Add files via upload

helpers/oanda_api_helpers.py

@@ -22,8 +22,9 @@ class TradingSession:
     # initiate objects
     def __init__(self, accountID, access_token):
         self.accountID = accountID
+        self.access_token = access_token
         self.api = oandapyV20.API(access_token=access_token, environment="practice")
-        self.order_book = self.sync_open_positions()
+        self.order_book = self.oanda_order_book()
 
     # initiate methods
     def send_request(self, request):
@@ -62,7 +63,7 @@ def open_order(self, instrument, units):
         # check if request was fulfilled and save its ID
         if request_data is not 1:
             instrument = request_data['orderCreateTransaction']['instrument']
-            self.order_book[instrument]['tradeID'] = (request_data['lastTransactionID'])
+            self.order_book[instrument]['tradeID'] = request_data['lastTransactionID']
             self.order_book[instrument]['order_type'] = -1 if units < 0 else 1
             print('{}: {}'.format('Long' if units > 0 else 'Short', instrument))
             return 0
@@ -98,7 +99,7 @@ def check_account_summary(self):
         r = accounts.AccountSummary(self.accountID)
         return self.send_request(r)
 
-    def sync_open_positions(self):
+    def oanda_order_book(self):
         """Synchronize open positions with this object's order_book"""
         order_book_oanda = self.check_open_positions()
         order_book = {'EUR_USD': {'order_type': None, 'tradeID': None},
@@ -114,6 +115,37 @@ def sync_open_positions(self):
             order_book[pos['instrument']]['order_type'] = order_type
         return order_book
 
+    def sync_with_oanda(self):
+        self.order_book = self.oanda_order_book()
+
+    def close_all_open_positions(self):
+        """Close all opened positions"""
+
+        # check oanda for open positions
+        try:
+            open_positions = self.check_open_positions()['positions'][0]
+        except IndexError:
+            self.order_book = self.oanda_order_book()
+            print('No opened positions')
+            return 0
+
+        # get ID's of open positions
+        trade_ids = []
+        try:
+            [trade_ids.append(x) for x in open_positions['short']['tradeIDs']]
+        except KeyError:
+            pass
+        try:
+            [trade_ids.append(x) for x in open_positions['long']['tradeIDs']]
+        except KeyError:
+            pass
+
+        # close orders by ID
+        [close_order_manually(self.accountID, self.access_token, x) for x in trade_ids]
+        self.order_book = self.oanda_order_book()
+        print('All positions closed')
+        return 0
+
 
 def close_order_manually(accountID, access_token, tradeID):
     """

helpers/utils.py

@@ -93,20 +93,33 @@ def portfolio_value(price, signal, trans_costs=0.0001):
     return value
 
 
-def get_data_batch(list_of_items, batch_size):
-    """Returns a batch of sequential data"""
-    indexes = np.random.choice(len(list_of_items[0]) - (batch_size+1))
+def get_data_batch(list_of_items, batch_size, sequential):
+    """Returns a batch of data. A batch of sequence or random points."""
+    if sequential:
+        indexes = np.random.randint(len(list_of_items[0]) - (batch_size+1))
+    else:
+        indexes = np.random.randint(0, len(list_of_items[0]), batch_size)
     batch_list = []
     for item in list_of_items:
-        batch = item[indexes:indexes+batch_size, ...]
+        batch = item[indexes:indexes+batch_size, ...] if sequential else item[indexes, ...]
         batch_list.append(batch)
     return batch_list
 
 
 def get_lstm_input_output(x, y, time_steps):
-    """Returns a batch of sequential data for lstm shaped like [batch_size, time_steps, features]"""
+    """Returns sequential lstm shaped data [batch_size, time_steps, features]"""
     data_points, _ = np.shape(x)
     x_batch_reshaped = []
     for i in range(data_points - time_steps):
         x_batch_reshaped.append(x[i: i+time_steps, :])
     return np.array(x_batch_reshaped), y[time_steps:]
+
+
+def get_cnn_input_output(x, y, time_steps=12):
+    """Returns sequential cnn shaped data [batch_size, features, time_steps]"""
+    data_points, _ = np.shape(x)
+    x_batch_reshaped = []
+    for i in range(data_points - time_steps):
+        x_batch_reshaped.append(x[i:i+time_steps, :])
+    x_batch_reshaped = np.transpose(np.array([x_batch_reshaped]), axes=(1, 3, 2, 0))
+    return np.array(x_batch_reshaped), y[time_steps:]