Original paper

SSRN-id3111334.pdf519.3KB

Abstract

The distance between short- and long-run moving averages of prices (MAD) predicts future equity returns in the cross-section. Annualized value-weighted alphas from the accompanying hedge portfolios are around 9%, and the predictability goes beyond momentum, 52-week highs, profitability, and other prominent anomalies. MAD-based investment payoffs survive reasonable trading costs faced by institutions, and are stronger on the long side relative to the short counterpart.

Keywords: market efficiency, technical analysis, moving averages, crossing rules,anchoring bias

Trading rules

Focus on: U.S. companies listed on NYSE, AMEX, and NASDAQ with stock codes 10/11, having a positive book value of equity in Compustat, among other specific factors.
Omit shares: those priced under $5 at month-end, those not traded within the month, and those lacking sufficient data on returns or company attributes.
Derive MAD by: Dividing 21-day moving average (MA(21)) by the 200-day moving average (MA(200)).
Set a benchmark value at 0.2.
Take long positions on stocks: where MAD meets or surpasses 1.2.
Assume short positions on stocks: where MAD is 0.8 or below.
All stocks in the portfolio should have equal weightage.
Adjust the portfolio composition every month.

Python code

Backtrader

import backtrader as bt

class MADStrategy(bt.Strategy):
    params = (
        ('ma_short', 21),
        ('ma_long', 200),
        ('long_threshold', 1.2),
        ('short_threshold', 0.8),
    )

    def __init__(self):
        self.ma_short = bt.indicators.SimpleMovingAverage(
            self.data.close, period=self.params.ma_short
        )
        self.ma_long = bt.indicators.SimpleMovingAverage(
            self.data.close, period=self.params.ma_long
        )
        self.mad = self.ma_short / self.ma_long

    def next(self):
        if not self.position:
            if self.mad[0] >= self.params.long_threshold and self.data.close[0] >= 5:
                self.buy()
            elif self.mad[0] <= self.params.short_threshold and self.data.close[0] >= 5:
                self.sell()
        elif self.position.size > 0 and self.mad[0] < self.params.long_threshold:
            self.close()
        elif self.position.size < 0 and self.mad[0] > self.params.short_threshold:
            self.close()

if __name__ == '__main__':
    cerebro = bt.Cerebro()
    data = bt.feeds.GenericCSVData(
        dataname='your-data-file.csv',
        dtformat=('%Y-%m-%d'),
        openinterest=-1,
        timeframe=bt.TimeFrame.Days,
        compression=1,
    )
    cerebro.adddata(data)
    cerebro.addstrategy(MADStrategy)
    cerebro.broker.setcash(10000.0)
    cerebro.addsizer(bt.sizers.EqualWeight)
    cerebro.addanalyzer(bt.analyzers.SharpeRatio, _name='sharpe')
    cerebro.addanalyzer(bt.analyzers.DrawDown, _name='drawdown')
    cerebro.addanalyzer(bt.analyzers.AnnualReturn, _name='annualreturn')
    print('Starting Portfolio Value: %.2f' % cerebro.broker.getvalue())
    result = cerebro.run()
    print('Final Portfolio Value: %.2f' % cerebro.broker.getvalue())
    print('Sharpe Ratio:', result[0].analyzers.sharpe.get_analysis()['sharperatio'])
    print('Max Drawdown:', result[0].analyzers.drawdown.get_analysis()['max']['drawdown'])
    print('Annual Return:', result[0].analyzers.annualreturn.get_analysis())

Moving Average Distance as a Predictor of Equity Returns

Original paper

Abstract

Trading rules

Python code

Backtrader