This utility is designed to increase the efficiency of manual trading and practicing methods of automatic trading using neural networks, without the need to resort to third-party programs, using import/export of working data. The forward period allows not only testing the strategy on time segments unavailable for optimization, but also stopping the automatic trading under unfavorable conditions.
The genetic algorithm searches for the best parameters of stop loss, take profit, trailing stop for 20 different strategies, or separately for each.
It includes a trade panel for quick work in manual mode without being distracted from working with the program. New method of placing pending orders allows placing orders in two clicks, focusing only on the price chart, disregarding the long figures of the market. The ability to save templates will allow you to quickly assess the market situation without leaving the program.
Visual assessment of indicators from the market has never been so fast. Method for quick visual assessment of indicators allows changing the indicator parameters using a slider, without the need to call the indicator properties window.
Training and testing of the neural network is divided into three stages: training, test, trade.
Training - Network training.
Test - testing the network after each training epoch (if the error is less than the previous, the network is stored).
Trade - this shows the responses of the network to data that the network has not seen during training and testing.
The network can be fed any indicators normalized to values from -1 to 1. The indicator values must not exceed the extreme values during training.
You can create a set of up to 100 indicators.
As a training set, you can select any indicator, built-in Zig Zag or an output of another network.
Before training, it is necessary to determine the type of network, each of which has its own special qualities.
MLP - a back propagation network, it has good predictive qualities.
GRNN - network with absolute memory.
SOM - separates into the specified number of classes.
- Trading panel
It is necessary to allow automated trading for the panel to work. To set the price of a pending order, simple click the price chart and press the button of the required order.
Before starting the optimization, it is necessary to select the strategies to optimize.
Stop Loss, Take Profit, Trailing Stop and trade signal levels are available in optimization.
- Visual assessment of indicators
The indicator parameters can be changed using the slider. The indicator will be instantly recalculated.
Kohonen neural network,