y.append(original[i])
y_max.append(max_cumulative[i])
y_min.append(min_cumulative[i])
y_max_sharpe.append(max_sharpe[i])
y_min_sharpe.append(min_sharpe[i])
y_index.append(price_index[j])
# Draw Graphs
fig = plt.figure(figsize=(15, 6)) # Create a figure
title = "Piotroski Optimization: Growth of " + str(
initial_amount / 1000000) + " million"
plt.suptitle(title, fontsize=18)
# Cumulative Return
ax1 = plt.subplot(1, 2, 1) # Create axis
ax1.plot(x_axis,
functions.relative(y_max),
label=max_name[:-4] + ': Max Cumulative Return')
ax1.plot(x_axis,
functions.relative(y_min),
label=min_name[:-4] + ': Min Cumulative Return')
ax1.plot(x_axis, functions.relative(y), label='original weight')
ax1.plot(x_axis, functions.relative(y_index), label=market)
ax1.xaxis.set_major_locator(
plt.MaxNLocator(5)) # Set Maximum number of x-axis values to show
ax1.legend(loc='best')
ax1.grid(True)
plt.title("Max Cumulative Return", fontsize=12)
plt.ylabel("Cumulative Return", fontsize=8)
# Sharpe Ratio
ax2 = plt.subplot(1, 2, 2) # Create axis
ax2.plot(x_axis,
for i in range(len(timestamp)):
for j in range(len(date_index)):
if date_index[j] == timestamp[i]:
correctdate.append(i)
correctdate2.append(j)
# Prepare for a graph
x_axis = []
y = [] # Pure Market Trending
strategy1 = [] # Weight Scheme 1
strategy2 = [] # Weight Scheme 2
strategy3 = [] # Weight Optimized
y_index = [] # Market Index Portfolio
# Normalize to initial amount and fill up values for y-axis by searching for correct dates
portfolio_pure = functions.relative(portfolio_pure)
portfolio_weight1 = functions.relative(portfolio_weight1)
portfolio_weight2 = functions.relative(portfolio_weight2)
portfolio_weight_optimized = functions.relative(portfolio_weight_optimized)
for i in range(len(correctdate)):
correctindex = correctdate[i]
y.append(portfolio_pure[correctindex])
strategy1.append(portfolio_weight1[correctindex])
strategy2.append(portfolio_weight2[correctindex])
strategy3.append(portfolio_weight_optimized[correctindex])
for j in range(len(correctdate2)):
correctindex = correctdate2[j]
y_index.append(price_index[correctindex])
x_axis.append(date_index[correctindex])
title_stats = [
correctdate = []
correctdate2 = []
for i in range(len(timestamp)):
for j in range(len(date_index)):
if date_index[j] == timestamp[i]:
correctdate.append(i)
correctdate2.append(j)
# Prepare for a graph
x_axis = date_index
y1 = [] # Weight Scheme1 Portfolio
y2 = [] # Weight Scheme2 Portfolio
y_index = [] # Market Index Portfolio
# Normalize to initial amount and fill up values for y-axis by searching for correct dates
relative_portfolio = functions.relative(portfolio)
relative_portfolio2 = functions.relative(portfolio2)
for i in range(len(correctdate)):
correctindex = correctdate[i]
y1.append(relative_portfolio[correctindex])
y2.append(relative_portfolio2[correctindex])
for j in range(len(correctdate2)):
correctindex = correctdate2[j]
y_index.append(price_index[correctindex])
y_index = functions.relative(y_index)
# Plot a graph
fig = plt.figure() # Create a figure
ax = plt.axes() # Create axis
plt.plot(x_axis, y1, label='Weight Scheme 1')
def backtest(weight_scheme, number):
numberofday = 0
quarter_list = []
# Initialize matrix full of zeros
resultMatrix = [[0 for x in range(len(ticker) + 5)]
for y in range(len(timestamp) + 1)]
# Change the first row of result matrix to show headers
firstrow = resultMatrix[0]
secondrow = resultMatrix[1]
for i0 in range(len(firstrow)):
if i0 == 0:
firstrow[i0] = ""
elif i0 < len(ticker) + 1:
firstrow[i0] = ticker[i0 - 1]
elif i0 == len(ticker) + 1:
firstrow[i0] = "Daily Return"
elif i0 == len(ticker) + 2:
firstrow[i0] = "Cumulative"
elif i0 == len(ticker) + 3:
firstrow[i0] = "Stock Count"
elif i0 == len(ticker) + 4:
firstrow[i0] = "Portfolio"
secondrow[i0] = initial_amount
# Change the first column of result matrix to show time index
for row in range(len(resultMatrix)):
if row == 0:
resultMatrix[row][0] = ""
else:
resultMatrix[row][0] = timestamp[row - 1]
resultMatrix = np.array(resultMatrix)
column_return = np.array(resultMatrix[:, -1][1:]) # Portfolio
column_return = np.asfarray(column_return, float)
column_return[0] = initial_amount
# Loop through time index
for index in range(len(timestamp)):
bool_quarter = False
stockcount = 0
answer = 0 # search for index for weights
month = int(timestamp[index].split("/")[1])
year = int(timestamp[index].split("/")[2])
if month < 7:
quarter = "Q2"
else:
quarter = "Q4"
current = str(year) + quarter
# Q-2 quarter
if month < 4:
year -= 1
quarter = "Q2"
elif month < 7:
year -= 1
quarter = "Q4"
elif month < 10:
year -= 1
quarter = "Q4"
else:
quarter = "Q2"
date = str(year) + quarter
# Check if quarter is different from previous day's quarter
quarter_list.append(current)
if number == 1:
if index == 0:
store_index.append([index, current])
else:
if quarter_list[index] != quarter_list[index - 1]:
store_index.append([index, current])
if (index
== 0) or (index != 0
and quarter_list[index] != quarter_list[index - 1]):
bool_quarter = True
# Search for index for weights
for index2 in range(len(weight_scheme)):
if date == weight_scheme[index2][0]:
answer = index2
# Loop through stocks
aggregate = 0
for stock in range(len(ticker)):
gain = functions.percentageChange(
float(price[index - 1][stock + 1]),
float(price[index][stock + 1]))
if answer == 0:
stockweight = 0
else:
stockweight = float(weight_scheme[answer][stock + 1]) / 100
if bool_quarter: # When quarter changes
if index == 0:
value = initial_amount * stockweight
else:
value = column_return[index - 1] * stockweight
else: # When quarter does not change
value = float(
resultMatrix[index][stock + 1]) * (1 + gain / 100)
resultMatrix[index + 1][
stock +
1] = value # Fill in the money allocation instead of percentage change
aggregate += value
# Calculate number of stocks to be invested
if stockweight != 0 and float(price[index][stock + 1]) != 0:
stockcount += 1
if aggregate == 0:
aggregate = column_return[index - 1]
if column_return[index - 1] == 0 or index == 0:
aggregate = initial_amount
column_return[index] = aggregate
# Stock Count
resultMatrix[index + 1][len(ticker) + 3] = stockcount
numberofday += 1
resultMatrix[:, -1][1:] = column_return
daily_return = functions.percentagelist(column_return)
resultMatrix[:, -4][2:] = daily_return
cumulative = functions.cumulative_list(daily_return)
resultMatrix[:, -3][2:] = cumulative
portfolio_list = column_return
# Take the last index to get cumulative_return
cumulative_return = cumulative[-1] # Final cumulative return
# Calculate related stats
head = "Weight Scheme " + str(number)
print(head, ":")
print(numberofday, "days;", round(cumulative_return, 4), "%")
print("\n")
print("Average Daily Return =", round(average(daily_return), 4), "%")
print("SD of Daily Return =", round(SD(daily_return), 4), "%")
annual_return = average(daily_return) * oneyear
print("Annualized Return =", round(annual_return, 4), "%")
annual_volatility = SD(daily_return) * (oneyear**(1 / 2))
print("Annualized Volatility =", round(annual_volatility, 4), "%")
sharpe_ratio = sharpe(daily_return)
print("Annual Sharpe Ratio: ", round(sharpe_ratio, 4))
print("Max Consecutive Loss = ", maxconsecutive(daily_return), "days")
print("Win Rate:", round(winrate(daily_return), 4), "%")
max_drawdown = functions.findMin(daily_return)
print("Max Drawdown:", round(max_drawdown, 4), "%")
print("\n")
stats_line = [
head, cumulative_return, annual_return, annual_volatility,
sharpe_ratio, max_drawdown
]
stats_matrix.append(stats_line)
if number == 2:
# Calculate stats for market index
daily_return_index = functions.percentagelist(price_index)
cumulative_index = functions.percentageChange(price_index[0],
price_index[-1])
print(market, "Index: ")
print("Cumulative =", round(cumulative_index, 4), "%")
print("Average Daily Return =", round(average(daily_return_index), 4),
"%")
print("SD of Daily Return =", round(SD(daily_return_index), 4), "%")
annual_return = average(daily_return_index) * oneyear
print("Annualized Return =", round(annual_return, 4), "%")
annual_volatility = SD(daily_return_index) * (oneyear**(1 / 2))
print("Annualized Volatility =", round(annual_volatility, 4), "%")
sharpe_ratio = sharpe(daily_return_index)
print("Annual Sharpe Ratio: ", round(sharpe_ratio, 4))
print("Max Consecutive Loss = ", maxconsecutive(daily_return_index),
"days")
print("Win Rate:", round(winrate(daily_return_index), 4), "%")
max_drawdown = functions.findMin(daily_return)
print("Max Drawdown:", round(max_drawdown, 4), "%")
stats_line = [
market, cumulative_index, annual_return, annual_volatility,
sharpe_ratio, max_drawdown
]
stats_matrix.append(stats_line)
# Find correct date to match dates for market index and our portfolio
correctdate = []
correctdate2 = []
for i in range(len(timestamp)):
for j in range(len(date_index)):
if date_index[j] == timestamp[i]:
correctdate.append(i)
correctdate2.append(j)
# Prepare for a graph
x_axis = date_index
y = [] # Strategy Portfolio
y_index = [] # Market Index Portfolio
# Normalize to initial amount and fill up values for y-axis by searching for correct dates
relative_portfolio = functions.relative(portfolio_list)
for i in range(len(correctdate)):
correctindex = correctdate[i] - 1
y.append(relative_portfolio[correctindex])
for j in range(len(correctdate2)):
correctindex = correctdate2[j]
y_index.append(price_index[correctindex])
y_index = functions.relative(y_index)
# Plot a graph
fig = plt.figure() # Create a figure
ax = plt.axes() # Create axis
if len(x_axis) == len(y):
plt.plot(x_axis, y, label='Weight Scheme ' + str(number))
plt.plot(x_axis, y_index, label=market)
ax.xaxis.set_major_locator(
plt.MaxNLocator(5)) # Set Maximum number of x-axis values to show
fig.autofmt_xdate() # Rotate values to see more clearly
title = "Stock Selection: Growth of " + str(
initial_amount / 1000000) + " million"
plt.title(title)
plt.ylabel("Cumulative Return")
fig.savefig(directory + market + " Stock Selection Weight Scheme " +
str(number) + ".png")
plt.show()
# Write csv file of result matrix inside result folder
title = directory + 'weight scheme ' + str(number)
with open(title + ' result.csv', "w") as output:
writer = csv.writer(output, lineterminator='\n')
for val in resultMatrix:
writer.writerow(val)
# Find correct date to match dates for market index and our portfolio
correctdate = []
correctdate2 = []
for i in range(len(timestamp)):
for j in range(len(date_index)):
if date_index[j] == timestamp[i]:
correctdate.append(i)
correctdate2.append(j)
# Prepare for a graph
x_axis = date_index
y = [] # Strategy Portfolio
y_index = [] # Market Index Portfolio
# Normalize to initial amount and fill up values for y-axis by searching for correct dates
relative_portfolio = functions.relative(portfolio_list)
for i in range(len(correctdate)):
correctindex = correctdate[i] - 1
y.append(relative_portfolio[correctindex])
for j in range(len(correctdate2)):
correctindex = correctdate2[j]
y_index.append(price_index[correctindex])
y_index = functions.relative(y_index)
# Plot a graph
fig = plt.figure() # Create a figure
ax = plt.axes() # Create axis
if len(x_axis) == len(y):
plt.plot(x_axis, y, label='Strategy')
plt.plot(x_axis, y_index, label=market)
ax.xaxis.set_major_locator(