Overconfidence Factor in China

1from AlgorithmImports import *
2import statsmodels.api as sm
3import data_tools
4# endregion
5
6class OverconfidenceFactorInChina(QCAlgorithm):
7
8def Initialize(self):
9self.SetStartDate(2009, 1, 1)
10self.SetCash(100000)
11
12self.tickers:list[str] = [
13    '0001', '0002', '0004', '0006', '0011', '0012', '0016', '0017',
14    '0019', '0020', '0027', '0066', '0083', '0101', '0135', '0144', '0151',
15    '0175', '0241', '0267', '0268', '0270', '0288', '0291', '0316', '0322',
16    '0384', '0388', '0489', '0586', '0656', '0669', '0688', '0700', '0762',
17    '0788', '0836', '0868', '0881', '0883', '0916', '0960', '0968', '0981',
18    '0992', '1038', '1044', '1093', '1099',
19
20    # data error: '0003'
21
22    # '1109', '1113', '1177', '1193',
23    # '1299', '1308', '1359', '1378', '1810', '1876', '1913', '1928', '1929',
24    # '1972', '1997', '2007', '2020', '2066', '2269', '2313', '2319', '2328',
25    # '2331', '2382', '2388', '2638', '2688', '3311', '3323', '3690', '3692',
26    # '3799', '3800', '6098', '6823', '6862'
27]
28
29self.min_volumes:int = 15           # need atleast n daily volumes for one month
30self.regression_period:int = 12     # need m monthly data
31self.quantile:int = 2
32self.leverage:int = 5
33self.data:dict[Symbol, data_tools.SymbolData] = {}
34
35# NOTE: Download method is rate-limited to 100 calls (https://github.com/QuantConnect/Documentation/issues/345)
36for ticker in self.tickers:
37    security = self.AddData(data_tools.ChineseStock, ticker, Resolution.Daily)
38    security.SetFeeModel(CustomFeeModel())
39    security.SetLeverage(self.leverage)
40
41    stock_symbol:Symbol = security.Symbol
42    income_statement_symbol:Symbol = self.AddData(data_tools.IncomeStatement, ticker, Resolution.Daily).Symbol
43    self.data[stock_symbol] = data_tools.SymbolData(income_statement_symbol, self.regression_period)
44
45self.recent_month:int = -1
46
47def OnData(self, data):
48curr_date:datetime.date = self.Time.date()
49
50# daily data update
51# price and average shares are retrieved and stored separately, because they might not come at the same time
52for symbol, symbol_obj in self.data.items():
53    if symbol in data and data[symbol]:
54        if data[symbol].GetProperty('volume') != 0:
55            volume:float = data[symbol].GetProperty('volume')
56            
57            symbol_obj.update_daily_volumes(volume)
58        
59        if data[symbol].Value != 0: 
60            price:float = data[symbol].Value
61
62            symbol_obj.update_last_price(price)
63
64    income_statement_symbol:Symbol = symbol_obj.income_statement_symbol
65    if income_statement_symbol in data and data[income_statement_symbol] and \
66        data[income_statement_symbol].GetProperty('weightedAverageShsOutDil') != 0:
67
68        average_shares:float = data[income_statement_symbol].GetProperty('weightedAverageShsOutDil')
69
70        symbol_obj.update_last_average_shares(average_shares)
71
72# rebalance monthly
73if self.recent_month == curr_date.month:
74    return
75self.recent_month = curr_date.month
76
77filtered_symbols:list[Symbol] = []
78regression_x:np.array = np.zeros(self.regression_period)
79
80for symbol, symbol_obj in self.data.items():
81    if symbol_obj.are_daily_volumes_ready(self.min_volumes):
82        symbol_obj.update_monthly_volumes()
83        symbol_obj.reset_daily_volumes()
84    else:
85        # data have to be consecutive
86        symbol_obj.reset_volumes()
87
88    if symbol_obj.are_regression_data_ready() and symbol_obj.are_data_for_market_cap_ready():
89        regression_x += np.array(symbol_obj.get_monthly_volumes())
90
91        filtered_symbols.append(symbol)
92
93ATV_by_symbol:dict[Symbol, float] = {}
94market_cap_by_symbol:dict[Symbol, float] = {}
95
96for symbol in filtered_symbols:
97    market_cap:float = self.data[symbol].get_market_cap()
98    regression_y:list[float] = self.data[symbol].get_monthly_volumes()
99
100    regression_model = self.MultipleLinearRegression(regression_x, regression_y)
101
102    ATV:float = np.mean(regression_model.resid)
103    ATV_by_symbol[symbol] = ATV
104    market_cap_by_symbol[symbol] = market_cap
105
106# big and small cap will be sorted into halfs by ATV values and there has to be enough stocks
107if len(ATV_by_symbol) < (self.quantile * 2):
108    self.Liquidate()
109    return
110
111cap_quantile:int = int(len(ATV_by_symbol) / self.quantile)
112
113sorted_by_cap:list[Symbol] = [x[0] for x in sorted(market_cap_by_symbol.items(), key=lambda item: item[1])]
114large_cap:list[Symbol] = sorted_by_cap[-cap_quantile:]
115small_cap:list[Symbol] = sorted_by_cap[:cap_quantile]
116
117large_cap_sorted_by_ATV:list[Symbol] = sorted(large_cap, key=lambda symbol: ATV_by_symbol[symbol])
118small_cap_sorted_by_ATV:list[Symbol] = sorted(small_cap, key=lambda symbol: ATV_by_symbol[symbol])
119
120large_cap_quantile:int = int(len(large_cap_sorted_by_ATV) / self.quantile)
121small_cap_quantile:int = int(len(small_cap_sorted_by_ATV) / self.quantile)
122
123# long the big overconfident and small overconfident portfolios
124long_leg:list[Symbol] = large_cap_sorted_by_ATV[-large_cap_quantile:] + small_cap_sorted_by_ATV[-small_cap_quantile:]
125
126# short the big low confident and small low confident portfolios.
127short_leg:list[Symbol] = large_cap_sorted_by_ATV[:large_cap_quantile] + small_cap_sorted_by_ATV[:small_cap_quantile]
128
129# trade execution
130invested = [x.Key for x in self.Portfolio if x.Value.Invested]
131for symbol in invested:
132    if symbol not in long_leg + short_leg:
133        self.Liquidate(symbol)
134
135total_long_cap:float = sum(list(map(lambda symbol: market_cap_by_symbol[symbol], long_leg)))
136for symbol in long_leg:
137    if symbol in data and data[symbol]:
138        self.SetHoldings(symbol, market_cap_by_symbol[symbol] / total_long_cap)
139
140total_short_cap:float = sum(list(map(lambda symbol: market_cap_by_symbol[symbol], short_leg)))
141for symbol in short_leg:
142    if symbol in data and data[symbol]:
143        self.SetHoldings(symbol, -market_cap_by_symbol[symbol] / total_short_cap)
144
145def MultipleLinearRegression(self, x:np.array, y:list):
146x = np.array(x).T
147x = sm.add_constant(x)
148result = sm.OLS(endog=y, exog=x).fit()
149return result
150
151# Custom fee model
152class CustomFeeModel(FeeModel):
153def GetOrderFee(self, parameters):
154fee = parameters.Security.Price * parameters.Order.AbsoluteQuantity * 0.00005
155return OrderFee(CashAmount(fee, "USD"))