债券、商品、外汇、股票 · 全球 · 差价合约、期货 · 每日 · 全球 Free

期货动量策略中的月末效应

登录后收藏

学术论文

The MOM-TOM Effect: Detecting the Market Impact of CTA Trading

作者MOM-TOM效应：检测CTA交易的市场影响 [点击查看论文]

机构

NLAsser Institute
?Man AHL

论文链接http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2515900

策略概要

该策略使用52种流动性期货（货币、商品、股指、固定收益）复制一个指数，并应用时间序列动量方法。一个由波动率缩放的100天SMA生成多头/空头信号，投资组合采用波动率加权。投资者仅在为期3天的月末（ToM）期间持有该投资组合，捕捉超过50%的时间序列动量总回报。通过利用月末过渡期间可预测的回报模式，这种ToM动量策略实现了比全周期动量策略显著更高的风险调整后回报。

II. 策略合理性

动量策略在月末（ToM）期间的功能归因于买入压力。月末期间的大量资金流入促使管理者扩大现有头寸或重新平衡投资组合，从而产生暂时的价格上涨。系统性趋势跟踪/动量CTA基金放大了这种效应，尤其是在流动性较差的商品中，价格压力更为明显。该论文强调，组合动量和ToM效应导致非ToM期间的部分反转，这与暂时的价格压力一致。此外，这种异常现象不能仅用被动多头头寸的ToM效应来解释，强调了动量策略在这些期间的独特贡献。

回测表现

波动率1.5%

夏普比率3.2

索提诺比率-0.476

胜率51%

完整 Python 代码

1import numpy as np
2from AlgorithmImports import *
3from collections import deque
4from pandas.tseries.offsets import BDay
5from pandas.tseries.offsets import BMonthEnd
6class TOMEffectFuturesMomentumStrategy(QCAlgorithm):
7def Initialize(self):
8 self.SetStartDate(1991, 1, 1)
9 self.SetCash(100000)
10 self.symbols = [
11     "CME_S1",       # Soybean Futures, Continuous Contract 
12     "CME_W1",       # Wheat Futures, Continuous Contract 
13     "CME_BO1",      # Soybean Oil Futures, Continuous Contract
14     "CME_C1",       # Corn Futures, Continuous Contract
15     "CME_O1",       # Oats Futures, Continuous Contract
16     "CME_LC1",      # Live Cattle Futures, Continuous Contract
17     "CME_FC1",      # Feeder Cattle Futures, Continuous Contract
18     "CME_GC1",      # Gold Futures, Continuous Contract
19     "CME_SI1",      # Silver Futures, Continuous Contract
20     "CME_PL1",      # Platinum Futures, Continuous Contract
21     "CME_CL1",      # Crude Oil Futures, Continuous Contract
22     "CME_HG1",      # Copper Futures, Continuous Contract
23     "CME_PA1",      # Palladium Futures, Continuous Contract 
24     "CME_RR1",      # Rough Rice Futures, Continuous Contract
25     "ICE_CC1",      # Cocoa Futures, Continuous Contract 
26     "ICE_KC1",      # Coffee C Futures, Continuous Contract
27     "ICE_OJ1",      # Orange Juice Futures, Continuous Contract
28     "ICE_SB1",      # Sugar No. 11 Futures, Continuous Contract
29     "ICE_RS1",      # Canola Futures, Continuous Contract
30     "ICE_GO1",      # Gas Oil Futures, Continuous Contract
31     "CME_RB2",      # Gasoline Futures, Continuous Contract
32     "CME_KW2",      # Wheat Kansas, Continuous Contract
33     "ICE_WT1",      # WTI Crude Futures, Continuous Contract
34     "CME_AD1",      # Australian Dollar Futures, Continuous Contract #1
35     "CME_BP1",      # British Pound Futures, Continuous Contract #1
36     "CME_CD1",      # Canadian Dollar Futures, Continuous Contract #1
37     "CME_EC1",      # Euro FX Futures, Continuous Contract #1
38     "CME_JY1",      # Japanese Yen Futures, Continuous Contract #1
39     "CME_SF1",      # Swiss Franc Futures, Continuous Contract #1
40                 
41     "CME_NQ1",      # E-mini NASDAQ 100 Futures, Continuous Contract #1
42     "CME_ES1",      # E-mini S&P 500 Futures, Continuous Contract #1
43     "EUREX_FSMI1",  # SMI Futures, Continuous Contract #1
44     "EUREX_FSTX1",  # STOXX Europe 50 Index Futures, Continuous Contract #1
45     "LIFFE_FCE1",   # CAC40 Index Futures, Continuous Contract #1
46     "LIFFE_Z1",     # FTSE 100 Index Futures, Continuous Contract #1
47             
48     "CME_TY1",      # 10 Yr Note Futures, Continuous Contract #1 -5000
49     "CME_FV1",      # 5 Yr Note Futures, Continuous Contract #1-8000
50     "CME_TU1",      # 2 Yr Note Futures, Continuous Contract #1 -10000
51     "ASX_XT1",      # 10 Year Commonwealth Treasury Bond Futures, Continuous Contract #1   # 'Settlement price' instead of 'settle' on quandl. 
52     "ASX_YT1",      # 3 Year Commonwealth Treasury Bond Futures, Continuous Contract #1    # 'Settlement price' instead of 'settle' on quandl.
53     "EUREX_FGBL1",  # Euro-Bund (10Y) Futures, Continuous Contract #1
54     "EUREX_FGBM1",  # Euro-Bobl Futures, Continuous Contract #1
55     "EUREX_FGBS1",  # Euro-Schatz Futures, Continuous Contract #1 
56     "SGX_JB1",      # SGX 10-Year Mini Japanese Government Bond Futures
57     "LIFFE_R1"      # Long Gilt Futures, Continuous Contract #1
58     "MX_CGB1",      # Ten-Year Government of Canada Bond Futures, Continuous Contract #1    # 'Settlement price' instead of 'settle' on quandl.
59 ]
60 
61 ma_period = 100
62 vol_period = 60
63 
64 self.SetWarmUp(vol_period)
65 self.data = {}
66 self.sma = {}
67 
68 self.days = 0
69 
70 for symbol in self.symbols:
71     data = self.AddData(QuantpediaFutures, symbol, Resolution.Daily)
72     data.SetLeverage(5)
73     data.SetFeeModel(CustomFeeModel())
74     
75     self.data[symbol] = deque(maxlen=vol_period)
76     self.sma[symbol] = self.SMA(symbol, ma_period, Resolution.Daily)
77     
78def OnData(self, data):
79 for symbol in self.symbols:
80     # data is still coming
81     if self.securities[symbol].get_last_data() and self.time.date() > QuantpediaFutures.get_last_update_date()[symbol]:
82         self.liquidate(symbol)
83         self.data[symbol].clear()
84         continue
85     if symbol in data and data[symbol]:
86         price = data[symbol].Value
87         self.data[symbol].append(price)        
88 
89 if self.IsWarmingUp: return
90 
91 if self.Portfolio.Invested:
92     self.days += 1
93     if self.days == 3:
94         self.Liquidate()
95         self.days = 0
96 offset = BMonthEnd()
97 last_day = offset.rollforward(self.Time)
98 # day before EOM
99 if self.Time.date() == last_day.date():
100     # Volatility calculation
101     volatility = {}
102     for symbol in self.symbols:
103         if len(self.data[symbol]) == self.data[symbol].maxlen:
104             volatility[symbol] = self.Volatility(self.data[symbol])
105     
106     if len(volatility) == 0: return
107
108     # MA sorting
109     long = [x[0] for x in volatility.items() if self.data[x[0]][-1] > self.sma[x[0]].Current.Value]
110     short = [x[0] for x in volatility.items() if self.data[x[0]][-1] < self.sma[x[0]].Current.Value]
111
112     # Volatility weighting
113     weight = {}
114     
115     total_vol_long = sum([1/volatility[x] for x in long])
116     if total_vol_long != 0:
117         for symbol in long:
118             vol = volatility[symbol]
119             if vol != 0:
120                 weight[symbol] = (1.0 / vol) / total_vol_long
121             else: 
122                 weight[symbol] = 0
123     total_vol_short = sum([1/volatility[x] for x in short])
124     if total_vol_short != 0:
125         for symbol in short:
126             vol = volatility[symbol]
127             if vol != 0:
128                 weight[symbol] = (1.0 / vol) / total_vol_short
129             else: 
130                 weight[symbol] = 0
131     
132     # Trade execution
133     for symbol in long:
134         if data.contains_key(symbol) and data[symbol]:
135             self.SetHoldings(symbol, weight[symbol])
136         
137     for symbol in short:
138         if data.contains_key(symbol) and data[symbol]:
139             self.SetHoldings(symbol, -weight[symbol])
140     
141def Volatility(self, history):
142 prices = np.array(history)
143 returns = (prices[1:]-prices[:-1])/prices[:-1]
144 vol = np.std(returns) * np.sqrt(252)
145 return vol
146# Quantpedia data
147class QuantpediaFutures(PythonData):
148_last_update_date:Dict[Symbol, datetime.date] = {}
149@staticmethod
150def get_last_update_date() -> Dict[Symbol, datetime.date]:
151return QuantpediaFutures._last_update_date
152def GetSource(self, config, date, isLiveMode):
153 return SubscriptionDataSource("http://data.quantpedia.com/backtesting_data/futures/{0}.csv".format(config.Symbol.Value), SubscriptionTransportMedium.RemoteFile, FileFormat.Csv)
154def Reader(self, config, line, date, isLiveMode):
155 data = QuantpediaFutures()
156 data.Symbol = config.Symbol
157 
158 try:
159     if not line[0].isdigit(): return None
160     split = line.split(';')
161     
162     data.Time = datetime.strptime(split[0], "%d.%m.%Y") + timedelta(days=1)
163     data['settle'] = float(split[1])
164     data.Value = float(split[1])
165     if config.Symbol.Value not in QuantpediaFutures._last_update_date:
166         QuantpediaFutures._last_update_date[config.Symbol.Value] = datetime(1,1,1).date()
167     if data.Time.date() > QuantpediaFutures._last_update_date[config.Symbol.Value]:
168         QuantpediaFutures._last_update_date[config.Symbol.Value] = data.Time.date()
169 except:
170     return None
171     
172 return data
173# Custom fee model
174class CustomFeeModel(FeeModel):
175def GetOrderFee(self, parameters):
176 fee = parameters.Security.Price * parameters.Order.AbsoluteQuantity * 0.00005
177 return OrderFee(CashAmount(fee, "USD"))