The fund, which is being run by a team of five Harvard University physics and astrophysics graduates, will use an approach termed ‘DarwinFX’. It will use a range of trading models that work as a unit, allowing it to efficiently analyse large datasets. "It’s almost impossible to build a single model that works across a large data set because there is a multitude of data and variations, so we build a set of smaller models that work as a group - each segment is specialised in a specific task," said Aaron Sokasian, a founder of the company and manager of the research team.
This approach allows the model to be very adaptive as those models that prove ineffective in managing money can be removed while the surviving models can be applied to new data, he added. "We can assess the performance of each model and only allow those that are successful to continue," Sokasian said. "The models will get better at managing inefficiencies in the financial markets - it is essentially survival of the fittest."
The system also allows components to ‘interbreed’ and some random models are also introduced, said Sokasian. "This means the programming language remains diverse and can adapt as market ineffiencies can obviously change form over time."
Financial Labs will launch with funds of $20 million under management - $17 million of which is in commitment and $3 million in capital. It plans to apply its investment approach to futures and equities over the coming months, and will also offer hedge funds at a later date.