The Case for Using Deep Learning in Crypto Asset Predictions
New areas of research offer solid promises to tackle the current challenges in crypto-asset price predictions.
In a previous article, we explore some of the challenges of using intelligent models to forecast the price of cryptocurrencies. At a high level, developing a predictive model requires three key main points:
a) A Prediction Thesis: Deciding the factor that will be predicted. Ex: price, risk, momentum, etc.
b) A Data Source: Predictions will based on order book datasets, blockchain records or derivative order books.
c) A Prediction Technique: Deciding which methodology to use. Ex: Time series forecasting, machine learning and deep learning.
From the prediction techniques in the market, there are three fundamental schools analytic schools that should be considered:
· Time-Series Forecasting
· Machine Learning
· Deep Learning