Hasan Amjad, head of algorithmic trading at GAM Systematic Cantab, explains how machine learning tools and techniques have enabled his firm to improve almost every aspect of its trading capabilities.
“It goes all the way really,” he says, “Starting with portfolio construction, all the way to the final trade and the post-trade analytics.”
For example, Amjad points out that machine learning can be used to improve pre-trade analytics by more effectively identifying what kind of trading the firm should be engaging in during current market conditions. He acknowledges that there are other techniques that enable firms to determine market conditions, but that “machine learning just takes it that one step further by being able to ingest a lot more data and give you the answer”.
Not only does machine learning enable trading firms to consume more data, but also more types of data. By using machine learning to consume unstructured data sets, Amjad says that it is possible to expand the scope of alpha generation beyond just average asset return time series models.
“For instance, you have people looking at satellite imagery of the shadows cast by oil tanks, the number of cars in parking lots, weather patterns in forecasting data, so in that sense it has broadened the types of models that you can build and that of course leads to more diversification in your alpha,” he explains.
Further, Amjad claims that machine learning can also be used to discover and exploit patterns in existing systems that could not have been detected by standard statistical techniques. This is why the technology can also be used to help improve execution and trade analytics, rather than just for alpha generation.
But for all the benefits that machine learning can bring to trading firms, Amjad warns that “it’s not magic” and stresses that the quality of the data being used is vitally important.
“A machine learning algo lives and dies by the quality of the data it consumes,” he says.
You can watch the full video here: