ONNX.Price.Prediction/ONNX.Price.Prediction2.mq5

<EFBFBD><EFBFBD>//+------------------------------------------------------------------+
//|                                        ONNX.Price.Prediction.mq5 |
//|                                  Copyright 2023, MetaQuotes Ltd. |
//|                                             https://www.mql5.com |
//+------------------------------------------------------------------+
#property copyright "Copyright 2023, MetaQuotes Ltd."
#property link      "https://www.mql5.com"
#property version   "1.00"

#define SAMPLE_SIZE 10  // bars count in sample as defined in Python\PricePredictionTraining.py
// X, y = collect_dataset(df, history_size=10)

#resource "Python/model2.onnx" as uchar model[]

//+------------------------------------------------------------------+
//| Script for demonstration of identical execution of               |
//| Python\PricePrediction.py and                                    |
//| ONNX.Price.Prediction.mq5                                        |
//+------------------------------------------------------------------+
int OnStart(void)
  {
//--- create a model from static buffer
//--- model.onnx file from resources is represented as const uchar model[89591]
   long handle=OnnxCreateFromBuffer(model,ONNX_DEFAULT);
   if(handle==INVALID_HANDLE)
     {
      Print("OnnxCreateFromBuffer error ",GetLastError());
      return(-1);
     }

//--- since not all sizes defined in the input tensor we must set them explicitly
//--- first index - batch size, second index - series size, third index - number of series (OHLC)
   const long input_shape[] = {1,SAMPLE_SIZE,4};
   if(!OnnxSetInputShape(handle,0,input_shape))
     {
      Print("OnnxSetInputShape error ",GetLastError());
      OnnxRelease(handle);
      return(-1);
     }

//--- since not all sizes defined in the output tensor we must set them explicitly
//--- first index - batch size, must match the batch size of the input tensor
//--- second index - number of predicted prices (we only predict Close)
   const long output_shape[] = {1,1};
   if(!OnnxSetOutputShape(handle,0,output_shape))
     {
      Print("OnnxSetOutputShape error ",GetLastError());
      OnnxRelease(handle);
      return(-1);
     }

//--- prepare input data
   matrix  rates;                         // matrix to get test sample OHLC
   matrix  mm(SAMPLE_SIZE,4);             // matrix of horizontal vectors Mean
   matrix  ms(SAMPLE_SIZE,4);             // matrix of horizontal vectors Std
   matrix  x_norm(SAMPLE_SIZE,4);         // matrix for prices normalize
   matrixf input_data(SAMPLE_SIZE,4);     // matrix for prepared input data
   vectorf output_data(1);                // vector to get result

//--- request last bars
   if(!rates.CopyRates("EURUSD",PERIOD_H1,COPY_RATES_OHLC,1,SAMPLE_SIZE))
     {
      Print("CopyRates error ",GetLastError());
      OnnxRelease(handle);
      return(-1);
     }
//--- the input of the model must be a set of vertical OHLC vectors
   x_norm=rates.Transpose();
//Print(x_norm);
//--- get series Mean
   vector m=x_norm.Mean(0);
//Print(m);
//--- get series Std
   vector s=x_norm.Std(0);
//Print(s);
//--- prepare matrices for prices normalization
   for(int i=0; i<SAMPLE_SIZE; i++)
     {
      mm.Row(m,i);
      ms.Row(s,i);
     }
//--- get normalized prices
   x_norm-=mm;
   x_norm/=ms;
//--- run the inference
   input_data.Assign(x_norm);
   if(!OnnxRun(handle,ONNX_NO_CONVERSION,input_data,output_data))
     {
      Print("OnnxRun error ",GetLastError());
      OnnxRelease(handle);
      return(-1);
     }

   Print("raw output: ",output_data);
//--- denormalize the price from the output value
   double y_pred = output_data[0] * ((s[1] - s[0]) - (s[0] - s[2])) + ((m[1] - m[0]) - (m[0] - m[2]));

   Print("predicted: ",y_pred);
//--- finalize
   OnnxRelease(handle);
   return(0);
  }
//+------------------------------------------------------------------+