A trading system will help you to automate your trading strategy. When you choose to build this kind of software, you need to take the following into consideration:

• Asset class: When you code, knowing which asset class will be used in your trading system will modify the data structure of this software. Each asset class is idiosyncratic and has its own set of features. US stocks are mainly traded on two exchanges (NY Stock Exchange and NASDAQ). There are about 6,000 companies (symbols) listed on these two exchanges. Unlike equities, Foreign Exchange (FX) has six major currency pairs, six minor currency pairs, and six more exotic currency pairs. We can add more currency pairs, but there will not be more than 100 currency pairs. However, there will be hundreds of market players (banks, brokers).
• Trading strategy type (high frequency, long-term position): Depending upon the type of strategies, the design of the software architecture will be impacted. High-frequency trading strategies require sending orders very rapidly. A regular trading system for US equities will decide to send an order within microseconds. A system trading on the Chicago Mercantile Exchange (CME) could work within nanoseconds. Based on this observation, the technology will be critical in the choice of designing the software. If we just refer to the programming language, Python is not adapted to speed and we will preferably choose C++ or Java. If we want to take a long-term position such as many days, the speed allowing a trader to get a liquidity faster than others will not be important. A programming language such as Python will be fast enough to reach this goal.
• The number of users (the number of trading strategies): When the number of traders increases, the number of trading strategies increases. This means that the number of orders is higher. Before sending an order to an exchange, we need to check the validity of the orders we are about to send: checking whether the overall position for a given instrument has not been reached. In trading world, we have more and more regulations moderating trading strategies. To follow that our trading strategy respect the regulation, we will test the compliance of the orders that we want to send. All these checks will add some calculation time. If we have too many orders, we will need to have all these verification done sequentially for one given instrument. If the software is not fast enough, it will slow down the orders to go out. So having more users will require a faster trading system.

These parameters modify the conception of the trading system you are going to build. It is essential to have a clear description of the requirements when you build a trading system.

Because the goal of a trading system is to support your trading ideas. The trading system will collect the information that your trading strategy needs and be in charge of sending orders and receiving responses from the market regarding this order. The main functionalities will be to collect the data (most of the time this will be price updates). If the trading strategy needs to get some quantitative data involving earnings, fed announcements (more generally news), these news will also trigger orders.  When the trading strategy decide the direction of the position. the trading system will send orders accordingly . The trading system will also decide which specific exchange will be the best to get the order filled for the requested price and for the requested volume.