As Bitcoin surges through the roof to reach heights that people didn’t imagine, people have begun to wonder how we can predict the rise and fall of bitcoin and other cryptocurrencies. For every crypto coin, there’s no physical attribute that can raise the value of the coin so it’s quite hard to track the progress of these crypto coins.

Different researchers have met success with tracking different bigger crypto coins but these methods have different variables that determine the progress of those models. This makes it difficult to decide how to replicate such a process for smaller models. At this point, it is hard to know and understand what drives bitcoins on the charts. However, a group of researchers believe that they could use an AI network known as Artificial Neural Network to predict the rise and fall of bitcoin prices.

How Artificial Neural Network works?

Predicting Fluctuations

To get this model to predict fluctuations, they used a predictor known as Moving Averages. This predictor functions by collecting the average of prices over time and plotting a graph with the prices along a line. The idea behind this move is for when the price of these cryptocurrency rises above or falls below the average price in the last fifty to two hundred days so traders know the type of trend to expect.

The test of this model provides some interesting results about bitcoin’s predictability through a period and when the trend table experiences unusual volatility. The success of this model has led to the further development of other predictors of this model.

New Model predictors

After observing the model for 7 years, three new predictors were created for this AI network known as returns, 50-day buy-sell signal and 200-day buy-sell signal. These new predictors were tested on ANN that includes the Chicago Board Options Exchange Volatility Index to see if the stock market volatility affected the trend table.

The result of this action is that the ANN model takes note of the predictors, input and output and tries to find a pattern between all the data. It keeps on checking these patterns until it gets to a point where the testing becomes redundant.

Results of prediction

The ANN model when tested on predicting the future of the value of coins succeeded in reducing the prediction error by 5 to 10 per cent throughout the observation period. Our observation showed that there was an improvement in the forecast indicating that in the future, predicting the rise and fall of bitcoin prices daily will no longer be by guesswork. You will be able to use prediction models to predict the future of crypto coins like it states in this Bitcoin formula trading review

From the results, we can see that bitcoin and other crypto coins are not affected by movement on the stock market which suggest that investors in bitcoins and traditional market investors are different entities.

The researchers also tried to zoom the market further for inefficiencies by separating the data into four subsamples and the AI model worked better when the data was shared into subsamples. The result of such an experiment aside from price accuracy includes the accuracy of the system’s prediction on the fall or rise of prices.


With this ANN model, trading bitcoin would become more profitable on average than buying coins randomly and expecting to make a profit. With this experiment, we believe that the future of bitcoin trading will be forecasting the coins and investing in the abilities of AI and AI network like Artificial Neural Network as stated here.