def writejson(back_data, budget, baseline_data, base_index):
daily_return = back_data['DAILY_PNL'] / budget
total_return = back_data['TOTAL_PNL'] / budget
stats = metrics(daily_return, total_return, baseline_data, base_index)
# multiply by 100 for readability purposes
daily_return_percent = daily_return * 100
total_return_percent = total_return * 100
d = {'dates': back_data['DAILY_PNL'].index.format(),
'daily_pnl': daily_return_percent.sum(axis=1).values.tolist(),
'total_pnl': total_return_percent.sum(axis=1).values.tolist(),
'stocks': back_data['DAILY_PNL'].columns.tolist(),
'stock_pnl': daily_return_percent.values.tolist(),
'stock_position': back_data['POSITION'].values.tolist(),
'metrics': stats.keys(),
'metrics_values': stats.values()}
return d
def loadgui(back_data, exchange, base_index, budget, logger):
######################
# Setup data
######################
position = back_data['POSITION']
close = back_data['CLOSE']
# position as % of total portfolio
long_position = (position * close).div(back_data['VALUE'], axis=0)
short_position = long_position.copy()
long_position[long_position < 0] = 0
short_position[short_position > 0] = 0
daily_pnl = back_data['DAILY_PNL'] / budget
total_pnl = back_data['TOTAL_PNL'] / budget
if base_index:
baseline_data = baseline(exchange, base_index, total_pnl.index, logger)
stats = metrics(daily_pnl, total_pnl, baseline_data, base_index)
else:
baseline_data = {}
stats = metrics(daily_pnl, total_pnl, {}, base_index)
daily_return = daily_pnl.sum(axis=1)
total_return = total_pnl.sum(axis=1)
long_exposure = long_position.sum(axis=1)
short_exposure = short_position.sum(axis=1)
zero_line = np.zeros(daily_pnl.index.size)
# print to logger
for x in stats.keys():
logger.info('%s : %0.2f' % (x, stats[x]))
def isDate(val):
# Function to validate if a given entry is valid date
try:
d = pd.to_datetime(val)
if d > daily_pnl.index[0] and d < daily_pnl.index[-1]:
return True
else:
return False
except ValueError:
raise ValueError("Not a Valid Date")
return False
def newselection(event):
# Function to autoupdate chart on new selection from dropdown
i = dropdown.current()
market = ['TOTAL PORTFOLIO'] + daily_pnl.columns.values.tolist()
plot(daily_pnl, total_pnl, long_position, short_position,
baseline_data, base_index, market[i], box_value2.get(),
box_value3.get())
def plot(daily_pnl,
total_pnl,
long_position,
short_position,
baseline_data,
base_index,
market='TOTAL PORTFOLIO',
start=daily_pnl.index.format()[0],
end=daily_pnl.index.format()[-1]):
# New plot when custom fields are changed
plt.clf()
# plt.style.use("seaborn-whitegrid")
daily_pnl = daily_pnl.loc[start:end]
total_pnl = total_pnl.loc[start:end]
long_position = long_position.loc[start:end]
short_position = short_position.loc[start:end]
if market == 'TOTAL PORTFOLIO':
daily_return = daily_pnl.sum(axis=1)
total_return = total_pnl.sum(axis=1)
long_exposure = long_position.sum(axis=1)
short_exposure = short_position.sum(axis=1)
else:
daily_return = daily_pnl[market]
total_return = total_pnl[market]
long_exposure = long_position[market]
short_exposure = short_position[market]
zero_line = np.zeros(daily_pnl.index.size)
# f, plot_arr = plt.subplots(3, sharex=True)
total_plot = plt.subplot2grid((10, 8), (0, 0), colspan=12, rowspan=4)
daily_plot = plt.subplot2grid((10, 8), (5, 0),
colspan=12,
rowspan=2,
sharex=total_plot)
position_plot = plt.subplot2grid((10, 8), (8, 0),
colspan=12,
rowspan=2,
sharex=total_plot)
ind = np.arange(len(daily_pnl.index))
total_plot.set_title('Total PnL')
total_plot.plot(ind, zero_line, 'k')
total_plot.plot(ind,
total_return.values,
'b',
linewidth=0.5,
label='strategy')
total_plot.legend(loc='upper left')
total_plot.autoscale(tight=True)
plt.setp(total_plot.get_xticklabels(), visible=False)
total_plot.yaxis.set_major_formatter(mtick.FuncFormatter(format_perc))
total_plot.set_ylabel('Cumulative Performance')
total_plot.legend(bbox_to_anchor=(0.03, 0.97),
loc='lower left',
borderaxespad=0.)
if base_index:
total_plot.plot(ind,
baseline_data['TOTAL_PNL'],
'g',
linewidth=0.5,
label=base_index)
daily_plot.set_title('Daily PnL')
daily_plot.plot(ind, zero_line, 'k')
daily_plot.bar(ind,
daily_return.values,
0.2,
align='center',
color='c',
label='strategy')
daily_plot.legend(loc='upper left')
daily_plot.autoscale(tight=True)
plt.setp(daily_plot.get_xticklabels(), visible=False)
daily_plot.yaxis.set_major_formatter(mtick.FuncFormatter(format_perc))
daily_plot.set_ylabel('Daily Performance')
daily_plot.legend(bbox_to_anchor=(0.03, 0.97),
loc='lower left',
borderaxespad=0.)
position_plot.set_title('Daily Exposure')
position_plot.plot(ind, zero_line, 'k')
position_plot.bar(ind,
short_exposure.values,
0.3,
linewidth=0,
align='center',
color='r',
label='short')
position_plot.bar(ind,
long_exposure.values,
0.3,
linewidth=0,
align='center',
color='b',
label='long')
position_plot.legend(loc='upper left')
position_plot.autoscale(tight=True)
position_plot.xaxis.set_major_formatter(
mtick.FuncFormatter(format_date))
position_plot.yaxis.set_major_formatter(
mtick.FuncFormatter(format_perc))
position_plot.set_ylabel('Long/Short %')
position_plot.legend(bbox_to_anchor=(0.03, 0.97),
loc='lower left',
borderaxespad=0.)
plt.gcf().canvas.draw()
def update_plot():
# Callback Function for plot button
try:
d1 = pd.to_datetime(box_value2.get())
d2 = pd.to_datetime(box_value3.get())
if d1 >= daily_pnl.index[0] and d2 <= daily_pnl.index[-1]:
plot(daily_pnl, total_pnl, long_position, short_position,
baseline_data, base_index, box_value.get(),
box_value2.get(), box_value3.get())
else:
tkMessageBox.showinfo(
"Date out of Range", "Please enter a date from %s to %s" %
(daily_pnl.index[0].strftime('%Y-%m-%d'),
daily_pnl.index[-1].strftime('%Y-%m-%d')))
except ValueError:
raise ValueError("Not a Valid Date")
def close_window():
# Callback function for Quit Button
GUI.destroy()
GUI.quit()
def format_date(x, pos=None):
# Format axis ticklabels to dates
thisind = np.clip(int(x + 0.5), 0, len(daily_pnl.index) - 1)
return daily_pnl.index[thisind].strftime('%b-%y')
def format_perc(y, pos=None):
# Format axis ticklabels to %
if budget > 1:
return '{percent:.2%}'.format(percent=y)
else:
return y
def onFrameConfigure(canvas):
canvas.configure(scrollregion=canvas.bbox("all"))
######################
# GUI mainloop
######################
# Create widget
GUI = tk.Tk()
GUI.title('Backtest Results')
winCanvas = tk.Canvas(GUI,
borderwidth=0,
background="#ffffff",
width=1500,
height=1000)
frame = tk.Frame(winCanvas, background="#ffffff")
vsb = tk.Scrollbar(GUI, orient="vertical", command=winCanvas.yview)
hsb = tk.Scrollbar(GUI, orient="horizontal", command=winCanvas.xview)
winCanvas.configure(yscrollcommand=vsb.set)
winCanvas.configure(xscrollcommand=hsb.set)
vsb.pack(side="left", fill="y")
hsb.pack(side="bottom", fill="x")
winCanvas.pack(side="right", fill="both", expand=True)
winCanvas.create_window((50, 50), window=frame, anchor="nw")
frame.bind("<Configure>",
lambda event, canvas=winCanvas: onFrameConfigure(winCanvas))
# Create dropdown for market
Label_1 = tk.Label(frame, text="Trading Performance:")
Label_1.grid(row=0, column=0, sticky=tk.EW)
box_value = tk.StringVar()
dropdown = ttk.Combobox(frame, textvariable=box_value, state='readonly')
dropdown['values'] = ['TOTAL PORTFOLIO'
] + daily_pnl.columns.values.tolist()
dropdown.grid(row=0, column=1, sticky=tk.EW)
dropdown.current(0)
dropdown.bind('<<ComboboxSelected>>', newselection)
# Create entry field for start date
Label_2 = tk.Label(frame, text="Start Date")
Label_2.grid(row=0, column=2, sticky=tk.EW)
box_value2 = tk.StringVar(frame, value=daily_pnl.index.format()[0])
start = tk.Entry(frame,
textvariable=box_value2,
validate='key',
validatecommand=(GUI.register(isDate), '%P'))
start.grid(row=0, column=3, sticky=tk.EW)
# Create entry field for end date
Label_3 = tk.Label(frame, text="End Date")
Label_3.grid(row=0, column=4, sticky=tk.EW)
box_value3 = tk.StringVar(frame, value=daily_pnl.index.format()[-1])
end = tk.Entry(frame,
textvariable=box_value3,
validate='key',
validatecommand=(GUI.register(isDate), '%P'))
end.grid(row=0, column=5, sticky=tk.EW)
# Create Plot button to reload chart
button1 = tk.Button(frame, text='PLOT', command=update_plot)
button1.grid(row=0, column=6, sticky=tk.EW)
# Create text widget with backtest results
customFont1 = tkFont.Font(family="Helvetica", size=9, weight="bold")
customFont2 = tkFont.Font(family="Helvetica", size=12)
text = tk.Text(frame,
height=3,
width=50,
wrap=tk.WORD,
bd=5,
padx=10,
pady=5)
text.grid(row=1, column=0, columnspan=7, sticky=tk.EW)
String1 = ''
String2 = ''
for y in stats.keys():
String1 = String1 + y + '\t\t'
x = stats[y]
if budget > 1 and 'Ratio' not in y:
String2 = String2 + '{percent:.2%}'.format(percent=x) + '\t\t'
else:
String2 = String2 + '%0.2f' % x + '\t\t'
text.insert(tk.END, String1)
text.insert(tk.END, '\n')
text.insert(tk.END, String2)
text.tag_add("keys", "1.0", "1.end")
text.tag_config("keys", font=customFont1)
text.tag_add("values", "2.0", "2.end")
text.tag_config("values", foreground="red", font=customFont2)
# Create canvas to plot chart
f = plt.figure(figsize=(16, 8))
canvas = FigureCanvasTkAgg(f, master=frame)
canvas.get_tk_widget().grid(row=2,
column=0,
columnspan=7,
rowspan=1,
sticky=tk.NSEW)
toolbar_frame = tk.Frame(frame)
toolbar_frame.grid(row=4, column=0, columnspan=7)
# plot 3 subplots for total position, daily position and exposure
plt.style.use("seaborn-whitegrid")
total_plot = plt.subplot2grid((10, 8), (0, 0), colspan=12, rowspan=4)
daily_plot = plt.subplot2grid((10, 8), (5, 0),
colspan=12,
rowspan=2,
sharex=total_plot)
position_plot = plt.subplot2grid((10, 8), (8, 0),
colspan=12,
rowspan=2,
sharex=total_plot)
ind = np.arange(len(daily_pnl.index))
total_plot.set_title('Total PnL')
total_plot.plot(ind, zero_line, 'k')
total_plot.plot(ind,
total_return.values,
'b',
linewidth=0.5,
label='strategy')
total_plot.legend(loc='upper left')
total_plot.autoscale(tight=True)
plt.setp(total_plot.get_xticklabels(), visible=False)
total_plot.yaxis.set_major_formatter(mtick.FuncFormatter(format_perc))
total_plot.set_ylabel('Cumulative Performance')
total_plot.legend(bbox_to_anchor=(0.03, 0.97),
loc='lower left',
borderaxespad=0.)
if base_index:
total_plot.plot(ind,
baseline_data['TOTAL_PNL'],
'g',
linewidth=0.5,
label=base_index)
daily_plot.set_title('Daily PnL')
daily_plot.plot(ind, zero_line, 'k')
daily_plot.bar(ind,
daily_return.values,
0.2,
align='center',
color='c',
label='strategy')
daily_plot.legend(loc='upper left')
daily_plot.autoscale(tight=True)
plt.setp(daily_plot.get_xticklabels(), visible=False)
daily_plot.yaxis.set_major_formatter(mtick.FuncFormatter(format_perc))
daily_plot.set_ylabel('Daily Performance')
daily_plot.legend(bbox_to_anchor=(0.03, 0.97),
loc='lower left',
borderaxespad=0.)
position_plot.set_title('Daily Exposure')
position_plot.plot(ind, zero_line, 'k')
position_plot.bar(ind,
short_exposure.values,
0.3,
linewidth=0,
align='center',
color='r',
label='short')
position_plot.bar(ind,
long_exposure.values,
0.3,
linewidth=0,
align='center',
color='b',
label='long')
position_plot.legend(loc='upper left')
position_plot.autoscale(tight=True)
position_plot.xaxis.set_major_formatter(mtick.FuncFormatter(format_date))
position_plot.yaxis.set_major_formatter(mtick.FuncFormatter(format_perc))
position_plot.set_ylabel('Long/Short')
position_plot.legend(bbox_to_anchor=(0.03, 0.97),
loc='lower left',
borderaxespad=0.)
plt.gcf().canvas.draw()
# Create Quit Button
button2 = tk.Button(frame, text='QUIT', command=close_window)
button2.grid(row=4, column=6, sticky=tk.EW)
GUI.mainloop()