forked from renat/ONNX.Price.Prediction
50 lines
1.4 KiB
Python
50 lines
1.4 KiB
Python
# Copyright 2023, MetaQuotes Ltd.
|
|
# https://www.mql5.com
|
|
|
|
import MetaTrader5 as mt5
|
|
import numpy as np
|
|
import onnxruntime as ort
|
|
import pandas as pd
|
|
from sys import argv
|
|
|
|
if not mt5.initialize():
|
|
print("initialize() failed, error code =",mt5.last_error())
|
|
quit()
|
|
|
|
# you code here
|
|
#
|
|
data_path=argv[0]
|
|
last_index=data_path.rfind("\\")+1
|
|
data_path=data_path[0:last_index]
|
|
print("data path to load onnx model",data_path)
|
|
|
|
eurusd_rates = mt5.copy_rates_from_pos("EURUSD", mt5.TIMEFRAME_H1, 1, 10)
|
|
#print(eurusd_rates)
|
|
df = pd.DataFrame(eurusd_rates)
|
|
X = df[['open', 'high', 'low', 'close']].values
|
|
X = np.expand_dims(X, axis=0)
|
|
#print(X)
|
|
|
|
m = X.mean(axis=1, keepdims=True)
|
|
#print(m)
|
|
s = X.std(axis=1, keepdims=True)
|
|
#print(s)
|
|
X_norm = (X - m) / s
|
|
#print(X_norm)
|
|
|
|
model_path = data_path+"model2.onnx"
|
|
ort_sess = ort.InferenceSession(model_path)
|
|
output_names = [out.name for out in ort_sess.get_outputs()]
|
|
|
|
outputs = ort_sess.run(output_names, {'lstm_input': X_norm.astype(np.float32)})
|
|
y_pred_norm = outputs[0]
|
|
print("raw output:",y_pred_norm)
|
|
|
|
# --- y_pred ---{
|
|
# y_pred = np.round(y_pred_norm.flatten() * ((s[:, 0, 1] - s[:, 0, 0]) - (s[:, 0, 0] - s[:, 0, 2])) + ((m[:, 0, 1] - m[:, 0, 0]) - (m[:, 0, 0] - m[:, 0, 2])), decimals=5)
|
|
# ---
|
|
y_pred = np.round(y_pred_norm.flatten(), decimals=5)
|
|
# --- y_pred ---}
|
|
print("predicted:",y_pred)
|
|
|
|
mt5.shutdown()
|