Although a relative newcomer to the world of algorithmic trading, QuantConnect has been making inroads into increasing the functionality of its algorithm backtesting platform.
This week, the firm added 15 years worth of US equities and 5 years worth of FX tick data to its offering in order to facilitate ease of coding, testing and fast iteration by independent software engineers.
Virtual Infrastructure Model
QuantConnect considers itself to be contributing to the process of democratizing algorithmic trading by giving engineers access to free financial data, powerful cloud computing and strategy back-testing.
“We launched QuantConnect with the goal of bringing algorithmic strategies to the mainstream engineer and investor, so they can be empowered with cutting edge investment strategies,” stated the firm’s CEO Jared Broad on launching the tick data.
“By providing a platform with unlimited free financial data, we’re giving access to tools that would typically cost $50,000-$100,000” he said.
QuantConnect has built an environment which can complete back-tests in less than two minutes. “This product is way ahead of retail options currently available,” said Alejandro Cañete Báez, advisor and Head Quant at Pan Alpha Trading.
The actual tick data concerned has been integrated into the platform from a wide array of sources to enable a diverse range of user-built models.
Forex data has been sourced from FXCM, sentiment data from Estimize and StockPulse and made available on a single platform the idea behind which is to enable users to bring algorithms from conception to reality in minutes rather than days.
Playing A Part In The Arms Race
Big Data, News, Sentiment Analytics & NLPGo to article >>
Algorithmic trading has become the focus of regulators and technology firms alike recently, with many of the leading organizations engaging in a very expensive and fast moving arms race in order to ensure that their technology operates more efficiently than the competition.
In some cases, anonymity is preferable in order to avert the eyes of competitors and regulators, resulting in dark pools joining large trading communities to connect them to venues.
The speed at which this is occurring is increasing, and there are now many institutional organizations which are placing emphasis on this. Examples of this are on a large scale, whereby microwave technology is being used to provide high speed connectivity, and large software firms have connected to infrastructure providers to increase the speed of connectivity worldwide, all of which is aimed at high frequency trading where latency is critical to the nature of such trading.
Added to that, last year’s disastrous incident at Knight Capital in which a test algorithm executed a trade on the production server resulting in $440 million in lost trades and severely damaging the firm’s solvency served as a reminder as to exercise caution with regard to electronic methods of trading.
Despite dissent from regulators and some technology providers, the race to nanosecond-level latency is in full swing and the infrastructure now demonstrates this ability.
In order to make full use of today’s infrastructure and for an algorithmic trader to continue to measure his advantages and disadvantages against other participants, the algorithms which execute the trades at such speed require accurate testing, which is where such systems as QuantConnect fit in.
The company provides a GIT API for collaborating in teams, and allows upload of encrypted algorithms to keep intellectual property safe. For the first time engineers have unfettered access to a decade’s worth of trading data for back-testing, without giving away any valuable intellectual property.
“We are building a global network of engineers designing diverse new strategies while helping investors across the world earn better returns with the strategies that suit them,” said COO, Shai Rosen.
“We are working hard to make our vision a reality by building partnerships with data providers, listening to our Quant community and continuously adding features to our platform.”
As far as algorithmic development is concerned, an engineer can find correlations, build an algorithm and test his results in less than 20 minutes.