加密货币 · 全球 · 加密货币 · 每周 · 全球 Free

加密货币中的混合因子

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，该策略涉及构建等权重基准投资组合和基于动量、价值和套利的因子复合投资组合。基准在每个重新平衡日将10%的风险敞口预算平均分配给所有可用代币，并持有投资组合直至下一次重新平衡。因子型投资组合与基准相结合，以创建增强型投资组合。如果因子投资组合导致任何货币出现负权重，则将其调整为零。投资组合每周重新平衡。

II. 策略合理性

尽管样本期仅四年多，但由于加密货币市场的高波动性，该研究得出了有意义的结论。研究表明，动量、价值和套利因子相结合，可提供强大的风险调整后回报，超越了单个动量策略的表现。尽管动量单独表现优于套利和价值，但这些三个因子的结合增强了整体回报，证实了它们的互补性。这表明混合动量、价值和套利可以提高表现，表明这些因子在加密货币投资组合中具有有效的作用，因为它们分散了风险并产生了优于单独使用动量的回报。

回测表现

波动率13.2%

夏普比率2.91

索提诺比率1.163

胜率64%

完整 Python 代码

1import numpy as np
2from AlgorithmImports import *
3from typing import List, Dict
4class BlendedFactorsinCryptocurrencies(QCAlgorithm):
5def Initialize(self) -> None:
6 self.SetStartDate(2015, 1, 1)
7 self.SetCash(1_000_000)
8 
9 self.period: int = 7
10 self.count_days: int = 1
11 self.percentage_traded: float = 0.1
12 
13 self.symbols: Dict[str, str] = {
14     'BTC' : 'BTCUSD',
15     'ETH' : 'ETHUSD', 
16     'LTC' : 'LTCUSD', 
17     'ETC' : 'ETCUSD',
18     'XMR' : 'XMRUSD',
19     'ZEC' : 'ZECUSD'
20 }
21 
22 self.data: Dict[str, SymbolData] = {}
23 self.SetBrokerageModel(BrokerageName.Bitfinex)
24                 
25 for crypto, ticker in self.symbols.items():
26     data: Securities = self.AddCrypto(ticker, Resolution.Daily, Market.Bitfinex)
27     self.AddData(CryptoNetworkData, crypto, Resolution.Daily)
28     self.data[crypto] = SymbolData(self.period)
29     
30def OnData(self, data: Slice) -> None:
31 crypto_data_last_update_date: Dict[Symbol, datetime.date] = CryptoNetworkData.get_last_update_date()
32 # Store daily price 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         coin_issuance: float = data[crypto].Price
38         
39         if cap_mrkt_cur_usd != 0 and txtfr_val_adj_usd != 0 and coin_issuance != 0:
40             self.data[crypto].update_data(cap_mrkt_cur_usd, txtfr_val_adj_usd, coin_issuance)
41     
42     if ticker in data:
43         if data[ticker]:
44             self.data[crypto].update_price(data[ticker].Price)
45 
46 if self.Time.date().weekday() != 0:
47     return
48 if self.count_days == 7:
49     self.count_days = 1
50 else:
51     self.count_days = self.count_days + 1
52     return
53 
54 symbols_ready = [x for x in self.symbols if self.data[x].is_ready() and self.Securities[x].GetLastData() and self.Time.date() < crypto_data_last_update_date[x]]
55 if len(symbols_ready) == 0:
56     self.Liquidate()
57     return
58 
59 weight: Dict[ticker, float] = {}
60 partial_weight: float = self.percentage_traded / len(symbols_ready)
61 
62 carry_metric_long: List[str] = []
63 carry_metric_short: List[str] = []
64 valuation_metric_long: List[str] = []
65 valuation_metric_short: List[str] = []
66 momentum_long: List[str] = []
67 momentum_short: List[str] = []
68 
69 for crypto in symbols_ready:
70     ticker: str = self.symbols[crypto]
71     weight[ticker] = partial_weight   # Set benchmark weight.
72     
73     carry_metric: float = self.data[crypto].carry_metric()
74     valuation_metric: float = self.data[crypto].valuation_metric()
75     momentum: float = self.data[crypto].momentum()
76     
77     carry_metric_long.append(ticker) if carry_metric > 0 else carry_metric_short.append(ticker)
78     valuation_metric_long.append(ticker) if valuation_metric > 0 else valuation_metric_short.append(ticker)
79     momentum_long.append(ticker) if momentum > 0 else momentum_short.append(ticker)
80 
81 for i, portfolio in enumerate([[carry_metric_long, valuation_metric_long, momentum_long], [carry_metric_short, valuation_metric_short, momentum_short]]):
82     for sub_portfolio in portfolio:
83         for ticker in sub_portfolio:
84             weight[ticker] += ((-1)**i) * self.percentage_traded / len(sub_portfolio)
85 # trade execution
86 invested: List[str] = [x.Key.Value for x in self.Portfolio if x.Value.Invested]
87 for symbol in invested:
88     if symbol not in weight:
89         self.Liquidate(symbol)
90 for symbol, w in weight.items():
91     if symbol in data and data[symbol]:
92         self.SetHoldings(symbol, w)
93class SymbolData():
94def __init__(self, period: int) -> None:
95 self.coin_issuance: RollingWindow = RollingWindow[float](period)
96 self.transactions: RollingWindow = RollingWindow[float](period)
97 self.curr_market_cap: float = 0.
98 self.Price: RollingWindow = RollingWindow[float](period)
99 
100def update_data(self, current_market_value: float, num_of_transactions: int, coin_issuance: float) -> None:
101 self.transactions.Add(num_of_transactions)
102 self.curr_market_cap = current_market_value
103 self.coin_issuance.Add(coin_issuance)
104 
105def update_price(self, price: float) -> None:
106 self.Price.Add(price)
107 
108def carry_metric(self) -> float:
109 seven_days_coin_issuance: List[float] = [x for x in self.coin_issuance]
110 # return -1 * (sum(seven_days_coin_issuance) / seven_days_coin_issuance[-1])
111 return (sum(seven_days_coin_issuance) / seven_days_coin_issuance[-1])
112def valuation_metric(self) -> float:
113 trailing_data: List[float] = [x for x in self.transactions]
114 return self.curr_market_cap / np.mean(trailing_data)
115
116def momentum(self) -> float:
117 prices: List[float] = [x for x in self.Price]
118 return prices[0] / prices[-1] - 1
119 
120def is_ready(self) -> bool:
121 return self.coin_issuance.IsReady and self.transactions.IsReady and self.Price.IsReady
122 
123# Crypto network data.
124# NOTE: IMPORTANT: Data order must be ascending (datewise)
125# Data source: https://coinmetrics.io/community-network-data/
126class CryptoNetworkData(PythonData):
127_last_update_date: Dict[Symbol, datetime.date] = {}
128@staticmethod
129def get_last_update_date() -> Dict[Symbol, datetime.date]:
130return CryptoNetworkData._last_update_date
131def GetSource(self, config: SubscriptionDataConfig, date: datetime, isLiveMode: bool) -> SubscriptionDataSource:
132 return SubscriptionDataSource(f"data.quantpedia.com/backtesting_data/crypto/{config.Symbol.Value}_network_data.csv", SubscriptionTransportMedium.RemoteFile, FileFormat.Csv)
133# File exmaple:
134# 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
135# 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
136def Reader(self, config: SubscriptionDataConfig, line: str, date: datetime, isLiveMode: bool) -> BaseData:
137 data: CryptoNetworkData = CryptoNetworkData()
138 data.Symbol = config.Symbol
139 try:
140     cols:str = ['SplyCur', 'CapMrktCurUSD', 'TxTfrValAdjUSD']
141     if not line[0].isdigit():
142         header_split = line.split(',')
143         self.col_index = [header_split.index(x) for x in cols]
144         return None
145     split = line.split(',')
146     
147     data.Time = datetime.strptime(split[0], "%Y-%m-%d") + timedelta(days=1)
148     for i, col in enumerate(cols):
149         data[col] = float(split[self.col_index[i]])
150     data.Value = float(split[self.col_index[0]])
151     if config.Symbol.Value not in CryptoNetworkData._last_update_date:
152         CryptoNetworkData._last_update_date[config.Symbol.Value] = datetime(1,1,1).date()
153     if data.Time.date() > CryptoNetworkData._last_update_date[config.Symbol.Value]:
154         CryptoNetworkData._last_update_date[config.Symbol.Value] = data.Time.date()
155     
156 except:
157     return None
158 return data