Long-term Price Predictions
Machine learning can help us predict the long-term price oscillations that we see every day on Enroyd. Our forecasts are based on the Prophet module, an open source software released by Facebook’s Core Data Science team. Are our predictions perfect? They’re not: they’re basically an estimation of the future trend based on past prices. The confidence intervals (the “grey lines” in the chart below) indicate the possible future price ranges predicted by the model.
Why do we only predict the next three months? As Crypto is a very volatile asset, any prediction beyond three months doesn’t make a lot of sense. Event the confidence intervals of our short-term predictions can be relatively large sometimes.
Therefore you shouldn’t base your investment decisions only on the model’s predictions and we cannot take any responsibility if the predictions are wrong. Especially “pumps ‘n dumps” cannot be predicted by the models because normally they’re not based on any fundamentals that would justify such a sudden price increase or decrease.
Daily Price Predictions
Machine learning can help us predict the price oscillations that we see every day in the market. Based on past prices and on more than 35 technical and other indicators we aggregate several machine learning models to predict what the future coin price could be 24 hours from now. Are these predictions perfect? They’re not: they’re only an estimation based on past prices and they cannot account for future events (e.g. important announcements, political decisions or “Black Swans”).
In the chart below you can see the percentage of models that predict whether the future price will be below or above the respective threshold price 24 hours from now. When volatility is high, the range of the price predictions will tend to broaden. Nevertheless, we remind you that you shouldn’t base your investment decisions only on the model predictions and we cannot take any responsibility if the predictions are wrong. Especially “pumps ‘n dumps” cannot be predicted by the models because normally they’re not based on any fundamentals that would justify such a sudden price increase or decrease. Our predictions are just supposed to guide you and to give you a hint as to where the “journey” might go… And also make sure to check out our wiki if you want to learn more about machine learning.
Historical Performance Daily Price Predictions
As we try to be as transparent as possible about our models, we decided to display the past performance of our price prediction models in the chart below. The two grey lines show the past daily 5% and 95% prediction intervals, whereas the orange line shows the real average prices. Why would this be useful? Well, this chart helps you assess how well the models have performed lately, if they’re still adjusting to the new market situation after a trend reversal, how good they deal with extreme values, etc.