Topic: Neural network based indicator (Read 34 times)

We have created a self-learning neural network for determining Bitcoin price movement.
 The system analyzes repeating patterns simultaneously on timeframes 15, 30, 45 minutes, 1,2,3,4 hours and 1 day. With a number of exact matches, with a coefficient above 0.67, the data is analyzed for the accuracy of the pattern with 8 main indicators of technical analysis. The neural network does not rely on technical analysis - the priority direction is recognition of the continuation of the trend of a clone pattern from the past, with the likely direction of price movement.
 With sharp jumps in volatility, a huge number of false signals are received, the system in this case ignores them, showing only a classified and stable price direction, backed up by historical data.
 At all stages of testing, our neural network showed the so-called trader's feeling - that is, clear and confident signals at the beginning of a new trend. Likewise, experienced traders intuitively see if there is a suitable entry point.
  We start testing the network today, on the channel - https://t.me/neuron_indicator
 The current accuracy is 67%, the number of signals per week: 10 - 30. The first goal is to increase the accuracy up to 75% with the same number of signals. The second goal is to visualize a possible chart with a correction over time (the model is redrawn more accurately every 15 minute timeframe).
 We start open testing for several weeks, upon reaching the first goal, we introduce paid access for a month at the lowest price. When the second goal is achieved, we create our own token with an exchanger on the site and integrate it into the infrastructure of all branches of the neural network.
 Wish us luck!
 Team "Neuronic".