Merge pull request #82 from hudson-and-thames/develop

Release for 0.8.0

.bumpversion.cfg

@@ -1,5 +1,5 @@
 [bumpversion]
-current_version = 0.7.0
+current_version = 0.8.0
 commit = True
 tag = True
 tag_name = {new_version}

.circleci/config.yml

@@ -1,10 +1,15 @@
 version: 2.1
 
+workflows:
+  version: 2
+  main:
+    jobs:
+      - run_tests
+
 jobs:
   run_tests:
-    machine:
-      image: ubuntu-2004:202010-01
-    resource_class: medium
+    docker:
+      - image: circleci/python:3.8
     steps:
       # Step 1: obtain repo from GitHub
       - checkout
@@ -34,7 +39,3 @@ jobs:
       - store_artifacts:
           path: test-reports
 
-workflows:
-  main:
-    jobs:
-      - run_tests

.coveragerc

@@ -9,6 +9,7 @@ omit =
      arbitragelab/util/segment.py
      tests/*
      /opt/conda/*
+     venv/*
      # Ensure we exclude any files in .local
      */.local/*
      */tmp/*

arbitragelab/cointegration_approach/johansen.py

@@ -87,12 +87,13 @@ def fit(self, price_data: pd.DataFrame, dependent_variable: str = None, det_orde
             hedge_ratios = cointegration_vectors.iloc[vector].to_dict()
             for ticker, ratio in hedge_ratios.items():
                 if ticker != dependent_variable:
-                    hedge_ratios[ticker] = -ratio / hedge_ratios[dependent_variable]
-            hedge_ratios[dependent_variable] = 1.0
+                    # Set value to be list to make it easier to read into pandas DataFrame
+                    hedge_ratios[ticker] = [-ratio / hedge_ratios[dependent_variable]]
+            # Set value to be list to make it easier to read into pandas DataFrame
+            hedge_ratios[dependent_variable] = [1.0]
 
-            # Add all to one dataframe
-            all_hedge_ratios = all_hedge_ratios.append(hedge_ratios, ignore_index=True)
-            all_hedge_ratios = all_hedge_ratios[price_data.columns]
+            # Concat together in one DataFrame
+            all_hedge_ratios = pd.concat([all_hedge_ratios, pd.DataFrame(hedge_ratios)])
 
         self.hedge_ratios = all_hedge_ratios
 

arbitragelab/cointegration_approach/minimum_profit.py

@@ -56,7 +56,7 @@ def set_train_dataset(self, price_df: pd.DataFrame):
             raise Exception("Data Format Error. Should only contain two price series.")
 
         # Verify the index is indeed pd.DatetimeIndex
-        assert price_df.index.is_all_dates, "Index is not of pd.DatetimeIndex type."
+        assert isinstance(price_df.index, pd.DatetimeIndex), "Index is not of pd.DatetimeIndex type."
 
         self.price_df = price_df
 

arbitragelab/copula_approach/base.py

@@ -8,6 +8,8 @@
 
 # pylint: disable = invalid-name, too-many-function-args
 from abc import ABC, abstractmethod
+from matplotlib.axes import Axes
+from matplotlib.figure import Figure
 
 import numpy as np
 import pandas as pd
@@ -202,7 +204,7 @@ def _get_param(self):
         """
 
     @staticmethod
-    def _3d_surface_plot(x: np.array, y: np.array, z: np.array, bounds: list, title: str, **kwargs) -> plt.axis:
+    def _3d_surface_plot(x: np.array, y: np.array, z: np.array, bounds: list, title: str, **kwargs) -> Figure:
         """
         Helper function to plot 3-d plot.
 
@@ -223,13 +225,12 @@ def _3d_surface_plot(x: np.array, y: np.array, z: np.array, bounds: list, title:
         ax.set_ylim(bounds)
         ax.plot_surface(x, y, z, **kwargs)
         plt.title(title)
-        plt.show()
 
-        return ax
+        return fig
 
     @staticmethod
     def _2d_contour_plot(x: np.array, y: np.array, z: np.array, bounds: float, title: str,
-                         levels: list, **kwargs) -> plt.axis:
+                         levels: list, **kwargs) -> Figure:
         """
         Helper function to plot 2-d contour plot.
 
@@ -243,15 +244,14 @@ def _2d_contour_plot(x: np.array, y: np.array, z: np.array, bounds: float, title
         :return: (plt.axis) Axis object.
         """
 
-        plt.figure()
+        fig = plt.figure()
         contour_plot = plt.contour(x, y, z, levels, colors='k', linewidths=1., linestyles=None, **kwargs)
         plt.clabel(contour_plot, fontsize=8, inline=1)
         plt.xlim(bounds)
         plt.ylim(bounds)
         plt.title(title)
-        plt.show()
 
-        return contour_plot
+        return fig
 
     def plot_cdf(self, plot_type: str = '3d', grid_size: int = 50, levels: list = None, **kwargs) -> plt.axis:
         """
@@ -293,7 +293,7 @@ def plot_cdf(self, plot_type: str = '3d', grid_size: int = 50, levels: list = No
 
         return ax
 
-    def plot_scatter(self, num_points: int = 100) -> plt.axis:
+    def plot_scatter(self, num_points: int = 100) -> Axes:
         """
         Plot copula scatter plot of generated pseudo-observations.
 
@@ -307,7 +307,7 @@ def plot_scatter(self, num_points: int = 100) -> plt.axis:
 
         return ax
 
-    def plot_pdf(self, plot_type: str = '3d', grid_size: int = 50, levels: list = None, **kwargs) -> plt.axis:
+    def plot_pdf(self, plot_type: str = '3d', grid_size: int = 50, levels: list = None, **kwargs) -> Figure:
         """
         Plot either '3d' or 'contour' plot of copula PDF.
 

arbitragelab/distance_approach/basic_distance_approach.py

@@ -494,7 +494,7 @@ def find_portfolios(data, pairs):
         of the second element.
 
         :param data: (pd.DataFrame) Dataframe with price series for elements.
-        :param pairs: (list) List of tuples with two elements to use for calculation.
+        :param pairs: (list) List of tuples with two str elements to use for calculation.
         :return: (pd.DataFrame) Dataframe with pairs as columns and their portfolio
             values as rows.
         """
@@ -504,20 +504,9 @@ def find_portfolios(data, pairs):
 
         # Iterating through pairs
         for pair in pairs:
-            # Getting two price series for elements in a pair
-            par = data[list(pair)]
-
             # Difference between price series - a portfolio
-            par_diff = par.iloc[:, 0] - par.iloc[:, 1]
-
-            # Naming the portfolio
-            par_diff.name = str(pair)
-
-            # Adding portfolio series to dataframe
-            portfolios = portfolios.append(par_diff)
-
-        # Transposing to make portfolios as columns
-        portfolios = portfolios.transpose()
+            par_diff = data.loc[:, pair[0]] - data.loc[:, pair[1]]
+            portfolios[str(pair)] = par_diff
 
         return portfolios
 
@@ -579,10 +568,6 @@ def signals(portfolios, variation, divergence):
             portfolio['target_quantity'] = portfolio['long_units'] + portfolio['short_units']
 
             # Adding target quantity to signals dataframe
-            signals = signals.append(portfolio['target_quantity'])
-
-        # Adjusting the final signals dataframe
-        signals = signals.transpose()
-        signals.columns = portfolios.columns
+            signals[str(pair)] = portfolio['target_quantity']
 
         return signals

arbitragelab/distance_approach/pearson_distance_approach.py

@@ -279,7 +279,7 @@ def _data_preprocess(self, price_data, risk_free, phase='train'):
 
         else:
             self.test_monthly_return = monthly_return
-            self.risk_free = self.risk_free.append(risk_free)
+            self.risk_free = pd.concat([self.risk_free, risk_free])
 
     def _beta_pairs_formation(self, num_pairs, weight):
         """

arbitragelab/ml_approach/neural_networks.py

@@ -13,6 +13,11 @@
 logging.getLogger('tensorflow').setLevel(logging.ERROR)
 os.environ['TF_CPP_MIN_LOG_LEVEL'] = '3'
 
+
+# Importing needed packages
+import tensorflow as tf
+from keras.layers import Input, LSTM, Dense, Activation, Lambda
+from keras.models import Model
 import matplotlib.pyplot as plt
 
 from arbitragelab.util import segment
@@ -27,12 +32,6 @@ def __init__(self):
         Initializing variables.
         """
 
-        # Importing needed packages
-        import tensorflow as tf
-        from keras.models import Model
-        from keras.callbacks import History
-        from keras.layers import Input, LSTM, Dense, Activation, Lambda
-
         self.fitted_model = None
 
     def fit(self, *args, **kwargs):
@@ -109,10 +108,6 @@ def build(self):
         :return: (Model) Resulting model.
         """
 
-        # Importing needed packages
-        from keras.models import Model
-        from keras.layers import Input, Dense
-
         input_layer = Input((self.frame_size,))
 
         hidden_layer = Dense(self.hidden_size,
@@ -130,6 +125,7 @@ def build(self):
 
         return model
 
+
 class RecurrentNeuralNetwork(BaseNeuralNetwork):
     """
     Recurrent Neural Network implementation.
@@ -175,10 +171,6 @@ def build(self):
         :return: (Model) Resulting model.
         """
 
-        # Importing needed packages
-        from keras.models import Model
-        from keras.layers import Input, LSTM, Dense
-
         input_layer = Input(self.input_shape)
 
         hidden_layer = LSTM(self.hidden_size, activation=self.hidden_layer_activation_function,
@@ -196,6 +188,7 @@ def build(self):
 
         return model
 
+
 class PiSigmaNeuralNetwork(BaseNeuralNetwork):
     """
     Pi Sigma Neural Network implementation.
@@ -240,10 +233,6 @@ def build(self):
         :return: (Model) Resulting model.
         """
 
-        # Importing needed packages
-        from keras.models import Model
-        from keras.layers import Input, Dense, Activation, Lambda
-
         input_layer = Input((self.frame_size,))
 
         second_sigma_layer = Dense(self.hidden_size,
@@ -272,9 +261,6 @@ def _pi_this(tensor):
         :return: (tf.Tensor) Product of input tensor.
         """
 
-        # Importing needed packages
-        import tensorflow as tf
-
         prod = tf.math.reduce_prod(tensor, keepdims=True, axis=1)
 
         return prod

arbitragelab/ml_approach/optics_dbscan_pairs_clustering.py

@@ -228,8 +228,6 @@ def plot_3d_scatter_plot(self, tsne_df: pd.DataFrame, no_of_classes: int,
         # Set the chart title.
         ax_object.set_title('Automatic Clustering\n' + method)
 
-        plt.show()
-
         return ax_object
 
     def plot_2d_scatter_plot(self, fig: Figure, tsne_df: pd.DataFrame, no_of_classes: int,
@@ -287,8 +285,6 @@ def plot_2d_scatter_plot(self, fig: Figure, tsne_df: pd.DataFrame, no_of_classes
         # Set the chart title.
         ax_object.set_title('Automatic Clustering\n' + method)
 
-        plt.show()
-
         return ax_object
 
     def plot_knee_plot(self) -> Axes: