Hi,
I dont really know much about technical analysis but I have been running a neural network for some while predicting btc prices. All results are provable documented on the steem blockchain in these posts : btc neural network. So far the results are very mixed but the entire time I am doing this was a bear market. It will be interesting to see what happens in the next bull run. The network has preformed slightly better than hodling btc.
I have tested the network on data that was not used in training and it does work there. But we are not talking about %-level advances to predictions, but in the permille level. The random noise is huge and the network is finding tiny trends below that noise.
The network predictions are always very smooth as all the noise is filtered out. That also means that the prediction is not realistic and should be overlayed with a random walk. When the prediction is up, it should be understood that there will be big fluctuations, but fluctuations up are slightly more likely than fluctuations down.
I have trained my network on the eth prices for your task and here are the results:
.png)
Red areas are predicted below the current price, blue/yellow areas indicate the expected passive growth, green areas are better than the expected growth. The vertical line is one week into the future and its size provides the prediction variance. The last point is one month in the future. The network prediction is smooth as the random noise is filtered out.
The network sees small losses in the next few days because it does not trust the rapid rise of yesterday. After that it predicts a boring week, while the general outlook for the next month is positive.
Please do not consider this post as financial advice! I am simply exploring the possibility to use neural networks to find hidden patterns in crypto prices.
ill have to think about this. mhhhh