High frequency, low latency – where Python fails
Our overview would be incomplete without mentioning HFT. Its roots are in the financial crisis of 2008 when liquidity became the main issue in most, if not all, developed markets. Exchanges started to offer an incentive to those who provided liquidity, waiving many restrictions that previously required liquidity providers to be regulated. As a result, many market participants started to offer liquidity, or, rather, demonstrated this liquidity in the order book – because they sent an order only to withdraw it from the book some milliseconds later. In other words, they started to bluff creating an illusion of liquidity.
Of course, to be successful here, you need to be able to process thousands of transactions per second and reduce the latency (that is, the time between the order is sent to the exchange and the time it appears in the order book) to the absolute minimum. That’s why HFT requires very expensive computers...
