static void TrainWithIndexData()
{
List <double[]> data = new List <double[]>();
var symbol = YSQSymbol.YSQIndex.SNP;
var begin = new DateTime(2004, 1, 1);
var end = new DateTime(2008, 1, 1);
var samplePackage = new YSQProvider().GetHistory(new Asset(symbol, AssetType.Index), begin, end, null);
var barPackage = (IBarPackage)samplePackage;
var bars = barPackage.Samples;
TaResult indicator0 = bars.SMA(7);
TaResult indicator1 = bars.SMA(14);
TaResult indicator2 = bars.SMA(24);
TaResult indicator3 = bars.SMA(60);
List <TaResult> indicatorsList = new List <TaResult> {
indicator0, indicator1, indicator2, indicator3
};
// discard bars and indicators previous to valid series
int firstValidSampleOnSeries = TaResult.GetFirstValidSample(indicatorsList);
Console.WriteLine("First valid sample on series: {0}", firstValidSampleOnSeries);
bars.RemoveRange(0, firstValidSampleOnSeries);
TaResult.DiscardFirstSamples(indicatorsList, firstValidSampleOnSeries);
Console.WriteLine("Invalid samples removed");
// possible answers
var stay = new double[] { 1, 0, 0 };
var goLong = new double[] { 0, 1, 0 };
var goShort = new double[] { 0, 0, 1 };
var deltaPrices = bars.Select(x => x.Close - x.Open).ToList();
var action = deltaPrices.Select(deltaPrice => {
var abs = Math.Abs(deltaPrice);
if (abs < 5)
{
return(stay);
}
if (deltaPrice < 0)
{
return(goShort);
}
return(goLong);
}).ToList();
Console.WriteLine("Times to stay: {0}", action.Count(x => x == stay));
Console.WriteLine("Times to go long: {0}", action.Count(x => x == goLong));
Console.WriteLine("Times to go short: {0}", action.Count(x => x == goShort));
int samplesCount = bars.Count;
for (int i = 0; i < samplesCount; i++)
{
var entry = new double[] {
indicator0.InstantValues[i].Value,
indicator1.InstantValues[i].Value,
indicator2.InstantValues[i].Value,
indicator3.InstantValues[i].Value,
action[i][0],
action[i][1],
action[i][2]
};
data.Add(entry);
}
NeuralNetwork neuralNetwork = Trainer.Execute(data);
ChartPool.CreateChart();
int takeProfitPoints = 15;
int stopLossPoints = 10;
decimal intensityThreshold = 0.0m;
indicator0.Normalize();
indicator1.Normalize();
indicator2.Normalize();
indicator3.Normalize();
var simulation = new NNSimulation(neuralNetwork, intensityThreshold, takeProfitPoints, stopLossPoints, true);
SimulationRunner simulationRunner = new SimulationRunner(bars, simulation);
simulationRunner.AddSerie("INDICATOR_0", indicator0);
simulationRunner.AddSerie("INDICATOR_1", indicator1);
simulationRunner.AddSerie("INDICATOR_2", indicator2);
simulationRunner.AddSerie("INDICATOR_3", indicator3);
simulationRunner.Execute();
ShowSimulation(simulation, bars, indicator0, indicator1, indicator2, indicator3);
}
