Funded by the National Association of Securities, dealers the Nasdaq became the world’s first electronic stock market when it began trading on February eighth 1971. Since then, people have used computers to sell and buy stocks at speeds and frequency that are unmatched by any human trader. The use of computers with pre-programmed algorithms to perform traits in the stock market has become known as algorithmic trading. In the past decade, high-frequency trading or HFT for short, has become one of the most popular methods of algorithmic trading. High frequency trading consists of using powerful computers, with dedicated connections to the stock exchanges to analyze stock data and execute a large number of transactions across multiple markets at extremely high speeds. Despite its assess, high-frequency trading has been losing popularity due to many factors including; the decline in profit margins, attributed to a decrease in volatility and trading volume, high maintenance costs and a large amount of competition among trading firms that use HFT algorithms to try to detect and outbid each other. Trading algorithms such as HFT make decisions are nanosecond time scales, taking human decisions and interaction out of the loop. By contrast, people often look at the news, use high-level analysis, and sometimes intuition before they make any decisions about a particular transaction. Ideally, one would like to create an algorithm that falls somewhere in the middle of these two extremes. For example, it’ll be ideal to create an algorithm that can analyze stock data faster than any person possibly could, but that can make smart decisions like a human trader would. Allowing you to perform better than any other trading algorithms such as HFT. Due to its assess in many fields, it’s not surprising that Machine Learning techniques such as reinforcement learning, has gained popularity in creating such fully automated and smart trading algorithms. But why use reinforcement learning? Why not use more conventional Machine Learning techniques such as supervised learning? In the next lesson, we’ll see some of the challenges faced by supervised learning in creating such automated trading systems.