Let's start with the facts. The idea of trading robots is not new, there are many publications about them, they are used by financial institutions and banks, but do they really help? There is a lot of research and thought on the topic of algorithmic trading, but they all come to a blurry conclusion.
Most often, the authors of these bots are either traders with some thoughts and reflections, or programmers with no trading experience. I will try to dissect this topic as much as I can. As a developer and as a trader. By the way, there are companies that can help you build a trading bot. As a person who lost on trading and raised large sums.
If you are a programmer or familiar with the basics of scripting, you can write your own bot in a few hours. But will it make enough money for you? Will there be a stable income? Will he ever reach an annual income of $100,000? In this post, I will answer these questions and give you some tips on how to move forward.
What is a trading bot?
A trading bot is an algorithm that transforms market conditions into trade decisions (usually buy, sell or hold). Nothing special here.
What types of trading bots are there?
All types that any trader could take, since a bot is an algorithm written by programmers, they can put any logic into it.
By type of traders and types of trading, the following options can be distinguished:
• Long-term traders (long-time traders) - investors;
• Swing traders (swing traders) - those who place trades for a week, month or year;
• Day traders (day traders) - those who place a small number of transactions during the day, without moving them overnight;
• Scalpers (scalpers) - those who put a lot of transactions per day, hour or even minute.
All these types of traders can be implemented as a bot.
Where can a trading bot place orders?
Wherever you want and can access the broker's API or through reverse engineering (some brokers may consider it a scam / hack, so it's better to clarify this point in their policy).
But no one forbids you to simulate your own brokerage conditions, commissions, spread, price, and receive current stock or cryptocurrency prices without having funds directly from the broker. Thus, it is possible to test the bot without resorting to real trading, while depending on the quality of the implementation of the simulator and the type of trading chosen, it will depend on how the strategy fits the real market.
What are the types of bot implementations?
Simply put, types of algorithmic trading. Since the bot is a complete program that implements the trader's trading behavior in automatic mode.
• Neural network or artificial intelligence - can be simple single-perceptron bots based on multiple neurons, complex LSTM networks, or even news analysis based on keyword heuristics, semantic search with NLP.
• Quantitative trading - based on a strategy that combines any criteria for making a decision, this can be a comparison of indicators, price behavior, search for patterns, etc.
• Semi-automated alert bot - will use some algorithms to tell traders what to do. This type includes indicators based on any strategy.
• Genetic Algorithms - This could be classified as part of machine learning/neural networks, but is not really well understood to define this type of algorithm as an approach to machine learning. Implementations of genetic algorithms vary, are researched by universities around the world, and are part of a larger topic beyond the scope of this post.
So, I answered four main questions that can already reveal a little theory before you start writing your bot. But what's next? Now that we understand what a bot is, we can think about measuring its KPI.
Do trading bots help you make money?
Yes, sure. This is why many hedge funds, banks and large financial companies hire specialists in machine learning and algorithms. These people are responsible for implementing automated trading bots to trade big markets with big money.