top of page


Intraday trading requires fast tracking and analysis of a perpetual stream of ever changing price movements.   Multicore Dynamics embarked on perhaps its most ambitious project todate,  to evaluate whether novel analysis techniques together with Machine Learning could be utilised to predict price action and to what extent this would translate into profitable trades.


Target : to design and build an AI based algorithmic trader. 

To start the process,  certain key ingredients were sought :


  1. A willingness to learn from the ground up

  2. Massive amounts of financial data

  3. Machine learning knowhow

  4. Novel analysis techniques

  5. Lots of processing power

  6. Stamina


Determination proved to be a key requirement as the path to progress,  from learning to development, to evaluation and delivery of rresults would prove arduous.

Image by DeepMind
Image by DeepMind


As is the case with most complex data,  we soon discovered that trading data would also require much pre-process filtering in-order to detect meanginful signals. 


Evaluating and applying the right machine learning methods would necessatate countless hours of optimisation and further fine tuning if our AI virtual trader was to detect patterns and correlations from within the data.  


Our AI trader monitors and detects signals arising from a live stream of market data.  From this, buy and sell instructions are issued according to patterns found within the processed data.  

STATUS : Evolving.

Image by julien Tromeur
bottom of page