股票 · 美国 · 股票 · 每日 · 美国 Free

股票市场中的盈利公告季节性效应

登录后收藏

学术论文

Being Surprised by the Unsurprising: Earnings Seasonality and Stock Returns

作者Tom Chang; Samuel M. Hartzmark; David H. Solomon; Eugene F. Soltes

机构

University of Southern California
?University of Southern California - Marshall School of Business - Finance and Business Economics Department
University of Chicago
?University of Chicago - Booth School of Business
Boston College
?Boston College - Carroll School of Management
Harvard University
?Harvard University - Business School (HBS)

论文链接https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2460166

策略概要

投资范围包括在纽约证券交易所、美国证券交易所和纳斯达克上市的普通股，不包括价格低于5美元的股票或缺少市值或盈利数据的股票。为了预测季节性，该策略使用过去20个季度的五年盈利数据。对于20个季度中的每一个季度，公司的每股收益（经股票拆分调整后）都从最大到最小进行排名。季度t的盈利排名（earn-rank）是季度t-4、t-8、t-12、t-16和t-20的平均排名。投资者做多盈利排名最高的5%的公司，做空排名最低的5%的公司。当多头和空头都至少包含十家公司时，形成投资组合。持仓在t-1日开仓，持有至t+1日，t日为盈利公告日。投资组合等权重，每日重新平衡。该策略旨在利用盈利公告的季节性，即历史盈利排名较高的公司预计相对于排名较低的公司表现更好。

II. 策略合理性

近因效应导致投资者对近期信息的重视程度高于较早的数据。在盈利方面，投资者往往会过度强调近期较低的盈利，而忽略去年同期较高的盈利。这会导致对未来盈利公告的过度悲观预期，从而增加积极意外的可能性。虽然投资者会考虑前几个季度的盈利数据，但最近的盈利往往更为突出，使近期信息在形成他们的预期和决策时更具影响力。

回测表现

索提诺比率-0.576

胜率48%

完整 Python 代码

1from AlgorithmImports import *
2from data_tools import TradeManager
3import numpy as np
4from collections import deque
5from pandas.tseries.offsets import BDay
6from dateutil.relativedelta import relativedelta
7from typing import List, Dict
8#endregion
9class EarningsAnnouncementSeasonalityEffectinEquities(QCAlgorithm):
10def Initialize(self) -> None:
11self.SetStartDate(2010, 1, 1)
12self.SetCash(100_000)
13self.period: int = 20
14self.long_symbols: int = 10
15self.short_symbols: int = 10
16self.holding_period: int = 4
17self.leverage: int = 5
18self.quantile: int = 20
19self.prev_month: int = -1
20self.prev_month_year: int = -1
21symbol: Symbol = self.AddEquity('SPY', Resolution.Daily).Symbol
22self.long: List[Symbol] = []
23self.short: List[Symbol] = []
24   
25self.eps: Dict[Symbol, deque] = {}
26self.earnings: Dict[datetime.date, List[str]] = {} 
27self.eps_data: Dict[int, Dict[int, Dict[str, Dict[datetime.date, float]]]] = {}
28# parse earnings dataset - Source: https://www.nasdaq.com/market-activity/earnings
29earnings_data: str = self.Download('data.quantpedia.com/backtesting_data/economic/earnings_dates_eps.json')
30earnings_data_json: List[dict] = json.loads(earnings_data)
31
32for obj in earnings_data_json:
33    date: datetime.date = datetime.strptime(obj['date'], '%Y-%m-%d').date()
34    year: int = date.year
35    month: int = date.month
36    self.earnings[date] = []
37    
38    for stock_data in obj['stocks']:
39        ticker: str = stock_data['ticker']
40        self.earnings[date].append(ticker)
41        if stock_data['eps'] == '':
42            continue
43        if year not in self.eps_data:
44            self.eps_data[year] = {}
45        if month not in self.eps_data[year]:
46            self.eps_data[year][month] = {}
47        if ticker not in self.eps_data[year][month]:
48            self.eps_data[year][month][ticker] = {}
49        self.eps_data[year][month][ticker][date] = float(stock_data['eps'])
50# equally weighted brackets for traded symbols
51# hodling period 3 days + 1 day due to QC midnight offset (00:00 trading)
52self.trade_manager: TradeManager = TradeManager(self, self.long_symbols, self.short_symbols, self.holding_period)
53
54self.selection_flag: bool = False
55self.settings.daily_precise_end_time = False
56self.Settings.MinimumOrderMarginPortfolioPercentage = 0.
57self.UniverseSettings.Resolution = Resolution.Daily
58self.AddUniverse(self.FundamentalSelectionFunction)
59self.Schedule.On(self.DateRules.MonthStart(symbol), self.TimeRules.AfterMarketOpen(symbol), self.Selection)
60def OnSecuritiesChanged(self, changes: SecurityChanges) -> None:
61for security in changes.AddedSecurities:
62    security.SetFeeModel(CustomFeeModel())
63    security.SetLeverage(self.leverage)
64
65def FundamentalSelectionFunction(self, fundamental: List[Fundamental]) -> List[Symbol]:
66if not self.selection_flag: 
67    return Universe.Unchanged
68self.selection_flag = False
69prev_month_date: datetime.date = (self.Time - relativedelta(months=1)).date()
70self.prev_month_year: int = prev_month_date.year
71self.prev_month: int = prev_month_date.month
72if self.prev_month_year not in self.eps_data or self.prev_month not in self.eps_data[self.prev_month_year]:
73    return Universe.Unchanged
74
75# select stocks, which has earning in previous month
76stocks_with_prev_month_eps: Dict[str, Dict[datetime.date, float]] = self.eps_data[self.prev_month_year][self.prev_month]
77selected: List[Fundamental] = [x for x in fundamental if x.Symbol.Value in stocks_with_prev_month_eps]
78for stock in selected: 
79    symbol: Symbol = stock.Symbol
80    ticker: str = symbol.Value
81    
82    # store eps data
83    if symbol not in self.eps:
84        self.eps[symbol] = deque(maxlen=self.period)
85    
86    # get all stock's eps from previous month
87    stock_prev_month_eps: Dict[datetime.date, float] = self.eps_data[self.prev_month_year][self.prev_month][ticker]
88    # get the date of the latest eps in previous month
89    stock_latest_eps_date:datetime.date = list(stock_prev_month_eps.keys())[-1]
90    # store stock's latest eps date in previous month and it's eps value
91    data_eps: List[datetime.date, float] = [stock_latest_eps_date, stock_prev_month_eps[stock_latest_eps_date]]
92    self.eps[symbol].append(data_eps)
93
94# Earn rank calc
95earn_rank: Dict[Symbol, float] = {}
96for symbol in self.eps:
97    if len(self.eps[symbol]) != self.eps[symbol].maxlen:
98        continue
99    relevant_earnings_dates:List[datetime.date] = [self.eps[symbol][-4][0], self.eps[symbol][-8][0], \
100        self.eps[symbol][-12][0] ,self.eps[symbol][-16][0], self.eps[symbol][-20][0]]
101        
102    # stock with potencial upcoming month earnings
103    eps_months: List[int] = [x.month for x in relevant_earnings_dates]
104    if self.Time.month in eps_months:
105        # rank the 20 quarters of earnings data from largest to smallest
106        ranked_earnings: List[Tuple[datetime.date, float]] = [i for i in sorted(list(self.eps[symbol]), key=lambda x: x[1], reverse=True)]
107        
108        ranks: List[int] = [ranked_earnings.index(earnings_data)+1 for earnings_data in ranked_earnings if earnings_data[0] in relevant_earnings_dates]
109        
110        earn_rank[symbol] = np.mean(ranks)
111            
112if len(earn_rank) < self.quantile:
113    return Universe.Unchanged
114sorted_by_earn_rank: List[Tuple[Symbol, float]] = sorted(earn_rank.items(), key=lambda x: x[1], reverse=True)
115quantile: int = int(len(sorted_by_earn_rank) / self.quantile)
116
117# symbols to trade this month
118self.long = [x[0] for x in sorted_by_earn_rank[:quantile]]
119self.short = [x[0] for x in sorted_by_earn_rank[-quantile:]]
120    
121return self.long + self.short
122def OnData(self, data: Slice) -> None:
123earnings_date: datetime.date = (self.Time + BDay(1)).date()
124
125if earnings_date in self.earnings:
126    for symbol in self.long + self.short:
127        if symbol not in data or not data[symbol]:
128            continue
129        ticker:str = symbol.Value
130        # symbol has earnings in 1 day
131        if ticker not in self.earnings[earnings_date]:
132            continue
133        if symbol in self.long:
134            self.trade_manager.Add(symbol, True)
135        else:
136            self.trade_manager.Add(symbol, False)
137
138self.trade_manager.TryLiquidate()
139
140def Selection(self) -> None:
141self.long.clear()
142self.short.clear()
143self.selection_flag = True
144# Custom fee model
145class CustomFeeModel(FeeModel):
146def GetOrderFee(self, parameters: OrderFeeParameters) -> OrderFee:
147fee: float = parameters.Security.Price * parameters.Order.AbsoluteQuantity * 0.00005
148return OrderFee(CashAmount(fee, "USD"))