Classical Equity Anomalies Combined with Trendfollowing Filter

1import numpy as np
2from AlgorithmImports import *
3from functools import reduce
4from numpy import isnan
5from pandas.core.frame import DataFrame
6class ClassicalEquityAnomaliesCombinedwithTrendfollowingFilter(QCAlgorithm):
7def Initialize(self):
8self.SetStartDate(2000, 1, 1)
9self.SetCash(100000)
10market:Symbol = self.AddEquity('SPY', Resolution.Daily).Symbol
11self.financial_statement_names:List[str] = [
12    'ValuationRatios.PBRatio',
13    'FinancialStatements.IncomeStatement.GrossProfit.TwelveMonths',
14    'FinancialStatements.IncomeStatement.TotalOperatingIncomeAsReported.TwelveMonths',
15    'OperationRatios.TotalAssetsGrowth.OneYear',
16    'FinancialStatements.CashFlowStatement.CapitalExpenditure.TwelveMonths',
17    'EarningReports.BasicAverageShares.TwelveMonths',
18    # accruals data
19    'FinancialStatements.BalanceSheet.CurrentAssets.TwelveMonths',
20    'FinancialStatements.BalanceSheet.CashAndCashEquivalents.TwelveMonths',
21    'FinancialStatements.BalanceSheet.CurrentLiabilities.TwelveMonths',
22    'FinancialStatements.BalanceSheet.CurrentDebt.TwelveMonths',
23    'FinancialStatements.BalanceSheet.IncomeTaxPayable.TwelveMonths',
24    'FinancialStatements.IncomeStatement.DepreciationAndAmortization.TwelveMonths',
25    'FinancialStatements.BalanceSheet.TotalAssets.TwelveMonths'
26]
27# Daily sma data.
28self.data:Dict[Symbol, SymbolData] = {}
29self.short_period:int = 50
30self.long_period:int = 200
31self.performance_period:int = 12*21
32self.quantile:int = 10
33self.leverage:int = 5
34self.long:List[Symbol] = []
35self.short:List[Symbol] = []
36self.fundamental_count:int = 1000
37self.fundamental_sorting_key = lambda x: x.DollarVolume
38
39self.selection_flag:bool = False
40self.UniverseSettings.Resolution = Resolution.Daily
41self.AddUniverse(self.FundamentalSelectionFunction)
42self.Settings.MinimumOrderMarginPortfolioPercentage = 0.
43self.Schedule.On(self.DateRules.MonthEnd(market), self.TimeRules.BeforeMarketClose(market), self.Selection)
44self.settings.daily_precise_end_time = False
45def OnSecuritiesChanged(self, changes: SecurityChanges) -> None:
46for security in changes.AddedSecurities:
47    security.SetFeeModel(CustomFeeModel())
48    security.SetLeverage(self.leverage)
49    
50def FundamentalSelectionFunction(self, fundamental: List[Fundamental]) -> List[Symbol]:
51# Update SMA every day.
52for stock in fundamental:
53    symbol:Symbol = stock.Symbol
54    # Store monthly price.
55    if symbol in self.data:
56        self.data[symbol].update(self.Time, stock.AdjustedPrice)
57if not self.selection_flag:
58    return Universe.Unchanged
59selected:List[Fundamental] = [
60    x for x in fundamental if x.HasFundamentalData and x.Market == 'usa' and \
61    all((not isnan(self.rgetattr(x, statement_name)) and self.rgetattr(x, statement_name) != 0) for statement_name in self.financial_statement_names)
62]
63if len(selected) > self.fundamental_count:
64    selected = [x for x in sorted(selected, key=self.fundamental_sorting_key, reverse=True)[:self.fundamental_count]]
65# Ranked stocks.
66rank = {}
67current_accruals_data = {}
68acc = {}
69momentum = {}
70net_stock_issue = {}
71capex = {}
72noa = {}
73selected_stocks:List[Fundamental] = []
74# Warmup price rolling windows.
75for stock in selected:
76    symbol:Symbol = stock.Symbol
77    if symbol not in self.data:
78        self.data[symbol] = SymbolData(self.short_period, self.long_period, self.performance_period)
79        history:DataFrame = self.History(symbol, self.performance_period, Resolution.Daily)
80        if history.empty:
81            self.Log(f"Not enough data for {symbol} yet.")
82            continue
83        
84        closes:pd.Series = history.loc[symbol].close
85        for time, close in closes.items():
86            self.data[symbol].update(time, close)
87    
88    if self.data[symbol].is_ready():
89        selected_stocks.append(stock)
90        rank[symbol] = 0
91        momentum[symbol] = self.data[symbol].performance()
92    
93        # Accural calc
94        current_accruals_data = AccrualsData(stock.FinancialStatements.BalanceSheet.CurrentAssets.TwelveMonths, stock.FinancialStatements.BalanceSheet.CashAndCashEquivalents.TwelveMonths,
95                                                    stock.FinancialStatements.BalanceSheet.CurrentLiabilities.TwelveMonths, stock.FinancialStatements.BalanceSheet.CurrentDebt.TwelveMonths, stock.FinancialStatements.BalanceSheet.IncomeTaxPayable.TwelveMonths,
96                                                    stock.FinancialStatements.IncomeStatement.DepreciationAndAmortization.TwelveMonths, stock.FinancialStatements.BalanceSheet.TotalAssets.TwelveMonths)
97        if self.data[symbol].accruals_data:
98            acc[symbol] = self.CalculateAccruals(current_accruals_data, self.data[symbol].accruals_data)
99    
100            # Calculate NOA.
101            noa[symbol] = (stock.FinancialStatements.BalanceSheet.CurrentAssets.TwelveMonths - stock.FinancialStatements.BalanceSheet.CurrentLiabilities.TwelveMonths) / self.data[symbol].accruals_data.TotalAssets
102        
103        # Calculate stock stock issue.
104        num_of_shares = stock.EarningReports.BasicAverageShares.TwelveMonths
105        if self.data[symbol].shares_outstanding:
106            net_stock_issue[symbol] = num_of_shares / self.data[symbol].shares_outstanding - 1
107        self.data[symbol].shares_outstanding = num_of_shares
108        
109        # Calculate stock capital expenditure.
110        cap_expenditures = stock.FinancialStatements.CashFlowStatement.CapitalExpenditure.TwelveMonths
111        if self.data[symbol].capex:
112            capex[symbol] = cap_expenditures / self.data[symbol].capex - 1
113        self.data[symbol].capex = cap_expenditures
114        
115        # Update accruals data for next year's calculation.
116        self.data[symbol].accruals_data = current_accruals_data            
117    
118# Ensure that consecutive accruals and shares data are available.
119for symbol in self.data:
120    if symbol not in list(map(lambda x: x.Symbol, selected_stocks)):
121        self.data[symbol].accruals_data = None
122        self.data[symbol].shares_outstanding = None
123        self.data[symbol].capex = None
124
125# Sort by book to market.
126sorted_by_bm = sorted(selected_stocks, key=lambda x: 1 / x.ValuationRatios.PBRatio)
127for index, stock in enumerate(sorted_by_bm):
128    rank[stock.Symbol] += (index+1)
129
130# Sort by gross profit.
131sorted_by_gp = sorted(selected_stocks, key=lambda x: x.FinancialStatements.IncomeStatement.GrossProfit.TwelveMonths / x.FinancialStatements.BalanceSheet.TotalAssets.TwelveMonths)
132for index, stock in enumerate(sorted_by_gp):
133    rank[stock.Symbol] += (index+1)
134# Sort by operating profit.
135sorted_by_op = sorted(selected_stocks, key=lambda x: x.FinancialStatements.IncomeStatement.TotalOperatingIncomeAsReported.TwelveMonths)
136for index, stock in enumerate(sorted_by_op):
137    rank[stock.Symbol] += (index+1)
138# Sort by asset growth.
139sorted_by_ag = sorted(selected_stocks, key=lambda x: x.OperationRatios.TotalAssetsGrowth.OneYear, reverse = True)
140for index, stock in enumerate(sorted_by_ag):
141    rank[stock.Symbol] += (index+1)
142# Sort by investments.
143sorted_by_inv = sorted(capex.items(), key=lambda x: x[1], reverse = True)
144for index, symbol_si in enumerate(sorted_by_inv):
145    rank[symbol_si[0]] += (index+1)
146    
147# Sort by net stock issue.
148sorted_by_si = sorted(net_stock_issue.items(), key=lambda x: x[1], reverse = True)
149for index, symbol_si in enumerate(sorted_by_si):
150    rank[symbol_si[0]] += (index+1)
151# Sort by accruals.
152sorted_by_acc = sorted(acc.items(), key=lambda x: x[1])
153for index, symbol_acc in enumerate(sorted_by_acc):
154    rank[symbol_acc[0]] += (index+1)
155# Sort by NOA.
156sorted_by_noa = sorted(noa.items(), key=lambda x: x[1], reverse = True)
157for index, symbol_noa in enumerate(sorted_by_noa):
158    rank[symbol_noa[0]] += (index+1)
159
160if len(rank) > self.quantile:
161    # Rank sorting.
162    sorted_by_rank:List = sorted(rank.items(), key=lambda x: x[1], reverse = True)
163    quantile:int = int(len(sorted_by_rank) / self.quantile)
164    top_decile:List[Symbol] = [x[0] for x in sorted_by_rank[:quantile]]
165    bottom_decile:List[Symbol] = [x[0] for x in sorted_by_rank[-quantile:]]
166    
167    top_decile_perf:float = np.mean([momentum[x] for x in top_decile])
168    bottom_decile_perf:float = np.mean([momentum[x] for x in bottom_decile])
169    
170    if top_decile_perf > bottom_decile_perf:
171        self.long = top_decile
172        self.short = bottom_decile
173    else:
174        self.long = bottom_decile
175        self.short = top_decile
176
177return self.long + self.short
178def OnData(self, data: Slice) -> None:
179if not self.selection_flag:
180    return
181self.selection_flag = False
182# order execution
183targets:List[PortfolioTarget] = []
184for i, portfolio in enumerate([self.long, self.short]):
185    for symbol in portfolio:
186        if symbol in data and data[symbol]:
187            targets.append(PortfolioTarget(symbol, ((-1) ** i) / len(portfolio)))
188
189self.SetHoldings(targets, True)
190self.long.clear()
191self.short.clear()
192
193def Selection(self) -> None:
194# Drop stocks according to trend filter.
195symbols_invested:List[Symbol] = [x.Key for x in self.Portfolio if x.Value.Invested]
196
197for symbol in symbols_invested:
198    if self.Portfolio[symbol].IsLong:
199        if not self.data[symbol].is_in_uptrend():
200            self.Liquidate(symbol)
201    
202    elif self.Portfolio[symbol].IsShort:
203        if self.data[symbol].is_in_uptrend():
204            self.Liquidate(symbol)
205
206if self.Time.month == 6:
207    self.selection_flag = True
208
209def CalculateAccruals(self, current_accrual_data, prev_accrual_data) -> float:
210delta_assets = current_accrual_data.CurrentAssets - prev_accrual_data.CurrentAssets
211delta_cash = current_accrual_data.CashAndCashEquivalents - prev_accrual_data.CashAndCashEquivalents
212delta_liabilities = current_accrual_data.CurrentLiabilities - prev_accrual_data.CurrentLiabilities
213delta_debt = current_accrual_data.CurrentDebt - prev_accrual_data.CurrentDebt
214delta_tax = current_accrual_data.IncomeTaxPayable - prev_accrual_data.IncomeTaxPayable
215dep = current_accrual_data.DepreciationAndAmortization
216avg_total = (current_accrual_data.TotalAssets + prev_accrual_data.TotalAssets) / 2
217
218bs_acc = ((delta_assets - delta_cash) - (delta_liabilities - delta_debt-delta_tax) - dep) / avg_total
219return bs_acc
220# https://gist.github.com/wonderbeyond/d293e7a2af1de4873f2d757edd580288
221def rgetattr(self, obj, attr, *args):
222def _getattr(obj, attr):
223    return getattr(obj, attr, *args)
224return reduce(_getattr, [obj] + attr.split('.'))
225
226class AccrualsData():
227def __init__(self, current_assets, cash_and_cash_equivalents, current_liabilities, current_debt, income_tax_payable, depreciation_and_amortization, total_assets):
228self.CurrentAssets = current_assets
229self.CashAndCashEquivalents = cash_and_cash_equivalents
230self.CurrentLiabilities = current_liabilities
231self.CurrentDebt = current_debt
232self.IncomeTaxPayable = income_tax_payable
233self.DepreciationAndAmortization = depreciation_and_amortization
234self.TotalAssets = total_assets
235
236class SymbolData():
237def __init__(self, short_sma_period, long_sma_period, performance_period):
238self.price = RollingWindow[float](performance_period)
239self.short_sma = SimpleMovingAverage(short_sma_period)
240self.long_sma = SimpleMovingAverage(long_sma_period)
241
242self.accruals_data = None
243self.shares_outstanding = None
244self.capex = None
245
246def update(self, time: datetime, value: float):
247self.short_sma.Update(time, value)
248self.long_sma.Update(time, value)
249self.price.Add(value)
250
251def is_ready(self) -> bool:
252return (self.short_sma.IsReady and self.long_sma.IsReady and self.price.IsReady)
253
254def is_in_uptrend(self) -> bool:
255return (self.short_sma.Current.Value > self.long_sma.Current.Value)
256
257def performance(self) -> float:
258return self.price[0] / self.price[self.price.Count - 1] - 1
259
260# Custom fee model.
261class CustomFeeModel(FeeModel):
262def GetOrderFee(self, parameters):
263fee = parameters.Security.Price * parameters.Order.AbsoluteQuantity * 0.00005
264return OrderFee(CashAmount(fee, "USD"))