CAPE Sector Picking Strategy

1from AlgorithmImports import *
2#endregion
3class CAPESectorPickingStrategy(QCAlgorithm):
4def Initialize(self):
5self.SetStartDate(2000, 1, 1) 
6self.SetCash(100000)  
7
8self.symbols = [
9    "XLC", # Community Services
10    "XLRE", # Real Estate
11    "XLV", # Health Care
12    "XLI", # Industrial
13    "XLY", # Consumer Cyclical    
14    "XLP", # Consumer Defensive    
15    "XLB", # Basic Materials    
16    "XLK", # Technology    
17    "XLU", # Utilities    
18    "XLE", # Energy    
19    "XLF", # Financial Services  
20]
21
22# Daily prices.
23self.data = {}
24
25self.period = 21 * 12
26self.max_missing_days = 5
27self.min_symbol_count = 5
28
29for symbol in self.symbols:
30    data = self.AddEquity(symbol, Resolution.Daily)
31    data.SetFeeModel(CustomFeeModel())
32    data.SetLeverage(5)
33    
34    self.data[symbol] = SymbolData(self.period)
35    
36    self.AddData(QuantpediaSectorCAPE, symbol, Resolution.Daily)
37
38self.recent_month = -1
39
40def OnData(self, data):
41for symbol in self.symbols:
42    if symbol in data and data[symbol]:
43        self.data[symbol].update(data[symbol].Value)
44
45if self.recent_month == self.Time.month:
46    return
47self.recent_month = self.Time.month
48# CAPE data are available since 2010
49cape_values = {}
50for symbol in self.symbols:
51    cape_data = self.Securities[symbol + '.QuantpediaSectorCAPE'].GetLastData()
52    if cape_data and (self.Time.date() - cape_data.Time.date()).days <= self.max_missing_days:
53        cape_values[symbol] = cape_data['Shiller']
54    
55# performance and cape ratio tuple.
56performance_cape = { x : (self.data[x].performance(), cape_values[x]) for x in self.symbols if self.data[x].is_ready() and x in cape_values }
57
58long = []
59if len(performance_cape) >= self.min_symbol_count:
60    # sorted by cape, selecting 5 lowest
61    five_lowest = [(x[0], x[1][0]) for x in sorted(performance_cape.items(), key=lambda item: item[1][1])][:self.min_symbol_count]
62    # choosing five with highest 12 month momentum
63    long = [x[0] for x in sorted(five_lowest, key=lambda item: item[1], reverse=True)][:(self.min_symbol_count-1)]
64    
65# Trade execution
66invested = [x.Key.Value for x in self.Portfolio if x.Value.Invested]
67for symbol in invested:
68    if symbol not in long:
69        self.Liquidate(symbol)
70length = len(long)
71for symbol in long:
72    if symbol in data and data[symbol]:
73        self.SetHoldings(symbol, 1 / length)
74class SymbolData():
75def __init__(self, period):
76self.Price = RollingWindow[float](period)
77    
78def update(self, value):
79self.Price.Add(value)
80    
81def is_ready(self):
82return self.Price.IsReady
83
84def performance(self):
85prices = [x for x in self.Price]
86return prices[0] / prices[-1] - 1
87
88# Custom fee model
89class CustomFeeModel(FeeModel):
90def GetOrderFee(self, parameters):
91fee = parameters.Security.Price * parameters.Order.AbsoluteQuantity * 0.00005
92return OrderFee(CashAmount(fee, "USD"))
93
94# Quantpedia Sectors
95# NOTE: IMPORTANT: Data order must be ascending (datewise)
96class QuantpediaSectorCAPE(PythonData):
97def GetSource(self, config, date, isLiveMode):
98return SubscriptionDataSource("data.quantpedia.com/backtesting_data/economic/sector_cape/{0}.csv".format(config.Symbol.Value), SubscriptionTransportMedium.RemoteFile, FileFormat.Csv)
99def Reader(self, config, line, date, isLiveMode):
100data = QuantpediaSectorCAPE()
101data.Symbol = config.Symbol
102
103if not line[0].isdigit(): return None
104split = line.split(';')
105
106# Own formatting based on CSV file structure
107data.Time = datetime.strptime(split[0], "%Y-%m-%d") + timedelta(days=1)
108if split[1] != '' and split[2] != '':
109    data['Shiller'] = float(split[1])
110    data['Regular'] = float(split[2])
111    data.Value = float(split[1])
112else:
113    return None
114    
115return data