Bond · United States · ETFs · Monthly · United States Free

The Crisis Alpha Portfolio

Academic paper

Paper linkhttps://papers.ssrn.com/sol3/papers.cfm?abstract_id=2739520

Strategy in a nutshell

This strategy trades five U.S. Treasury ETFs with maturities from 1 to 20 years. Absolute momentum is measured over nine months, while relative momentum ranks assets by risk-adjusted returns. Positive momentum assets are included, otherwise full weight goes to 1-year Treasuries. Portfolio weights are rank-proportional and rebalanced monthly, dynamically adjusting to momentum signals.

Economic rationale

U.S. Treasuries act as safe-haven assets during market crises due to low credit risk and high liquidity. Leveraging relative momentum and tactical overweighting, this strategy captures crisis alpha, enhancing returns and mitigating risk across market cycles.

Backtest performance

Annualised return7.29%

Volatility5.04%

Beta-0.017

Sharpe ratio0.32

Sortino ratio-0.459

Maximum drawdown-5.9%

Win rate51%

Full Python code

1import numpy as np
2from AlgorithmImports import *
3class TheCrisisAlphaPortfolio(QCAlgorithm):
4def Initialize(self):
5self.SetStartDate(2008, 1, 1)
6self.SetCash(100000)
7self.symbols = ['TLT', 'IEF', 'IEI', 'SHY']
8self.treasuries_1year = 'BIL'
9
10period = 10
11self.SetWarmUp(period)
12
13# Monthly etf price.
14self.data = {}
15
16for symbol in self.symbols + [self.treasuries_1year]:
17    data = self.AddEquity(symbol, Resolution.Daily)
18    self.data[symbol] = SymbolData(symbol, period)
19
20self.last_month = -1
21self.Schedule.On(self.DateRules.MonthStart(self.symbols[0]), self.TimeRules.At(0, 0), self.Rebalance)
22
23def OnData(self, data):
24if self.last_month == self.Time.month: return
25self.last_month = self.Time.month
26# Store monthly prices.
27for symbol in self.symbols + [self.treasuries_1year]:
28    symbol_obj = self.Symbol(symbol)
29    if symbol_obj in data.Keys:
30        if data[symbol_obj]:
31            price = data[symbol_obj].Value
32            if price != 0:
33                self.data[symbol].update(price)
34
35def Rebalance(self):
36self.Liquidate()
37
38# Etfs with positive momentum. - symbol -> (performance, volatility)
39positive_mom = {x : (self.data[x].performance(), self.data[x].volatility()) for x in self.symbols if self.data[x].is_ready() and self.data[x].performance() > 0}
40
41if not self.data[self.treasuries_1year].is_ready(): return
42if len(positive_mom) > 0:
43    bil_ret = self.data[self.treasuries_1year].performance()
44    
45    rank = {}
46    for symbol, perf_volatility in positive_mom.items():
47        excess_ret = perf_volatility[0] - bil_ret
48        variance = perf_volatility[1] ** 2
49        rank[symbol] = excess_ret / variance
50    
51    sorted_by_rank = [x[0] for x in sorted(rank.items(), key = lambda x: x[1], reverse = True)]
52    
53    rank_len = len(sorted_by_rank)
54    total_score = sum(range(0, rank_len + 1))
55    
56    rank_index = rank_len
57    for symbol in sorted_by_rank:
58        weight = (1 / total_score) * rank_index
59        self.SetHoldings(symbol, weight)
60        rank_index -= 1
61else:
62    self.SetHoldings(self.treasuries_1year, 1)
63    
64class SymbolData():
65def __init__(self, symbol, period):
66self.Symbol = symbol
67self.Price = RollingWindow[float](period)
68
69def update(self, value):
70self.Price.Add(value)
71
72def is_ready(self) -> bool:
73return self.Price.IsReady
74
75def performance(self, values_to_skip = 0) -> float:
76closes = [x for x in self.Price][values_to_skip:]
77return (closes[0] / closes[-1] - 1)
78
79def volatility(self):
80prices = np.array([x for x in self.Price])
81daily_returns = prices[:-1] / prices[1:] - 1
82return np.std(daily_returns)