Let’s zoom in on one of the types of trading strategies we talked about earlier, cross-sectional equity investing. This is a deep and important type of trading strategy. So, let’s talk in detail about the pieces needed to build this type of strategy. In general, you can think of the process of developing this type of strategy as having six stages: data selection, universe definition, alpha discovery, alpha combination, portfolio construction, and trading. In the first stage, you need to decide what data set or data sets you want to use. As we’ve said, you should start with a hypothesis. So, you’re going to want to get access to the data relevant to testing it. In the second stage, you need to pare down your data set to a subset that contains the stocks you wanted potentially trade. You might exclude stocks that have low training volume, and are therefore, hard to trade. But you also want to construct your portfolio, so that the stocks in it have similarities, so that ranking them is a reasonable thing to do. However, since you’re looking to benefit from their movements relative to each other, they shouldn’t be too similar. You’re also going to want to limit your universe to the stocks to which your hypothesis logically applies. For example, let’s imagine you want to work with the hypothesis discussed in the paper, Geographic Momentum by Quoc Nguyen. This hypothesis is that a momentum effect is created for multinational US companies, because investors do not pay attention to foreign market developments. So, there is an opportunity to predict increases in these company stock prices when markets and countries they’re operating in improve. If you’re working with this hypothesis, you would only want to work with companies who actually have significant foreign market operations.