I am professor of Computing and Systems Engineering at the National University of Colombia in Bogota and I been using Quantopian as teaching tool for the past two years in the undergraduate Algorithms course and this semester in the graduate Machine Learning course.
I have been using online competitive coding-programming and algorithms challenges like TopCoder, Codechief, SPOJ, Hackerearth and Hackerrank as teaching tools in the algorithms course for the past 8 years, this contests include educational materials that are very useful for students and allow them to test and train their skills with easy to hard problems competing with peers around the world. Students get involved because they know that many top companies are using these challenges as part of the tests in their hiring process and also sponsoring competitions with prizes as a way to scout for talent.
Two years ago I decided to introduce the Quantopian contest as part of the algorithms class. These are 6th semester undergraduate students with good programming skill but little financial background so I gave the students a brief introduction to trading and trading algorithms; and I included the participation in the Quantopian contest as an extra credit project. I recommended them to explore the examples of trading algorithms that are available in the educational material and the contributions blogs, pick one they could understand and try to improve it with any reasonable modification. Students were surprised by the quality and quantity of educational resources and contribution posts available in Quantopian. The response of students was surprising, even though the project was optional; almost all of them submitted entries to the contest, some of them let me know they were very really excited about possibilities of making a living developing trading algorithms and that they wished they could put more time to develop a really good algorithm. The following semester I decided to include the development of a simple trading algorithm that qualifies to the contest as the class final project. To my surprise many students continued developing algorithms in Quantopian after finishing the course and even some won contest daily prizes. Given their interest we started a Bogota Algorithmic Trading Meetup, to meet and learn, that now has some 370 member including students form other local universities and some practitioners.
Given the positive experience in algorithms, I decided to include this semester as final project in the graduate Machine Learning class, the development of machine learning based trading algorithm that qualifies to the contest in Quantopian. These are master and doctoral students from Computing and Systems Engineering, Statistics, Economics, Mathematical Finance and some few advanced undergraduate students; the course requisite was basic programming skills. I also gave them a brief introduction to machine learning based trading algorithms presenting "Analysis of a Naive Bayes High Low Return Predictor using Previous Returns" using previous returns based on the great post “Machine Learning on Quantopian ” by Thomas Wiecki and a "Naive Bayes High Low Return Prediction Algorithm" that is modification of the Simple Mean Reversion Algorithm based on the great post “How to Leverage the Pipeline to Conduct Machine Learning in the IDE” by Jim Obreen.
If any of you have used Quantopian as and educational tool I will be very happy to hear about your experience also if you know other resources about machine learning based trading algorithms that are not listed here please let me know