Original paper
Abstract
In this paper, we document that an application of a moving average strategy of technical analysis to portfolios sorted by volatility generates investment timing portfolios that often outperform the buy-and-hold strategy substantially. For high volatility portfolios, the abnormal returns, relative to the CAPM and the Fama-French three-factor models, are high, and higher than those from the well known momentum strategy. The abnormal returns remain high even after accounting for transaction costs. Although both the moving average and the momentum strategies are trend-following methods, their performances are surprisingly uncorrelated and behave differently over the business cycles, default and liquidity risk.
Keywords:Â Technical Analysis, Moving Average, Anomaly, Market Timing
Trading rules
- Categorize stocks from NYSE, AMEX, and NASDAQ based on their market capitalization.
- Focus on the lower tenth fraction, targeting the smallest stocks.
- Adopt a long position on stocks surpassing their 10-day moving average by the day's end, indicating an upward trajectory.
- Adjust the portfolio on a day-to-day basis.
- In the portfolio, apply a value-based weighting for stocks that align with the entry criteria.
- If stocks don't align with the 10-day SMA guideline, keep the equivalent value in cash.
- Opt between a solely long strategy or a combined long-short approach (short all stocks in size decile for pure alpha).
- For a fine-tuned strategy, sort small stocks into deciles based on volatility and focus on most volatile stocks.
Python code
Backtrader
import backtrader as bt
import backtrader.feeds as btfeeds
import datetime
class MovingAverageCrossStrategy(bt.Strategy):
params = (
('sma_period', 10),
)
def __init__(self):
self.sma = bt.indicators.SimpleMovingAverage(self.data.close, period=self.params.sma_period)
def next(self):
if self.data.close[0] > self.sma[0]:
self.order_target_percent(self.data, target=1.0)
else:
self.order_target_percent(self.data, target=0.0)
class LongShort(bt.Strategy):
params = (
('sma_period', 10),
)
def __init__(self):
self.sma = bt.indicators.SimpleMovingAverage(self.data.close, period=self.params.sma_period)
def next(self):
if self.data.close[0] > self.sma[0]:
self.order_target_percent(self.data, target=1.0)
else:
self.order_target_percent(self.data, target=-1.0)
if __name__ == '__main__':
cerebro = bt.Cerebro()
# Data loading
data = btfeeds.GenericCSVData(
dataname='your_data.csv',
fromdate=datetime.datetime(2004, 1, 1),
todate=datetime.datetime(2009, 12, 31),
dtformat=('%Y-%m-%d'),
openinterest=-1,
)
cerebro.adddata(data)
# Select strategy
strategy_choice = 'long_only' # 'long_short' for the long-short version
if strategy_choice == 'long_only':
cerebro.addstrategy(MovingAverageCrossStrategy)
else:
cerebro.addstrategy(LongShort)
cerebro.run()Replace 'your_data.csv' with the path to your data file. The data file should contain historical stock data for NYSE, AMEX, and NASDAQ stocks, sorted by size and focused on the bottom decile. You can choose between the long-only or long-short strategy version by changing the value of strategy_choice. The provided code does not include the value-weighted portfolio, daily rebalancing, and volatility decile sorting. These features would require custom implementation, depending on the available data and desired granularity.