EA Tutorial ~ How beginners learn neural network?

EA Tutorial ~ How beginners learn neural network?

# Expert Advisor Tutorial ~ How beginners learn neural network?

This expert advisor tutorial shares with you the experience about how beginners learn neural network via our expert advisor studio.

# A lot of traders who are not familiar with artificial intelligence asked me whether it's possible for a beginner to learn neural networks.

The answer is simple and positive. YES! you can learn neural networks without any prerequisite knowledge.

Investigating neural network further takes time and we need to explain the theory on the perspective of mathematics. It's difficult for a short post to discuss. A book with hundreds of pages may explain more clearly. Fortunately, we don't need to understand it like a scientist.

We just need to know why we need neural networks and how it helps us instead of knowing what it is.

Neural networks can help us find out the patterns of the market movements by training the model. It can tell us what the traditional indicators didn't tell. Some patterns are not explicit to us by using traditional indicators. So we need neural networks to supplement the missing information.

# The first thing we need to do is to gather data.

The patterns underlying the historical data can be found anywhere. An output of an indicator, a Fibonacci retracement, a piece of news... If the data can be gathered quantitatively, they can be used for a neural network model.

# The second step is to train the model by the gathered data.

Before training, you need to normalize them to be ranged from 0 to 1.

The third step is to verify the model trained by different data sets. The training set and the verification set should be different. For example, we trained a model by the historical data from Jam 2018 to Dec 2018, then we need to verify the model by the data from Jan 2019 to Dec 2019.

# We recommend you start learning neural networks from the simplest example, such as perceptrons to calculate XOR.

Perceptrons to calculate XOR have just 4 combinations of the input data: (1,1) -> 0, (0,0) -> 0, (1,0) -> 1, (0,1) -> 1. So the structure inside is quite simple as well. You can debug the program step by step to observe how it works.

I have written a post to introduce the new feature of Fintechee - Genetic algorithm for trading. Please refer to it to know more details about neural networks. In that post, we use a perceptron neural network model to calculate XOR.

We will use neural networks, genetic algorithms on our trading. Please feel free to track our trading records.

Read More