Artificial Intelligence for Trading : Course Review

Udacity was offering the nanodegree Artificial Intelligence for Trading free for a month. This was too good an offer to not sample the course. The expected time to complete the course is c.6months @10 hours a week. I just zipped along as fast as I can with focus on finishing the maximum amount within 1 month.

Pros:

• Exhaustive course with a lot of material, code and practical projects to do. The course, as expected of a nanodegree, is detailed and deep dives into a lot of theoretical, practical and technical background
• Introduces you to Factor Investing and how to use AI for it
• Good course for people looking to work in industry as gives detailed understanding of what being a quant involves
• Like the ISB course I finished earlier, this course highlights the importance of academic research papers in the quest for alpha
• Provides an excellent starting point and references, which I think a reasonably informed participant could build upon as they gain more experience

Cons:

• The modules in both terms start as too easy and then suddenly the difficulty level goes through the roof somewhere after the midpoint. The course would anyway be extremely hard for people with no background in technology and finance, so an intermediate level start gradually building up the difficulty could be better
• The projects that use zipline at end of term 1 have build problems, something I saw was very common in discussion boards. I could not complete the fourth project simply because it would not build and (given the speed of my review) I had no time looking for solutions. Given the course is pricey at $300/month, I would expect the assignments to work with minimal problem
• Very less practical application for retail traders looking to set up their own operation as it is more targeted towards an audience aspiring to become a Quant in the industry. Also, while the content is reasonably good, I think you will be able to find similar or better content cheaper elsewhere.

A lot of exercises in the second half of term 1 use zipline, which I reckon is the key reason behind the errors I referred to earlier. Further, to use it independently you may need to buy/ subscribe to a data source and spend time understanding zipline. That does not take away the usefulness of the lessons though. If you are not a fan of zipline, a quick Google search shows some open source alternatives to zipline: Ultra Finance, PyAlgoTrade, BackTrader.

Term 2 covers quite some ground. Unfortunately for course content creators advances in web technology means that you can pretty much not send requests from a computer to a website like before. A few lessons are outdated due to this simple reason.

Overall, a reasonably good course if you want to be a Quant, but not so for a retail trader.