I planned to set up my own website about automated trading long time ago, since I deeply believe that automated trading is a good application of AI and Machine Learning. But because I am too lazy, I didn’t start. Recently, I got inspired by Mr. Loveless from Cantor for his passion and insight. So, I decide to go …

As a researcher in intelligent systems, I am supposed to concentrate on my professional research. And I really know I should not be distracted. There are some questions like: why do you do something about trading? how did I get started in automated trading? Let me tell you my story.

First of all, I have a broad range of interests. I believe that different challenging applications can advance the development and innovation of AI theories. I always keep close eyes on different applications of AI techniques. Now and then, I read many papers about AI and machine learning for computational finance. I initially doubted the results there very much. Since I was told the efficient market assumption and the random walk thing.

Then, in 2007, I happened to find a competition hold by Interactive Brokers (IBG). The competition was open to any student in North America (now the competition is open to global students). I wanted to try and verify whether AI/machine learning for automated trading works or not. I wrote a proposal and jumped …

After about 5 evenings work, I completed the codes and put my algorithm to the competition. Surprisingly, my automated trading system worked well. http://www.interactivebrokers.com/en/p.php?f=olympiadWinners&ib_entity=llc In 2007, I won 3rd place prize and $10,000 award with 40.9% gain over two months. 

My 2008 year was quite busy. I wanted to graduate. But I was still a student in Spring 2008 and thus eligible for 2008 IBG Collegiate Trading Olympiad Competition. I didn’t change my algorithm much and put it again to the competition. Unfortunately, I just got an OK performance. This was mainly because my current algorithm is a prototype and could not well tackle big and sudden market drops, such as the Nasdaq loss during Feb.-March, 2008. But my algorithm was still kind of robust and helped me win $1000, making slightly better profit than bank saving and definitely beating the market of that time.

Now you may see how I got interested in automated trading and how my experience promotes my confidence in AI for finance. In another aspect, the more I understand the complexity of market data and its dynamics, the more deeply I believe the market makes an excellent research benchmark for AI and machine learning studies.

I will keep updating this website. Check me out now and then …

Chenghui Cai @Duke