October 28, 2013

Introduction to the business startup Quantopian

Founded in 2011 by John FawcettQuantopian is not the first business startup with the aim to reinvent algorithmic trading. We have earlier covered EquaMetrics, and they have developed a web application to make it easier for amateurs to use algorithms when trading. When developing trading algorithms, you need to know how to program the software yourself, or hire an expensive software developer. If you are using EquaMetrics, you don't need to know how to develop software because of the drag-and-drop interface. If you are using Quantopian, you have to know the programming language Python. You can't neither trade as frequently as with EquaMetrics that allows you to trade up to 3,000 times per second. With Quantopian, you can trade each minute.  

So the basic idea is that you build you own trading algorithm, or you can modify an already existing trading algorithm provided by Quantopian and other users of Quantopian. When you have an algorithm, Quantopian lets you test it against an 11-year history of minute-level stock data. Notice that only US stocks are available, but it may change in the future. Are you satisfied with the algorithm? If so, you can test it live against the real market. You also have the option to share the algorithm with other users, but you don't have to. It's currently impossible to make real money with Quantopian, but that feature will be available in the future - "Live trading is currently limited to a group of pilot testers." If you need a deeper explanation, you should watch this video: Quantopian Introduction and Tour Webinar.

It's currently unknown how Quantopian plans to make money. According to PandoDaily, the first goal is to grow a strong community of algorithmic traders.

Why Quantopian is a good idea
  • If you are trading, you have to make sure you have a good strategy. Most amateur traders tend to jump between different methods, such as "sell below MA200," and they have no idea if the methods are profitable. With Quantopian, it's easy to find out if your strategy is really profitable. 
  • Ed Seykota was one of the pioneers of algorithmic trading. One of his customer's account started with $5,000 in 1972 and had made over $15 million as of mid 1988. So algorithmic trading can be very profitable. 
  • One good thing with algorithmic trading is that you can't make stupid trades based on your emotions. If the computer tells you to buy, then you have to buy. 

Why Quantopian is a bad idea
  • You can test your algorithms against an 11-year history of minute-level stock data - but 11 years might be a too short time. 
  • It could be difficult for amateurs to compete with other professional algorithmic traders. To make faster trades, you have to be very close to the computer that executes the trades, and to be able to be close to that computer, you need to pay lots of money. 
  • One problem is fees. Each time you buy or sell a stock, you have to pay a fee to the bank. With algorithmic trading, the number of orders will increase and the fees might eat up your profits. Large professional banks can pay lower fees, but will amateurs be able to pay the same low fees?