refactoring and new function stubs`

Removed the distance functions class from the distance method. Added a few new stubs to basicdistanceapproach.

Brokerage/AbstractBrokerage.py

@@ -0,0 +1,5 @@
+'''
+This module will serve as an abstract class defining the methods on a brokerage fee structure class
+'''
+class AbstractBrokerage():
+    pass

Brokerage/InteractiveBrokers.py

@@ -0,0 +1,5 @@
+'''
+This class will simulate the fee structure associated with interactive brokers
+'''
+class InteractiveBrokers():
+    pass

PairsTrading/DistanceApproach/BasicDistanceApproach.py

@@ -1,32 +1,28 @@
 import pandas as pd
 from pandas.api.types import is_numeric_dtype
+from math import sqrt
+from DistanceFunctions import DistanceFunctions
+
+
 
-'''
-This class stores distance functions that can be passed to the calculate pairs function of the DistanceApproach Class.
-'''
-class DistanceFunctions():
-    def __init(self) -> None:
-        pass
-    '''
-    Simple Euclidean Distance
-    '''
-    def euclideanDistance(a,b):
-        return (b-a)**2
 
 '''
-This class implements the distance approach to a pairs trading signal in 3 steps:
-- Orders and ranks pairs according to some distance metric
-- Generates a signal according to a threshold
+This class implements the basic distance approach to a pairs trading signal in 3 steps:
+- normalizes price data
+- Orders proposed pairs according to a distance metric
+- Generates a signal according to a deviation threshold
 - Outputs a backtest according to a brokerage fee template
 '''
 class BasicDistanceApproach():
 
     def __init__(self) -> None:
+        self.pairs = None
         self.pairsData = None
         self.tradeData = None
         self.NotEnoughColumnsException = 'Dataframe is missing columns. Check that you have both securities and date columns'
         self.dateTimeException = 'Left-Most Column is not of type datetime64'
         self.nonNumericalException = 'Detected non-numerical data-types to the right of date column'
+        self.signals = None
 
 
 
@@ -57,7 +53,7 @@ def setPairsData(self, data: pd.DataFrame) -> bool:
     - If there are non-numerical data types 
     Returns true if we can successfully set the pairs data
     '''
-    def setTradeData(self, data: pd.DataFrame) -> bool:
+    def setSignalData(self, data: pd.DataFrame) -> bool:
         if self.isDataWellFormed(data):
             self.tradeData = data
             return True
@@ -80,9 +76,36 @@ def isDataWellFormed(data: pd.DataFrame) -> bool:
 
     '''
     Generates Pairs:
-    Returns a dataframe with all the pair names and their distance values
+    - Build list of all possible pairs
+    - Sort list according to distance function
+    - Sets the top n best pairs as the pairs field in the class
+    Args:
+    - distanceFunc: a function that measures the distance (see DistanceFunctions.py)
+    - top: the number of pairs to select from the n best candidates
+
+    The pairs field is a list of data frames structured as follows:
+    Date | Asset_Price1 | Asset_Price2 | Asset_Price1_Normalized | Asset_Price2_Normalized | Normalized_Spread
+    '''
+    def generatePairs(self, distanceFunc: DistanceFunctions.__call__, top: int):
+        pass
+
+    '''
+    Retrieves the stored pairs in the object. Raises an exception if the pairs have not yet been generated.
+    '''
+    def getPairs(self):
+        if self.pairs != None:
+            return self.pairs
+        else:
+            raise Exception('Error: Pairs have not yet been generated.')
+
+    '''
+    Adds a signal column to each data frame within the pairs list (self.pairs).
+    A signal  is in the form of an integer:
+    1 -> Buy the spread
+    -1 -> Sell the spread
+    0 -> Close the position
     '''
-    def generatePairs(self,distanceFunc):
+    def addTradeSignals():
         pass
 
 

PairsTrading/DistanceApproach/DistanceFunctions.py

@@ -0,0 +1,14 @@
+import pandas as pd
+
+'''
+This class stores distance functions that can be passed to the calculate pairs function of the DistanceApproach Class.
+'''
+class DistanceFunctions():
+    def __init__(self) -> None:
+        pass
+    '''
+    The sum of the squared distances between two securities
+    '''
+    def sumSquaredDistance(self, priceSeriesA: pd.Series, priceSeriesB: pd.Series) -> float:
+        squaredDifferenceSum = ((priceSeriesB-priceSeriesA)**2)
+        return sum(squaredDifferenceSum)

PairsTrading/DistanceApproach/TestDistanceFunctions.py

@@ -0,0 +1,26 @@
+from datetime import date, datetime
+import unittest
+from DistanceFunctions import DistanceFunctions
+import pandas as pd
+'''
+Tests the distance functions
+'''
+class TestDistanceFunctions(unittest.TestCase):
+    '''
+    Tests the sum of squared difference function
+    '''
+    def testSumSquaredDifference(self):
+        d = DistanceFunctions()
+        exampleDate1 = pd.to_datetime(datetime(2019,1,1))
+        exampleDate2 = pd.to_datetime(datetime(2019,1,2))
+        exampleDate3 = pd.to_datetime(datetime(2019,1,3))
+        exampleDate4 = pd.to_datetime(datetime(2019,1,4))
+        priceDF = pd.DataFrame({'Date' : [exampleDate1, exampleDate2, exampleDate3, exampleDate4], 'Price1' : [1, 2, 1, 3],'Price2' : [1, 3, 4, 1]}) 
+
+        sumSquares = d.sumSquaredDistance(priceDF["Price1"],priceDF["Price2"])
+        self.assertEqual(sumSquares,14)
+
+
+if __name__ == '__main__':
+    unittest.main()
+