Original paper
Abstract
In this study, we investigate the return enhancement ability of style momentum strategy: a strategy that switches between value and growth styles based on previous performance. We explore the variation in abnormal returns of long-only and long-short momentum strategies using various style based indexes (Russell value/growth indexes, Fama-French value/growth indexes, and MSCI value/growth indexes) where value and growth stocks are classified using different criteria. Our results show that the performance of style momentum does vary across different index families. We first find that in general the long-only strategies create significant positive abnormal returns whereas the long-short strategies do not. Second, for a fixed formation period, abnormal returns of the strategies tend to decrease when the length of holding periods increase. Third, abnormal returns are stronger and more significant when rotating within large cap value and growth indexes while abnormal returns are weaker and inconsistent when rotating within small cap value and growth indexes. Fourth, strategies based on rotating across all market cap levels do not generate consistently significant positive abnormal returns for Russell indexes or Fama-French indexes but they do for MSCI indexes. Fifth, individual stock momentum only explains a very small portion of the returns of style moment strategies
Keywords:Â Style Momentum, Value, Growth, Large Cap, Small Cap
Trading rules
- Use six Russell ETFs that depict different equity styles.
- Monthly, determine the one-year momentum for each of these styles.
- Adopt a long position on the style with the strongest momentum and a short position on the one with the weakest.
- Adjust the portfolio composition on a monthly basis.
Python code
Backtrader
import backtrader as bt
class MomentumStyleRotation(bt.Strategy):
params = (
("momentum_period", 12),
)
def __init__(self):
self.etfs = self.datas
self.momentum = [bt.indicators.Momentum(e, period=self.params.momentum_period) for e in self.etfs]
def next(self):
long_etf = max(zip(self.momentum, self.etfs), key=lambda x: x[0][0])
short_etf = min(zip(self.momentum, self.etfs), key=lambda x: x[0][0])
for etf in self.etfs:
if etf == long_etf[1]:
self.order_target_percent(etf, target=1)
elif etf == short_etf[1]:
self.order_target_percent(etf, target=-1)
else:
self.order_target_percent(etf, target=0)
if __name__ == "__main__":
cerebro = bt.Cerebro()
# Add six Russell ETFs data feeds
for etf_code in ["ETF1", "ETF2", "ETF3", "ETF4", "ETF5", "ETF6"]:
data = bt.feeds.YahooFinanceData(dataname=etf_code, fromdate="2000-01-01", todate="2021-12-31")
cerebro.adddata(data)
cerebro.addstrategy(MomentumStyleRotation)
cerebro.broker.setcash(100000)
cerebro.broker.setcommission(commission=0.001)
cerebro.run()