private static void RealStart(AnnVisualizer thisVisualizer, NeuralNetwork ann)
{
thisVisualizer.Show(ann);
int refreshCount = 0;
while (true)
{
if (refreshCount++ == 2)
{
refreshCount = 0;
thisVisualizer.Update();
}
Thread.Sleep(200);
}
}
private static void RealStart(AnnVisualizer thisVisualizer, NeuralNetwork ann)
{
thisVisualizer.Show(ann);
int refreshCount = 0;
while (true)
{
if (refreshCount++ == 2)
{
refreshCount = 0;
thisVisualizer.Update();
}
Thread.Sleep(200);
}
}
static void Main(string[] args)
{
DateTime start = Config.ParseDateTimeLocal(args[0]);
DateTime end = Config.ParseDateTimeLocal(args[1]);
PROBLEM pb = (PROBLEM)Enum.Parse(typeof(PROBLEM), args[2]);
Dictionary<string, string> dicSettings = new Dictionary<string, string>();
dicSettings["APP_NAME"] = "Midax";
dicSettings["LIMIT"] = "10";
dicSettings["DB_CONTACTPOINT"] = "192.168.1.25";
dicSettings["REPLAY_MODE"] = "CSV";
dicSettings["TRADING_MODE"] = "CALIBRATION";
Config.Settings = dicSettings;
// read market data and indicator values
var marketData = new Dictionary<string, List<CqlQuote>>();
var indicatorData = new Dictionary<string, List<CqlQuote>>();
var profitData = new Dictionary<string, List<double>>();
string[] ids = new string[1];
ids[0] = "CS.D.EURUSD.TODAY.IP";
while (start <= end)
{
List<string> mktdataFiles = new List<string>();
mktdataFiles.Add(string.Format("..\\..\\..\\MarketSelector\\MktSelectorData\\mktselectdata_{0}_{1}_{2}.csv", start.Day, start.Month, start.Year));
Config.Settings["REPLAY_CSV"] = Config.TestList(mktdataFiles);
Config.Settings["PUBLISHING_START_TIME"] = string.Format("{0}-{1}-{2} {3}:{4}:{5}", start.Year, start.Month, start.Day, 6, 45, 0);
Config.Settings["PUBLISHING_STOP_TIME"] = string.Format("{0}-{1}-{2} {3}:{4}:{5}", start.Year, start.Month, start.Day, 18, 0, 0);
Config.Settings["TRADING_START_TIME"] = string.Format("{0}-{1}-{2} {3}:{4}:{5}", start.Year, start.Month, start.Day, 8, 0, 0);
Config.Settings["TRADING_STOP_TIME"] = string.Format("{0}-{1}-{2} {3}:{4}:{5}", start.Year, start.Month, start.Day, 17, 0, 0);
Config.Settings["TRADING_CLOSING_TIME"] = string.Format("{0}-{1}-{2} {3}:{4}:{5}", start.Year, start.Month, start.Day, 16, 55, 0);
//Config.Settings["PUBLISHING_CSV"] = string.Format("..\\..\\CalibrationData\\calibdata_{0}_{1}_{2}.csv", start.Day, start.Month, start.Year);
var client = new ReplayStreamingClient();
client.Connect();
Dictionary<string, List<CqlQuote>> curDayMktData = client.GetReplayData(ids);
foreach (var keyVal in curDayMktData)
{
var processableMktData = new List<CqlQuote>();
foreach (var quote in keyVal.Value)
{
if (client.ExpectedIndicatorData["EMA_90_" + ids[0]].Select(cqlq => cqlq.t).Contains(quote.t))
processableMktData.Add(quote);
}
if (marketData.ContainsKey(keyVal.Key))
marketData[keyVal.Key].AddRange(processableMktData);
else
marketData[keyVal.Key] = processableMktData;
}
foreach (var keyVal in client.ExpectedIndicatorData)
{
if (indicatorData.ContainsKey(keyVal.Key))
indicatorData[keyVal.Key].AddRange(keyVal.Value);
else
indicatorData[keyVal.Key] = keyVal.Value;
}
foreach (var keyVal in client.ExpectedProfitData)
{
if (!profitData.ContainsKey(keyVal.Key.Key))
profitData[keyVal.Key.Key] = new List<double>();
profitData[keyVal.Key.Key].Add(keyVal.Value);
}
// process next day
do
{
start = start.AddDays(1);
}
while (start.DayOfWeek == DayOfWeek.Saturday || start.DayOfWeek == DayOfWeek.Sunday);
}
NeuralNetworkForCalibration ann = null;
var maxError = 1e-5;
switch (pb)
{
case PROBLEM.PARITY:
ann = new NeuralNetworkParity("Parity-3");
ann.Train();
break;
case PROBLEM.WMA:
ann = new NeuralNetworkWMA_5_2(ids[0], marketData, indicatorData, profitData);
maxError = 5.0;
break;
case PROBLEM.FX:
ann = new NeuralNetworkFX(ids[0], marketData, indicatorData, profitData);
maxError = 5.0;
break;
}
if (ann == null)
MessageBox.Show("Could not instanciate the Neural Network", "Error");
else
{
using (AnnVisualizer visualizer = new AnnVisualizer(ann))
{
ann.Train(maxError);
CassandraConnection DBconnection = (CassandraConnection)PublisherConnection.Instance.Database;
if (ann.Version > 0)
{
var prevError = DBconnection.GetAnnError(ann.AnnId, ann.StockId, ann.Version - 1);
var prevLearningRate = DBconnection.GetAnnLearningRate(ann.AnnId, ann.StockId, ann.Version - 1);
DialogResult dialogResult = MessageBox.Show(string.Format("The calibration error is {0} (Previously was {1}).\n The learning rate is {2}% (Previously was {3}).\n Would you like to publish the weights to production DB?",
ann.Error, prevError, ann.LearningRatePct, prevLearningRate), ann.GetType().ToString(), MessageBoxButtons.YesNo);
if (dialogResult == DialogResult.Yes)
PublisherConnection.Instance.Insert(DateTime.Now, ann);
}
else{
DialogResult dialogResult = MessageBox.Show(string.Format("The calibration error is {0}.\n The learning rate is {1}%.\n Would you like to publish the weights to production DB?",
ann.Error, ann.LearningRatePct), ann.GetType().ToString(), MessageBoxButtons.YesNo);
if (dialogResult == DialogResult.Yes)
PublisherConnection.Instance.Insert(DateTime.Now, ann);
}
}
}
}
