester, quicksim, that includes transaction cost modeling.
 
Quote from Craig66:

It doesn't look like anything you could not pick up off the internet if you were suitably motivated...CAPM & Bollinger Bands?
More...

CAPM & BBands? LOL.

The name GIT wreaks mook.
 
Quote from slacker:

The class first week should be getting the software needed up and running. All software is open source and free.

This is the third time the class has been presented so it is much improved over a year ago. In the first class offering over 4000 students enrolled from all over the world. The discussion forums were amazing with comments about alternatives to the approach Tucker took in class. Some of the students where managing large accounts and had some real good comments and suggestions. For me, the student forums were as informative as the formal videos.

The discussion forums are also a good place to get help with Python or the QSTK package. Unless you are managing a real fund it is hard for me to imagine needing more than this package. You could spend thousands and not do better IMHO.....

From Week1 class notes:

QSToolKit (QSTK) is a Python-based open source software framework designed to support portfolio construction and management. We are building the QSToolKit primarily for finance students, computing students, and quantitative analysts with programming experience. You should not expect to use it as a desktop app trading platform. Instead, think of it as a software infrastructure to support a workflow of modeling, testing and trading.
Portfolio optimization: Using the CVXOPT library.
Event studies: An efficient event analyzer, Event_Profiler.
Simulation: A simple backtester, quicksim, that includes transaction cost modeling.
More...

So out of the 4k students who took the course, how many are making money?
 
Quote from marketsurfer:

All right, Slacker. Thank you for the cogent reply. I'm sorry you feel that way. I'm really not a malicious or vindictive person. I also apprieciate the link to the course . The most cogent thing I have learned is that the market is full of deception. Things are not as they seem. Think about this. Regards, surf
More...

Really? I thought markets could be modeled and programmed in a nice neat package.

Gee, you bursted my bubble. I think I am gonna cry.
 
I found the best thing from the course is the QSTK package. The event study is very interesting. It draws the path of a stock from 10 or 20 days before your signal to 10 or 20 days after the signal -- but for as many stocks and instances as the signal is found. If you draw all the paths, it looks like spaghetti tied together in the middle; but the QSTK can switch the mode so it replaces that with bars of 1 standard deviation as the paths expand from the signal point. It kind of gives a picture of after the signal if things generally go up or down or flat on average. There is a better description on the quantwiki site. With a bit of Python programming, you can add to it, to make your own features or draw different kinds of studies. I've got it to draw a graph of a portfolio, to explore changing the weightings of the portfolio components. There are also a lot of features I haven't been able to explore, like some of the machine learning, I suppose they may be covered in the Part 2 and 3 of the course.

It is somewhat difficult because it is a large software package, and you are refactoring some parts, and learning other features it has. I've found, if I use some of the features in the package, I have to dig into the code to figure out what it is doing or how it is calculating something. Or sometimes I get an error with no simple message, and I have to figure out where that is happening in the source code. It was confusing at first because some folders are working scripts, some are source code, and it took me a while to figure out which were which, just took some getting used to. Sometimes that is a drag, but of course one always has to do some work.

I hope the course is improved from the Oct 2012 offering. In that run, several of the video lectures were not posted until late after the schedule, and it the mood on the forums was not great after that. But I stuck with it, and I hope they've been able to fix those issues, I think the professor has a lot to offer interested students, and I always want to focus on good positive outcomes.

There is also the point of using Linux and not Windows. I've used it on Windows with no problems, and didn't use the ubuntu virtual image. One poster noted all the variables to add to the class path, and there is a substitute python class for one of the solvers that will work on Win.

Good luck, I hope you enjoy the course, and would be great to hear your experience with the QSTK.
 
Quote from SteveNYC:

Really? I thought markets could be modeled and programmed in a nice neat package.

Gee, you bursted my bubble. I think I am gonna cry.
More...


You Are not following brother. I was referencing slackers total belief in how PTJ and the others say they trade. The markets a jungle and deception is how the predators feed.
 
You must log in or register to reply here.
Back
Top