Computer Vision BTC forecast


The information presented on this site should not be used as a basis for investment decisions.
Bitcoin forecast based on the computer vision technical analysis.




The figure shows the BTC course (olive graph), 12h course average (continuous line in peru color) and the computer vision forecast (red dots) based on the technical course analysis. Green dots are shown when the prediction of the social media (Bitcoin BTC 48h forecast) is combined and the predictions coincide, i.e. the probability should be stronger if the color is green. Even if the two prediction match they have fully independent data sources and the algorithms are complete different.

The computer visual forecast predicts the changes where the average rate changes more than 3.5% (up or down) in future. If the prediction gives a result that the change is positive (more than 3.5%) and it is very probable (probability near 100%) the dots are drawn on the top of the figure. Respectively, if the prediction gives a result that is negative (i.e. the change is more than -3.5% downwards) and it is very likely to happen (probability close to 100%) the dots are drawn to the bottom of the figure. The dots between the minimum and maximum are not relevant; it means the the prediction is not giving results for these values, or the probability is far from 100% that a larger change would occur.

The computer vision algorithm is based on the tensorflow framework and the keras machine learning library from which the predictions are made. It uses the BTC data up to the date of 2020-05-12. A more detailed theory for the algorithm is provided here.

The tuning of the algorithm is ongoing. For example, the algorithm should be refreshed regularly with a new data in order to adopt the new features in data. Also, better results would be obtained, if the results were combined to some other prediction data.