加密货币 · 全球 · 加密货币 · 每周 · 全球 Free

加密货币中的价值因子

登录后收藏

学术论文

'Know When to Hodl Em, Know When to Fodl Em': An Investigation of Factor Based Investing in the Cryptocurrency Space

作者Stefan Hubrich

机构

T. Rowe Price Program for Charitable Giving
?T.Rowe Price

论文链接https://ssrn.com/abstract=3055498

策略概要

投资范围包括11种加密货币。原始估值指标计算为当前市值与过去七天美元计价链上交易平均值之比。然后，该原始价值因子通过纵向去均值并除以标准差进行标准化，为每种货币创建一个均值为零、标准差为一的标准化变量。投资组合采用等权重，总敞口限制为10%，这意味着10%的投资组合分配给加密货币。当价值因子高于零时，权重为正；当价值因子低于零时，权重为负。虽然无法做空加密货币，但可以采用只做多策略。投资组合每周重新平衡。论文中有两种加权方案，其中等权重策略是出于说明目的而选择的。第二种基于风险的方案也在论文中有详细说明。

II. 策略合理性

对加密货币的研究虽然基于仅四年多的短期样本，但由于加密市场的高波动性，得出了稳健的结论。研究分析了每周回报，投资组合每周重新平衡。波动性是有益的，因为它允许以较小的资本配置获得显著回报，这是无杠杆投资组合的关键优势。高波动性资产提供了投资规模的灵活性。使用各种基准投资组合和线性回归进行的分析表明，价值因子提供了一些附加价值，并且可以成为加密货币市场中盈利策略的一部分，尽管动量策略产生了更强的结果。当价值与动量结合在复合策略中时，仍然可以带来益处。

回测表现

波动率7.1%

夏普比率1.13

索提诺比率1.093

胜率68%

完整 Python 代码

1from AlgorithmImports import *
2from typing import List, Dict
3class ValueFactorInCryptocurrencies(QCAlgorithm):
4def Initialize(self) -> None:
5 self.SetStartDate(2015, 1, 1)
6 self.SetCash(1_000_000)
7 
8 self.period: int = 7
9 self.percentage_traded: float = 0.1
10 
11 self.symbols: Dict[str, str] = {
12     'BTC' : 'BTCUSD',
13     'ETH' : 'ETHUSD', 
14     'LTC' : 'LTCUSD', 
15     'ETC' : 'ETCUSD',
16     'XMR' : 'XMRUSD',
17     'ZEC' : 'ZECUSD',
18 }
19  
20 self.data: Dict[str, SymbolData] = {}
21 
22 self.SetBrokerageModel(BrokerageName.Bitfinex)  
23 for crypto, ticker in self.symbols.items():
24     data: Securities = self.AddCrypto(ticker, Resolution.Daily, Market.Bitfinex)
25     
26     self.AddData(CryptoNetworkData, crypto, Resolution.Daily)
27     self.data[crypto] = SymbolData(self.period)
28 
29 self.Settings.MinimumOrderMarginPortfolioPercentage = 0.
30def OnData(self, data: Slice) -> None:
31 crypto_data_last_update_date: Dict[Symbol, datetime.date] = CryptoNetworkData.get_last_update_date()
32 # Store daily data.
33 for crypto, ticker in self.symbols.items():
34     if crypto in data and data[crypto]:
35         cap_mrkt_cur_usd: float = data[crypto].Capmrktcurusd
36         txtfr_val_adj_usd: float = data[crypto].Txtfrvaladjusd
37         self.data[crypto].update_attributes(cap_mrkt_cur_usd, txtfr_val_adj_usd)
38 if self.Time.date().weekday() != 0:
39     return
40 long: List[str] = []
41 short: List[str] = []
42 
43 for crypto, ticker in self.symbols.items():
44     if self.Securities[crypto].GetLastData() and self.Time.date() > crypto_data_last_update_date[crypto]:
45         continue
46     
47     if self.data[crypto].is_ready():
48         valuation_metric = self.data[crypto].valuation_metric()
49         
50         if valuation_metric > 0.:
51             long.append(self.Symbol(ticker))
52         else:
53             short.append(self.Symbol(ticker))
54 
55 # Trade Execution
56 invested: List[Symbol] = [x.Key for x in self.Portfolio if x.Value.Invested]
57 for symbol in invested:
58     if symbol not in long + short:
59         self.Liquidate(symbol)
60 for i, portfolio in enumerate([long, short]):
61     for symbol in portfolio:
62         if symbol in data and data[symbol]:
63             self.SetHoldings(symbol, round((((-1) ** i) / len(portfolio)) * self.percentage_traded, 2))
64 
65class SymbolData():
66def __init__(self, period: int) -> None:
67 self.transactions: RollingWindow = RollingWindow[float](period)
68 self.curr_market_cap: float = 0.
69 
70def update_attributes(self, current_market_value: float, num_of_transactions: int) -> None:
71 self.transactions.Add(num_of_transactions)
72 self.curr_market_cap = current_market_value
73 
74def valuation_metric(self) -> float:
75 trailing_data: np.ndarray = np.array([x for x in self.transactions])
76 return self.curr_market_cap / np.mean(trailing_data)
77 
78def is_ready(self) -> bool:
79 return self.transactions.IsReady
80 
81# Crypto network data.
82# NOTE: IMPORTANT: Data order must be ascending (datewise)
83# Data source: https://coinmetrics.io/community-network-data/
84class CryptoNetworkData(PythonData):
85_last_update_date: Dict[Symbol, datetime.date] = {}
86@staticmethod
87def get_last_update_date() -> Dict[Symbol, datetime.date]:
88return CryptoNetworkData._last_update_date
89def GetSource(self, config: SubscriptionDataConfig, date: datetime, isLiveMode: bool) -> SubscriptionDataSource:
90 return SubscriptionDataSource(f"data.quantpedia.com/backtesting_data/crypto/{config.Symbol.Value}_network_data.csv", SubscriptionTransportMedium.RemoteFile, FileFormat.Csv)
91# File exmaple:
92# date,AdrActCnt,AdrBal1in100KCnt,AdrBal1in100MCnt,AdrBal1in10BCnt,AdrBal1in10KCnt,AdrBal1in10MCnt,AdrBal1in1BCnt,AdrBal1in1KCnt,AdrBal1in1MCnt,AdrBalCnt,AdrBalNtv0.001Cnt,AdrBalNtv0.01Cnt,AdrBalNtv0.1Cnt,AdrBalNtv100Cnt,AdrBalNtv100KCnt,AdrBalNtv10Cnt,AdrBalNtv10KCnt,AdrBalNtv1Cnt,AdrBalNtv1KCnt,AdrBalNtv1MCnt,AdrBalUSD100Cnt,AdrBalUSD100KCnt,AdrBalUSD10Cnt,AdrBalUSD10KCnt,AdrBalUSD10MCnt,AdrBalUSD1Cnt,AdrBalUSD1KCnt,AdrBalUSD1MCnt,AssetEODCompletionTime,BlkCnt,BlkSizeMeanByte,BlkWghtMean,BlkWghtTot,CapAct1yrUSD,CapMVRVCur,CapMVRVFF,CapMrktCurUSD,CapMrktFFUSD,CapRealUSD,DiffLast,DiffMean,FeeByteMeanNtv,FeeMeanNtv,FeeMeanUSD,FeeMedNtv,FeeMedUSD,FeeTotNtv,FeeTotUSD,FlowInExNtv,FlowInExUSD,FlowOutExNtv,FlowOutExUSD,FlowTfrFromExCnt,HashRate,HashRate30d,IssContNtv,IssContPctAnn,IssContPctDay,IssContUSD,IssTotNtv,IssTotUSD,NDF,NVTAdj,NVTAdj90,NVTAdjFF,NVTAdjFF90,PriceBTC,PriceUSD,ROI1yr,ROI30d,RevAllTimeUSD,RevHashNtv,RevHashRateNtv,RevHashRateUSD,RevHashUSD,RevNtv,RevUSD,SER,SplyAct10yr,SplyAct180d,SplyAct1d,SplyAct1yr,SplyAct2yr,SplyAct30d,SplyAct3yr,SplyAct4yr,SplyAct5yr,SplyAct7d,SplyAct90d,SplyActEver,SplyActPct1yr,SplyAdrBal1in100K,SplyAdrBal1in100M,SplyAdrBal1in10B,SplyAdrBal1in10K,SplyAdrBal1in10M,SplyAdrBal1in1B,SplyAdrBal1in1K,SplyAdrBal1in1M,SplyAdrBalNtv0.001,SplyAdrBalNtv0.01,SplyAdrBalNtv0.1,SplyAdrBalNtv1,SplyAdrBalNtv10,SplyAdrBalNtv100,SplyAdrBalNtv100K,SplyAdrBalNtv10K,SplyAdrBalNtv1K,SplyAdrBalNtv1M,SplyAdrBalUSD1,SplyAdrBalUSD10,SplyAdrBalUSD100,SplyAdrBalUSD100K,SplyAdrBalUSD10K,SplyAdrBalUSD10M,SplyAdrBalUSD1K,SplyAdrBalUSD1M,SplyAdrTop100,SplyAdrTop10Pct,SplyAdrTop1Pct,SplyCur,SplyExpFut10yr,SplyFF,SplyMiner0HopAllNtv,SplyMiner0HopAllUSD,SplyMiner1HopAllNtv,SplyMiner1HopAllUSD,TxCnt,TxCntSec,TxTfrCnt,TxTfrValAdjNtv,TxTfrValAdjUSD,TxTfrValMeanNtv,TxTfrValMeanUSD,TxTfrValMedNtv,TxTfrValMedUSD,VelCur1yr,VtyDayRet180d,VtyDayRet30d
93# 2009-01-09,19,19,19,19,19,19,19,19,19,19,19,19,19,0,0,19,0,19,0,0,0,0,0,0,0,0,0,0,1614334886,19,215,860,16340,0,0,0,0,0,0,1,1,0,0,0,0,0,0,0,0,0,0,0,0,9.44495122962963E-7,0,950,36500,100,0,950,0,1,0,0,0,0,1,0,0,0,0,11641.53218269,1005828380.584716757433,0,0,950,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,950,950,950,950,950,950,950,950,950,950,950,950,950,0,0,0,0,0,0,0,0,0,0,0,0,0,950,50,50,950,17070250,950,1000,0,1000,0,0,0,0,0,0,0,0,0,0,0,0,0
94def Reader(self, config: SubscriptionDataConfig, line: str, date: datetime, isLiveMode: bool) -> BaseData:
95 data: CryptoNetworkData = CryptoNetworkData()
96 data.Symbol = config.Symbol
97 try:
98     cols:str = ['CapMrktCurUSD', 'TxTfrValAdjUSD']
99     if not line[0].isdigit():
100         header_split = line.split(',')
101         self.col_index = [header_split.index(x) for x in cols]
102         return None
103     split = line.split(',')
104     
105     data.Time = datetime.strptime(split[0], "%Y-%m-%d") + timedelta(days=1)
106     
107     for i, col in enumerate(cols):
108         data[col] = float(split[self.col_index[i]])
109     data.Value = float(split[self.col_index[0]])
110         
111     if config.Symbol.Value not in CryptoNetworkData._last_update_date:
112         CryptoNetworkData._last_update_date[config.Symbol.Value] = datetime(1,1,1).date()
113     if data.Time.date() > CryptoNetworkData._last_update_date[config.Symbol.Value]:
114         CryptoNetworkData._last_update_date[config.Symbol.Value] = data.Time.date()
115 except:
116     return None
117 return data