Edit to official CQF backtester

backtest_v2/backtest.py

@@ -161,47 +161,56 @@ def read_data(price_location, view_location):
 
 def backtest(strat_function, starting_value, prices_location, views_location, log, log_name):
     prices, views, names = read_data(prices_location, views_location)
+    #print(prices)
+    #print(views)
+    #print(names)
     acc = Account(starting_value, log, names, log_name)
     strat = Strategy(strat_function)    
-    for ind, (price, view) in enumerate(zip(prices, views)): 
+    for ind, (price, view) in enumerate(zip(prices, views)):
         if ind == 0:
             prev_weights = list(np.zeros(len(prices[0]))) 
         else:
             prev_weights = acc.weights[-1]
-        acc.update(strat.allocations(prices[0: ind  + 1], views[0: ind + 1], prev_weights), price)
 
+        acc.update(strat.allocations(prices[0: ind + 1], views[0: ind + 1], prev_weights), price)
     sharpe = acc.daily_sharpe()
     max_drawdown = acc.max_drawdown()
     calmar = acc.calmar()
     total_return, percent_return = acc.returns()
-
+    if acc.log: 
+        with open(log_name, mode='a', encoding='utf-8', newline='') as f:
+            writer = csv.writer(f)
+            [writer.writerow([idx, trade["Value"], trade["Return"], trade["Weights"]]) for idx, trade in enumerate(acc.trades)]
 
     print(f'Sharpe: {sharpe}')
-    print(f'Max Drawdon: {max_drawdown}')
+    print(f'Max Drawdown: {max_drawdown}')
     print(f'Total Return: {total_return}')
     print(f'Percent Return: {percent_return}')
-    plt.plot(acc.absolute_values)
-    plt.xlabel = "Time"
-    plt.ylabel = "Portfolio Value"
-
-    plt.figure()
-
-
-    plt.plot(np.array(acc.absolute_values[1:])/np.array(acc.absolute_values[:-1]))
-    plt.xlabel = "Time"
-    plt.ylabel = "Returns"
-
-    plt.figure()
-
-    plt.hist(np.array(acc.absolute_values[1:])/np.array(acc.absolute_values[:-1]), bins=50)
     print(f'Max Tick Return: {np.amax(np.array(acc.absolute_values[1:])/np.array(acc.absolute_values[:-1])) - 1}')
     print(f'Min Tick Return: {np.amin(np.array(acc.absolute_values[1:])/np.array(acc.absolute_values[:-1])) - 1}')
     print(f'Average Tick Return: {np.average(np.array(acc.absolute_values[1:])/np.array(acc.absolute_values[:-1])) - 1}')
 
-    if acc.log: 
-        with open(log_name, mode='a', encoding='utf-8', newline='') as f:
-            writer = csv.writer(f)
-            [writer.writerow([idx, trade["Value"], trade["Return"], trade["Weights"]]) for idx, trade in enumerate(acc.trades)]
-
+    fig, axs = plt.subplots(3, 1, constrained_layout=True)
+    axs[0].plot(acc.absolute_values)
+    axs[0].set_title('Portfolio')
+    axs[0].set_xlabel('Time')
+    axs[0].set_ylabel('Portfolio Value')
+    fig.suptitle('Visual Representation of Strategy Performance', fontsize=16)
+
+    axs[1].plot(np.array(acc.absolute_values[1:]) / np.array(acc.absolute_values[:-1]))
+    axs[1].set_xlabel('Time')
+    axs[1].set_title('Returns (side view)')
+    axs[1].set_ylabel('Returns')
+
+    axs[2].hist(np.array(acc.absolute_values[1:]) / np.array(acc.absolute_values[:-1]), bins=50)
+    axs[2].set_xlabel('% Return')
+    axs[2].set_title('Returns (front view)')
+    axs[2].set_ylabel('Frequency')
     plt.show()
+
+    # convert array into dataframe
+    DF = pd.DataFrame(acc.absolute_values)
+
+    # save the dataframe as a csv file
+    DF.to_csv("time_series_data.csv")
 

backtest_v2/demo.py

@@ -1,5 +1,7 @@
 from backtest import backtest
 import numpy as np
+import random
+import pandas as pd
 from scipy.optimize import minimize
 
 
@@ -15,20 +17,26 @@
 
 '''
 def strat_function(preds, prices, last_weights):
-
+    #print(last_weights)
     prices = np.array(prices)
     signal = prices[:, -1][-1]
-    if signal == -1:
-        return [0, 1, 0]
-    if signal == 1:
-        return [0, 0, 0]
-    if signal == 0:
-        return last_weights
-
-
+    #print(signal)
+
+    return [random.uniform(-1, 1),
+            random.uniform(-1, 1),
+            random.uniform(-1, 1),
+            random.uniform(-1, 1),
+            random.uniform(-1, 1),
+            random.uniform(-1, 1),
+            random.uniform(-1, 1),
+            random.uniform(-1, 1),
+            random.uniform(-1, 1)]
 
 '''
 Running the backtest - starting portfolio value of 10000, reading in data from these two locations.
 '''
-backtest(strat_function, 10000, '../test_datasets/SEDG-ENPH-Signal.csv', '../test_datasets/SEDG-ENPH-Signal.csv', True, "log.csv")
+backtest(strat_function, 2000,
+'C:/Users/jacky/OneDrive/Documents/GitHub/backtesting/test_datasets/Actual.csv',
+'C:/Users/jacky/OneDrive/Documents/GitHub/backtesting/test_datasets/Actual.csv',
+         True, "log.csv")