static void Main2(string[] args)
{
System.Console.WriteLine("Loading model ");
MLContext mlContext = new MLContext();
ITransformer mlModel = mlContext.Model.Load("MLModel.zip", out var modelInputSchema);
var predEngine = mlContext.Model.CreatePredictionEngine <ModelInput, ModelOutput>(mlModel);
System.Console.Write("[OK]\n");
System.Console.Write("Loading candles ");
var candlesbefore = Oanda.LoadCandles("EUR_USD", "S5", DateTime.Now.AddDays(-2));
var candles = Oanda.LoadLiveCandles("EUR_USD", "S5");
System.Console.Write($"[OK {candles.Length}]\n");
System.Console.Write("Analyse data ");
var collection = new CandleCollection("EUR_USD", Period.S, 5);
var minuteconsolidation = new ConsolidatedCandleCollection(collection, Period.M, 1);
var m15Consolidation = new ConsolidatedCandleCollection(collection, Period.M, 15);
var stochasticS5 = new StochasticStats(collection);
var stochasticM1 = new StochasticStats(minuteconsolidation);
var stochasticM15 = new StochasticStats(m15Consolidation);
var winpip = new PipExpectation(collection, 30);
foreach (OandaCandle candle in candlesbefore)
{
DateTime datetime = new DateTime(candle.Time.Year, candle.Time.Month, candle.Time.Day, candle.Time.Hour, candle.Time.Minute, candle.Time.Second);
collection.AddCandle(datetime, candle.Mid.O, candle.Mid.C, candle.Mid.H, candle.Mid.L, candle.Volume);
}
foreach (OandaCandle candle in candles)
{
DateTime datetime = new DateTime(candle.Time.Year, candle.Time.Month, candle.Time.Day, candle.Time.Hour, candle.Time.Minute, candle.Time.Second);
collection.AddCandle(datetime, candle.Mid.O, candle.Mid.C, candle.Mid.H, candle.Mid.L, candle.Volume);
}
System.Console.Write("[OK]\n");
System.Console.Write("Prediciction ");
foreach (DateTime dateTime in collection._list.Keys.OrderBy(c => c.Ticks))
{
var S5Candle = collection.GetCandle(dateTime);
if (S5Candle == null)
{
continue;
}
var M1Candle = minuteconsolidation.GetCandle(dateTime);
if (M1Candle == null)
{
continue;
}
var M15Candle = m15Consolidation.GetCandle(dateTime);
if (M15Candle == null)
{
continue;
}
var S5Stoch = stochasticS5.GetValue(dateTime);
if (S5Stoch == null)
{
continue;
}
var M1Stoch = stochasticM1.GetValue(dateTime);
if (M1Stoch == null)
{
continue;
}
var M15Stoch = stochasticM15.GetValue(dateTime);
if (M15Stoch == null)
{
continue;
}
ModelInput input = new ModelInput()
{
Time = (float)S5Candle.DateTime.TimeOfDay.TotalSeconds,
Decision = 0,
S5V = S5Candle.Volume,
S5C = decimal.ToSingle(S5Candle.CloseRelativ),
S5H = decimal.ToSingle(S5Candle.HighRaltiv),
S5L = decimal.ToSingle(S5Candle.LowRelativ),
S5StochK = (float)(S5Stoch.K),
S5StochD = (float)(S5Stoch.D),
M1V = M1Candle.Volume,
M1C = decimal.ToSingle(M1Candle.CloseRelativ),
M1H = decimal.ToSingle(M1Candle.HighRaltiv),
M1L = decimal.ToSingle(M1Candle.LowRelativ),
M1StochK = (float)M1Stoch.K,
M1StochD = (float)M1Stoch.D,
M15V = M15Candle.Volume,
M15C = decimal.ToSingle(M15Candle.CloseRelativ),
M15H = decimal.ToSingle(M15Candle.HighRaltiv),
M15L = decimal.ToSingle(M15Candle.LowRelativ),
M15StochK = (float)M15Stoch.K,
M15StochD = (float)M15Stoch.D
};
ModelOutput result = predEngine.Predict(input);
var expected = winpip.GetValue(dateTime);
if (expected == null)
{
continue;
}
float decision = 0f;
if (expected.Long > 5m && expected.Short < 1.6m)
{
decision = 1;
}
else if (expected.Short > 5m && expected.Long < 1.6m)
{
decision = 2;
}
else
{
decision = 0;
}
if (result.Prediction > 0 || decision > 0)
{
System.Console.WriteLine($"[{dateTime}] Decision {decision} result {result.Prediction} Predicted scores: [{String.Join(",", result.Score)}]");
}
else if (result.Prediction > 0)
{
System.Console.WriteLine($"[{dateTime}] Operation {result.Prediction} Predicted scores: [{String.Join(",", result.Score)}]");
}
}
while (true)
{
Task.Delay(2000).Wait();
var candleslive = Oanda.LoadLiveCandles("EUR_USD", "S5", 2);
foreach (OandaCandle candle in candleslive)
{
DateTime datetime = new DateTime(candle.Time.Year, candle.Time.Month, candle.Time.Day, candle.Time.Hour, candle.Time.Minute, candle.Time.Second);
collection.AddCandle(datetime, candle.Mid.O, candle.Mid.C, candle.Mid.H, candle.Mid.L, candle.Volume);
}
foreach (DateTime dateTime in collection._list.Keys.OrderByDescending(c => c.Ticks).Take(2))
{
var S5Candle = collection.GetCandle(dateTime);
if (S5Candle == null)
{
continue;
}
var M1Candle = minuteconsolidation.GetCandle(dateTime);
if (M1Candle == null)
{
continue;
}
var M15Candle = m15Consolidation.GetCandle(dateTime);
if (M15Candle == null)
{
continue;
}
var S5Stoch = stochasticS5.GetValue(dateTime);
if (S5Stoch == null)
{
continue;
}
var M1Stoch = stochasticM1.GetValue(dateTime);
if (M1Stoch == null)
{
continue;
}
var M15Stoch = stochasticM15.GetValue(dateTime);
if (M15Stoch == null)
{
continue;
}
ModelInput input = new ModelInput()
{
Time = (float)S5Candle.DateTime.TimeOfDay.TotalSeconds,
Decision = 0,
S5V = S5Candle.Volume,
S5C = decimal.ToSingle(S5Candle.CloseRelativ),
S5H = decimal.ToSingle(S5Candle.HighRaltiv),
S5L = decimal.ToSingle(S5Candle.LowRelativ),
S5StochK = (float)(S5Stoch.K),
S5StochD = (float)(S5Stoch.D),
M1V = M1Candle.Volume,
M1C = decimal.ToSingle(M1Candle.CloseRelativ),
M1H = decimal.ToSingle(M1Candle.HighRaltiv),
M1L = decimal.ToSingle(M1Candle.LowRelativ),
M1StochK = (float)M1Stoch.K,
M1StochD = (float)M1Stoch.D,
M15V = M15Candle.Volume,
M15C = decimal.ToSingle(M15Candle.CloseRelativ),
M15H = decimal.ToSingle(M15Candle.HighRaltiv),
M15L = decimal.ToSingle(M15Candle.LowRelativ),
M15StochK = (float)M15Stoch.K,
M15StochD = (float)M15Stoch.D
};
ModelOutput result = predEngine.Predict(input);
System.Console.WriteLine($"[{dateTime}] Operation {result.Prediction} Predicted scores: [{String.Join(",", result.Score)}]");
}
}
}
static void Main(string[] args)
{
var candles = Oanda.LoadCandles("EUR_USD", "S5", DateTime.Now.AddDays(-60));
var collection = new CandleCollection("EUR_USD", Period.S, 5);
var stochasticS5 = new StochasticStats(collection);
var M1Consolditation = new ConsolidatedCandleCollection(collection, Period.M, 1);
var M15Consolditation = new ConsolidatedCandleCollection(collection, Period.M, 15);
var stochasticM1 = new StochasticStats(M1Consolditation);
var H1Consolditation = new ConsolidatedCandleCollection(collection, Period.H, 1);
var stochastich1 = new StochasticStats(H1Consolditation);
var H2Consolditation = new ConsolidatedCandleCollection(collection, Period.H, 2);
var winpip = new PipExpectation(collection, 60);
float count = candles.Where(c => c.Time < DateTime.Now.AddDays(-10)).Count();
float iterator = 0;
float percentiterator = 1;
System.Console.Write("Loading candles cache [");
foreach (OandaCandle candle in candles)
{
DateTime datetime = new DateTime(candle.Time.Year, candle.Time.Month, candle.Time.Day, candle.Time.Hour, candle.Time.Minute, candle.Time.Second);
collection.AddCandle(datetime, candle.Mid.O, candle.Mid.C, candle.Mid.H, candle.Mid.L, candle.Volume);
iterator++;
if (iterator * 10 / count > percentiterator)
{
percentiterator++;
System.Console.Write(".");
}
}
System.Console.WriteLine("] DONE");
FeaturesFactory features = new FeaturesFactory();
features.SetCollection(collection);
features.AddFeaturable(M1Consolditation);
features.AddFeaturable(M15Consolditation);
features.AddFeaturable(H1Consolditation);
features.AddFeaturable(H2Consolditation);
features.AddFeaturable(stochasticM1);
features.AddFeaturable(stochastich1);
features.AddFeaturable(stochasticS5);
features.SetPredition(winpip);
MLContext mlContext = new MLContext();
var regressors = new List <IEstimator <ITransformer> >()
{
mlContext.Regression.Trainers.FastForest(),
mlContext.Regression.Trainers.FastTree(),
mlContext.Regression.Trainers.FastTreeTweedie(),
mlContext.Regression.Trainers.OnlineGradientDescent(),
mlContext.Regression.Trainers.LbfgsPoissonRegression(),
mlContext.Regression.Trainers.LightGbm(numberOfIterations: 300),
mlContext.Regression.Trainers.Sdca(),
mlContext.Regression.Trainers.Gam(),
};
foreach (var regressor in regressors)
{
var model = features.Fit(regressor, DateTime.Now.AddDays(-10));
LearnedTradeStrategy strategy = new LearnedTradeStrategy(features, model);
strategy.Initialize();
TradeSimulation simulation = new TradeSimulation(collection, strategy);
DateTime from = DateTime.Now.AddDays(-7).Date;
DateTime to = DateTime.Now.Date;
simulation.Simulate(from, to);
}
//TradeSimulation simulation = new TradeSimulation(collection, strategy);
//DateTime from = DateTime.Now.AddDays(-7).Date;
//DateTime to = DateTime.Now.Date;
//simulation.Simulate(from, to);
}
static void Main3(string[] args)
{
var candles = Oanda.LoadCandles("EUR_USD", "S5", DateTime.Now.AddDays(-45));
var collection = new CandleCollection("EUR_USD", Period.S, 5);
var minuteconsolidation = new ConsolidatedCandleCollection(collection, Period.M, 1);
var m15Consolidation = new ConsolidatedCandleCollection(collection, Period.M, 15);
var h1consolidation = new ConsolidatedCandleCollection(collection, Period.H, 1);
var stochasticS5 = new StochasticStats(collection);
var stochasticM1 = new StochasticStats(minuteconsolidation);
var stochasticM15 = new StochasticStats(m15Consolidation);
var h1stoch = new StochasticStats(h1consolidation);
var winpip = new PipExpectation(collection, 60);
foreach (OandaCandle candle in candles)
{
DateTime datetime = new DateTime(candle.Time.Year, candle.Time.Month, candle.Time.Day, candle.Time.Hour, candle.Time.Minute, candle.Time.Second);
collection.AddCandle(datetime, candle.Mid.O, candle.Mid.C, candle.Mid.H, candle.Mid.L, candle.Volume);
}
var maxlong = winpip._futurPip.Values.Max(c => c.Long);
var avglong = winpip._futurPip.Values.Average(c => c.Long);
int moreThanSpreadLong = winpip._futurPip.Values.Count(c => c.Long > 5m && c.Short < 1.5m);
var maxshort = winpip._futurPip.Values.Max(c => c.Short);
var avgshort = winpip._futurPip.Values.Average(c => c.Short);
int moreThanSpreadShort = winpip._futurPip.Values.Count(c => c.Short > 5m && c.Long < 1.5m);
var records = new List <dynamic>();
foreach (DateTime dateTime in collection._list.Keys.OrderBy(c => c.Ticks))
{
var S5Candle = collection.GetCandle(dateTime);
if (S5Candle == null)
{
continue;
}
var M1Candle = minuteconsolidation.GetCandle(dateTime);
if (M1Candle == null)
{
continue;
}
var M15Candle = m15Consolidation.GetCandle(dateTime);
if (M15Candle == null)
{
continue;
}
var H1Candle = h1consolidation.GetCandle(dateTime);
if (H1Candle == null)
{
continue;
}
var S5Stoch = stochasticS5.GetValue(dateTime);
if (S5Stoch == null)
{
continue;
}
var M1Stoch = stochasticM1.GetValue(dateTime);
if (M1Stoch == null)
{
continue;
}
var M15Stoch = stochasticM15.GetValue(dateTime);
if (M15Stoch == null)
{
continue;
}
var H1Stoch = h1stoch.GetValue(dateTime);
if (H1Stoch == null)
{
continue;
}
var expected = winpip.GetValue(dateTime);
if (expected == null)
{
continue;
}
dynamic record = new ExpandoObject();
record.Time = S5Candle.DateTime.TimeOfDay.TotalSeconds;
record.S5V = S5Candle.Volume;
record.S5C = decimal.ToSingle(S5Candle.CloseRelativ);
record.S5H = decimal.ToSingle(S5Candle.HighRaltiv);
record.S5L = decimal.ToSingle(S5Candle.LowRelativ);
record.S5StochK = (float)S5Stoch.K;
record.S5StochD = (float)S5Stoch.D;
record.M1V = M1Candle.Volume;
record.M1C = decimal.ToSingle(M1Candle.CloseRelativ);
record.M1H = decimal.ToSingle(M1Candle.HighRaltiv);
record.M1L = decimal.ToSingle(M1Candle.LowRelativ);
record.M1StochK = (float)M1Stoch.K;
record.M1StochD = (float)M1Stoch.D;
record.M15V = M15Candle.Volume;
record.M15C = decimal.ToSingle(M15Candle.CloseRelativ);
record.M15H = decimal.ToSingle(M15Candle.HighRaltiv);
record.M15L = decimal.ToSingle(M15Candle.LowRelativ);
record.M15StochK = (float)M15Stoch.K;
record.M15StochD = (float)M15Stoch.D;
record.H1V = H1Candle.Volume;
record.H1C = decimal.ToSingle(H1Candle.CloseRelativ);
record.H1H = decimal.ToSingle(H1Candle.HighRaltiv);
record.H1L = decimal.ToSingle(H1Candle.LowRelativ);
record.H1StochK = (float)H1Stoch.K;
record.H1StochD = (float)H1Stoch.D;
//record.FPL = expected.Long;
//record.FPS = expected.Short;
if (expected.Long > 5m)
{
record.Decision = 1;
}
//else if (expected.Long > 2m)
//{
// record.Decision = 1;
//}
else
{
record.Decision = 0;
}
records.Add(record);
}
using (var writer = new StreamWriter("train.csv"))
using (var csv = new CsvWriter(writer))
{
csv.Configuration.Delimiter = ",";
csv.WriteRecords(records);
}
MLContext mlContext = new MLContext();
ITransformer mlModel = mlContext.Model.Load("MLModel.zip", out var modelInputSchema);
var predEngine = mlContext.Model.CreatePredictionEngine <ModelInput, ModelOutput>(mlModel);
foreach (DateTime dateTime in collection._list.Keys.OrderBy(c => c.Ticks))
{
var S5Candle = collection.GetCandle(dateTime);
if (S5Candle == null)
{
continue;
}
var M1Candle = minuteconsolidation.GetCandle(dateTime);
if (M1Candle == null)
{
continue;
}
var M15Candle = m15Consolidation.GetCandle(dateTime);
if (M15Candle == null)
{
continue;
}
var H1Candle = h1consolidation.GetCandle(dateTime);
if (H1Candle == null)
{
continue;
}
var S5Stoch = stochasticS5.GetValue(dateTime);
if (S5Stoch == null)
{
continue;
}
var M1Stoch = stochasticM1.GetValue(dateTime);
if (M1Stoch == null)
{
continue;
}
var M15Stoch = stochasticM15.GetValue(dateTime);
if (M15Stoch == null)
{
continue;
}
var H1Stoch = h1stoch.GetValue(dateTime);
if (H1Stoch == null)
{
continue;
}
var expected = winpip.GetValue(dateTime);
if (expected == null)
{
continue;
}
float decision = 0f;
if (expected.Long > 5m && expected.Short < 1.6m)
{
decision = 1;
}
else if (expected.Short > 5m && expected.Long < 1.6m)
{
decision = 2;
}
else
{
decision = 0;
}
ModelInput input = new ModelInput()
{
Time = (float)S5Candle.DateTime.TimeOfDay.TotalSeconds,
Decision = 0,
S5V = S5Candle.Volume,
S5C = decimal.ToSingle(S5Candle.CloseRelativ),
S5H = decimal.ToSingle(S5Candle.HighRaltiv),
S5L = decimal.ToSingle(S5Candle.LowRelativ),
S5StochK = (float)(S5Stoch.K),
S5StochD = (float)(S5Stoch.D),
M1V = M1Candle.Volume,
M1C = decimal.ToSingle(M1Candle.CloseRelativ),
M1H = decimal.ToSingle(M1Candle.HighRaltiv),
M1L = decimal.ToSingle(M1Candle.LowRelativ),
M1StochK = (float)M1Stoch.K,
M1StochD = (float)M1Stoch.D,
M15V = M15Candle.Volume,
M15C = decimal.ToSingle(M15Candle.CloseRelativ),
M15H = decimal.ToSingle(M15Candle.HighRaltiv),
M15L = decimal.ToSingle(M15Candle.LowRelativ),
M15StochK = (float)M15Stoch.K,
M15StochD = (float)M15Stoch.D
};
ModelOutput result = predEngine.Predict(input);
if (result.Prediction > 0 || input.Decision > 0)
{
System.Console.WriteLine($"[{input.Time}] decision {input.Decision} result {result.Prediction} Predicted scores: [{String.Join(",", result.Score)}]");
}
}
}
internal void AddData(Candle candle)
{
_collection.AddCandle(candle);
}