Algorithmic Trading and Cloud Computing
Posted by: Brad
Tags: algo trading, algorithmic trading, high frequency trading
Ever since computers were first used in the stock markets in the 1970s and 1980s, trading and brokerage firms have continuously adapted, modified and improved their trading systems to the extent than the majority of all volume traded in US stock markets is now computer-driven.
Algorithmic trading (or “algo trading” as it has become known) is now the dominant force, particularly since virtually all exchanges and trading venues now use electronic matching systems rather than any kind of face-to-face dealing environment.
However, in order to stay ahead of the game, traders and brokers have had to participate in a kind of “arms race”, deploying faster and faster systems with ever more powerful hardware and more complex software, in a bid to realise advantages that their competitors don’t have. This is a very expensive proposition. It can be particularly expensive when you are developing and testing out a new algorithm before using it to trade in the live market.
To try to keep some of the costs down, market participants are increasingly looking at cloud computing, i.e. the sharing of remote computer resources. How can cloud computing be used in algo trading, you ask?
Well, the idea is that the computer programs are hosted on remote servers and accessed via the Internet on demnd, so that firms can test out new trading models and investment strategies on a kind of ad-hoc basis, thus saving on the cost of implementing the hardware and software locally.
Connectivity into exchanges and other liquidity venues is also something that could be managed remotely by a third party and accessed when needed by trading firms. Again, this kind of infrastructure is expensive to implement but if it can be accessed on demand via the Cloud, on a pay-as-you-go basis, the return on investment for a partiuclar trading strategy could be much higher.
So is the future of high frequency and algorithmic trading in cloud computing? Time will tell.