股票 · 欧洲 · 期货 · 每日 · 欧洲 Free

股息风险溢价策略

学术论文

作者提取股息风险溢价 [点击查看论文]

论文链接https://quantpedia.com/www/Credit_Suisse_Dividend_Risk_Premium.pdf

策略概要

该策略旨在通过模拟假设的1年期股息期货来保持对隐含股息的恒定敞口。这涉及每日在近月和远月股息期货之间滚动头寸，逐步将近月期货的权重降低至零，同时将远月期货的权重增加至100%。为了捕捉股息风险溢价，投资者做多“1年期恒定到期日”股息期货，并通过做空SX5E指数来对冲股票风险敞口。每日对冲比率通过过去两周股息期货回报与SX5E回报的线性回归确定，确保有效的风险管理。

II. 策略合理性

股息风险溢价的存在归因于几个因素。投资者高估了单只股票的股息削减风险，尽管年股息削减超过60%的情况很少见（自1991年以来为4%）。他们还低估了股息指数的多元化收益。流动性风险也是一个因素，因为股息期货的流动性较差。此外，系统性错误定价是由银行交易台持有大量SX5E股息多头头寸引起的。这些因素，包括对低概率、高影响力事件的风险厌恶和流动性挑战，在股息期货市场中创造了持续的溢价。

回测表现

波动率11.6%

夏普比率1.11

最大回撤-28.8%

胜率55%

完整 Python 代码

1from AlgorithmImports import *
2from scipy import stats
3import numpy as np
4#endregion
5class DividendRiskPremiumStrategy(QCAlgorithm):
6def Initialize(self):
7self.SetStartDate(2010, 1, 1)
8self.SetCash(100000)
9
10self.dividend_futures = ['FEXD1', 'FEXD2', 'FEXD3']
11
12self.fexd1 = None
13for i, div_symbol in enumerate(self.dividend_futures):
14    data = self.AddData(QuantpediaFutures, div_symbol, Resolution.Daily)
15    data.SetLeverage(10)
16    data.SetFeeModel(CustomFeeModel())
17    
18    # store nearby contract
19    if i == 0:
20        self.fexd1 = data.Symbol
21
22# STOXX 50 future
23data = self.AddData(QuantpediaFutures, "EUREX_FSTX1", Resolution.Daily)
24data.SetLeverage(10)
25data.SetFeeModel(CustomFeeModel())
26self.SX5E = data.Symbol
27
28# daily price
29self.data = {}
30self.period = 10
31self.max_missing_days = 5
32for symbol in [self.fexd1, self.SX5E]:
33    self.data[symbol] = RollingWindow[float](self.period)
34
35self.selection_flag = False
36self.recent_month:int = -1
37
38def OnData(self, data):
39# store daily prices
40if self.fexd1 in data and data[self.fexd1] and self.SX5E in data and data[self.SX5E]:
41    self.data[self.fexd1].Add(data[self.fexd1].Value)
42    self.data[self.SX5E].Add(data[self.SX5E].Value)
43else:
44    if (self.Securities[self.fexd1].GetLastData() and (self.Time.date() - self.Securities[self.fexd1].GetLastData().Time.date()).days > self.max_missing_days) or \
45        (self.Securities[self.SX5E].GetLastData() and (self.Time.date() - self.Securities[self.SX5E].GetLastData().Time.date()).days > self.max_missing_days):
46        self.data[self.fexd1].Reset()
47        self.data[self.SX5E].Reset()
48        self.Liquidate()
49if self.recent_month == self.Time.month:
50    return
51self.recent_month = self.Time.month
52if self.Time.month in [7, 1]:
53    self.Liquidate()
54    self.weights = [1, 0]   # divident future #2 and #3 weights
55    self.selection_flag = True
56
57if self.Portfolio.Invested or self.selection_flag:
58    if self.data[self.fexd1].IsReady and self.data[self.SX5E].IsReady:
59        dividend_future_prices = np.array([x for x in self.data[self.fexd1]])
60        market_prices = np.array([x for x in self.data[self.SX5E]])
61        
62        dividend_future_returns = dividend_future_prices[:-1] / dividend_future_prices[1:] - 1
63        market_returns = market_prices[:-1] / market_prices[1:] - 1
64        if not all(x==0 for x in dividend_future_returns):
65            slope, intercept, r_value, p_value, std_err = stats.linregress(market_returns, dividend_future_returns)
66            
67            # dividend futures exposure
68            self.SetHoldings(self.dividend_futures[1], self.weights[0])
69            self.SetHoldings(self.dividend_futures[2], self.weights[1])
70            
71            # adjust weights every invested day
72            if self.weights[0] >= 0.008:
73                self.weights[0] -= 0.008
74            
75            if self.weights[0] <= 0.992:
76                self.weights[1] += 0.008
77            
78            # hedge
79            self.SetHoldings(self.SX5E, -abs(slope))
80            
81            self.selection_flag = False
82        else:
83            self.Liquidate()
84# Quantpedia data.
85# NOTE: IMPORTANT: Data order must be ascending (datewise)
86class QuantpediaFutures(PythonData):
87def GetSource(self, config, date, isLiveMode):
88return SubscriptionDataSource("data.quantpedia.com/backtesting_data/futures/{0}.csv".format(config.Symbol.Value), SubscriptionTransportMedium.RemoteFile, FileFormat.Csv)
89def Reader(self, config, line, date, isLiveMode):
90data = QuantpediaFutures()
91data.Symbol = config.Symbol
92
93if not line[0].isdigit(): return None
94split = line.split(';')
95
96data.Time = datetime.strptime(split[0], "%d.%m.%Y") + timedelta(days=1)
97data['back_adjusted'] = float(split[1])
98data['spliced'] = float(split[2])
99data.Value = float(split[1])
100return data
101# Custom fee model.
102class CustomFeeModel(FeeModel):
103def GetOrderFee(self, parameters):
104fee = parameters.Security.Price * parameters.Order.AbsoluteQuantity * 0.00005
105return OrderFee(CashAmount(fee, "USD"))