Neutral networks experts (Free code) – Trading Systems – 16 August 2023

registerEA(
"nn_example",
"A test EA to run neuron model for XOR(v1.0)",
[

{name: "h11", value: 20, required: true, type: "Number", range: [-100, 100], step: 10},
{name: "h12", value: 20, required: true, type: "Number", range: [-100, 100], step: 10},
{name: "b1", value: -10, required: true, type: "Number", range: [-100, 100], step: 10},

{name: "h21", value: -20, required: true, type: "Number", range: [-100, 100], step: 10},
{name: "h22", value: -20, required: true, type: "Number", range: [-100, 100], step: 10},
{name: "b2", value: 30, required: true, type: "Number", range: [-100, 100], step: 10},

{name: "o1", value: 20, required: true, type: "Number", range: [-100, 100], step: 10},
{name: "o2", value: 20, required: true, type: "Number", range: [-100, 100], step: 10},
{name: "b", value: -30, required: true, type: "Number", range: [-100, 100], step: 10},
],
function (context) { 
    
        var account = getAccount(context, 0)
        var brokerName = getBrokerNameOfAccount(account)
        var accountId = getAccountIdOfAccount(account)
        var symbolName = "EUR/USD"

        getQuotes (context, brokerName, accountId, symbolName)

},
function (context) { 
        var h11 = getEAParameter(context, "h11") 
        var h12 = getEAParameter(context, "h12") 
        var b1 = getEAParameter(context, "b1") 
        var h21 = getEAParameter(context, "h21")
        var h22 = getEAParameter(context, "h22")
        var b2 = getEAParameter(context, "b2")
        var o1 = getEAParameter(context, "o1") 
        var o2 = getEAParameter(context, "o2") 
        var b = getEAParameter(context, "b") 

        var sigmoid = function 
                return 1 / (1 + Math.pow(Math.E, -t))
        }

        var nn = function (p1, p2) {
                return sigmoid(o1 * sigmoid(h11 * p1 + h12 * p2 + b1) + o2 * sigmoid(h21 * p1 + h22 * p2 + b2) + b)
        }

        var error = 0

        error += nn(1, 1)
        error += 1 - nn(1, 0)
        error += 1 - nn(0, 1)
        error += nn(0, 0)

        setOptimizationResult(context, -error)

        printMessage("error: " + error + ", " + Math.round(nn(1, 1)) + " " + Math.round(nn(1, 0)) + " " + Math.round(nn(0, 1)) + " " + Math.round(nn(0, 0)))
},
function (context) { 
})

2.

registerEA(
                "sample_run_neuron_model",
                "A test EA to run neuron model(v1.04)",
                [{ 
                        name: "period",
                        value: 20,
                        required: true,
                        type: PARAMETER_TYPE.INTEGER,
                        range: [1, 100]
                },{
                        name: "inputNum",
                        value: 20,
                        required: true,
                        type: PARAMETER_TYPE.INTEGER,
                        range: [1, 100]
                },{
                        name: "threshold",
                        value: 0.3,
                        required: true,
                        type: PARAMETER_TYPE.NUMBER,
                        range: [0, 1]
                },{
                        name: "takeProfit",
                        value: 0.0001,
                        required: true,
                        type: PARAMETER_TYPE.NUMBER,
                        range: [0, 100]
                }],
                function (context) { 
                        
                        
                        if (typeof localStorage.reservedZone == "undefined") return

                        var reservedZone = JSON.parse(localStorage.reservedZone)
                        if (typeof reservedZone.sample_training_neuron_model == "undefined") return
                        context.myPerceptron = synaptic.Network.fromJSON(reservedZone.sample_training_neuron_model)

                        var account = getAccount(context, 0)
                        var brokerName = getBrokerNameOfAccount(account)
                        var accountId = getAccountIdOfAccount(account)
                        var symbolName = "EUR/USD"

                        getQuotes (context, brokerName, accountId, symbolName)
                        context.chartHandle = getChartHandle(context, brokerName, accountId, symbolName, TIME_FRAME.M1)
                        var period = getEAParameter(context, "period")
                        context.indiHandle = getIndicatorHandle(context, brokerName, accountId, symbolName, TIME_FRAME.M1, "rsi", [{
                                name: "period",
                                value: period
                        }])
                },
                function (context) { 
                },
                function (context) { 
                        var arrTime = getData(context, context.chartHandle, DATA_NAME.TIME)
                        if (typeof context.currTime == "undefined") {
                                context.currTime = arrTime[arrTime.length - 1]
                        } else if (context.currTime != arrTime[arrTime.length - 1]) {
                                context.currTime = arrTime[arrTime.length - 1]
                        } else {
                                return
                        }

                        var account = getAccount(context, 0)
                        var brokerName = getBrokerNameOfAccount(account)
                        var accountId = getAccountIdOfAccount(account)
                        var symbolName = "EUR/USD"

                        var period = getEAParameter(context, "period")
                        var inputNum = getEAParameter(context, "inputNum")
                        var threshold = getEAParameter(context, "threshold")
                        var takeProfit = getEAParameter(context, "takeProfit")
                        var arrRsi = getData(context, context.indiHandle, "rsi")

                        if (inputNum + period - 1 > arrRsi.length) throw new Error("No enough data.")

                        var input = []

                        for (var i = arrRsi.length - inputNum - 1; i < arrRsi.length - 1; i++) {
                                input.push(arrRsi[i] / 100)
                        }

                        var result = context.myPerceptron.activate(input)[0]
                        printMessage(result)

                        var ask = null
                        var bid = null
                        var volume = 0.01

                        try {
                                ask = getAsk(context, brokerName, accountId, symbolName)
                                bid = getBid(context, brokerName, accountId, symbolName)
                        } catch (e) {
                                
                                printErrorMessage(e.message)
                                return
                        }

                        if (result < 0.5 - threshold) {
                                sendOrder(brokerName, accountId, symbolName, ORDER_TYPE.OP_BUY, 0, 0, volume, ask+takeProfit, bid-3*takeProfit, "", 0, 0)
                        } else if (result > 0.5 + threshold) {
                                sendOrder(brokerName, accountId, symbolName, ORDER_TYPE.OP_SELL, 0, 0, volume, bid-takeProfit, ask+3*takeProfit, "", 0, 0)
                        }
                }
        )

3.

registerEA(
                "sample_training_neuron_model",
                "A test EA to train neuron model(v1.03)",
                [{ 
                        name: "period",
                        value: 20,
                        required: true,
                        type: PARAMETER_TYPE.INTEGER,
                        range: [1, 100]
                },{
                        name: "inputNum",
                        value: 20,
                        required: true,
                        type: PARAMETER_TYPE.INTEGER,
                        range: [1, 100]
                },{
                        name: "hiddenNum",
                        value: 50,
                        required: true,
                        type: PARAMETER_TYPE.INTEGER,
                        range: [1, 100]
                },{
                        name: "diffPrice",
                        value: 0.0001,
                        required: true,
                        type: PARAMETER_TYPE.NUMBER,
                        range: [0, 10]
                }],
                function (context) { 
                        var account = getAccount(context, 0)
                        var brokerName = getBrokerNameOfAccount(account)
                        var accountId = getAccountIdOfAccount(account)
                        var symbolName = "EUR/USD"

                        context.chartHandle = getChartHandle(context, brokerName, accountId, symbolName, TIME_FRAME.M1)
                        var period = getEAParameter(context, "period")
                        context.indiHandle = getIndicatorHandle(context, brokerName, accountId, symbolName, TIME_FRAME.M1, "rsi", [{
                                name: "period",
                                value: period
                        }])
                },
                function (context) { 
                        var period = getEAParameter(context, "period")
                        var inputNum = getEAParameter(context, "inputNum")
                        var hiddenNum = getEAParameter(context, "hiddenNum")
                        var arrOpen = getData(context, context.chartHandle, DATA_NAME.OPEN)
                        var arrClose = getData(context, context.chartHandle, DATA_NAME.CLOSE)
                        var arrRsi = getData(context, context.indiHandle, "rsi")

                        if (arrRsi.length <= period + 1) return
                        if (inputNum + period - 1 > arrRsi.length) throw new Error("No enough data.")

                        
                        Perceptron.prototype = new synaptic.Network()
                        Perceptron.prototype.constructor = Perceptron

                        var myPerceptron = new Perceptron(inputNum, hiddenNum, 1)
                        var myTrainer = new synaptic.Trainer(myPerceptron)

                        var diffPrice = getEAParameter(context, "diffPrice")
                        var trainingSet = []
                        var longCount = 0
                        var shortCount = 0

                        for (var i = period - 1; i < arrRsi.length - inputNum; i++) {
                                if (arrClose[i * inputNum + inputNum] - arrOpen[i * inputNum + inputNum] > diffPrice) {
                                        var input = []

                                        for (var j = 0; j < inputNum; j++) {
                                                input.push(arrRsi[i * inputNum + j] / 100)
                                        }

                                        trainingSet.push({
                                                input: input,
                                                output: [0]
                                        })

                                        longCount++
                                } else if (arrOpen[i * inputNum + inputNum] - arrClose[i * inputNum + inputNum] > diffPrice) {
                                        var input = []

                                        for (var j = 0; j < inputNum; j++) {
                                                input.push(arrRsi[i * inputNum + j] / 100)
                                        }

                                        trainingSet.push({
                                                input: input,
                                                output: [1]
                                        })

                                        shortCount++
                                }
                        }

                        myTrainer.train(trainingSet)

                        
                        
                        if (typeof localStorage.reservedZone == "undefined") {
                                localStorage.reservedZone = JSON.stringify({sample_training_neuron_model: myPerceptron.toJSON()})
                        } else {
                                var reservedZone = JSON.parse(localStorage.reservedZone)
                                reservedZone.sample_training_neuron_model = myPerceptron.toJSON()
                                localStorage.reservedZone = JSON.stringify(reservedZone)
                        }

                        printMessage(longCount + ", " + shortCount)
                        printMessage(JSON.stringify(trainingSet))
                        printMessage(JSON.stringify(myPerceptron.toJSON()))
                },
                function (context) { 
                }
        )

ReadMe….

“sample_training_neuron_model” and “sample_run_neuron_model” are two neural network-based EAs that must be used together. One cannot function without the other.

“sample_training_neuron_model” trains the model, while “sample_run_neuron_model” generates signals based on the model trained by “sample_training_neuron_model”.

To help understand how a neural network works, i have included a simple example called “nn_example”.

The algorithm implemented here is based on Synaptic, an AI framework. Tensorflow, a more advanced framework, runs convolutional neural networks. For more details on Tensorflow, refer to plugin_for_tensorflow.

This is better than fire…

Best regards.

Forex Mastery, a network of Companies specializing in the development of Strict and complex Software used in trading development and automation.