Original paper
Abstract
This paper shows that market breadth, i.e. the difference between the average number of rising stocks and the average number of falling stocks within a portfolio, is a robust predictor of future stock returns on market and industry portfolios for 64 countries for the period between 1973 and 2018. We link the market breadth with herd behavior and show that high market breadth portfolios significantly outperform low market breadth portfolios, and that this effect is robust to effects such as size, style, volatility, skewness, momentum, and trend-following signals. In addition, the role of market breadth is particularly strong among markets characterized by high limits to arbitrage, following bullish periods, and in collectivistic societies, supporting behavioral explanations of the phenomenon. We also examine practical implications of the effect and our results indicate that the effect may be employed for equity allocation and market timing, although frequent portfolio rebalancing can lead to higher transaction costs that may affect profitability.
Keywords:Â Herding, Market Breadth, Return Predictability, Asset Pricing, Portfolio Allocation
Trading rules
- Investment universe: 64 country ETFs
- Monthly calculation: Market Breadth (MBR) for each country
Python code
Backtrader
import backtrader as bt
class MarketBreadth(bt.Strategy):
params = (
('n_countries', 64),
('long_pct', 0.1),
('short_pct', 0.1),
)
def __init__(self):
self.inds = dict()
for d in self.datas:
self.inds[d] = dict()
self.inds[d]['rising'] = bt.indicators.SumN(d.close > d.close(-1), period=30)
self.inds[d]['falling'] = bt.indicators.SumN(d.close < d.close(-1), period=30)
self.inds[d]['mbr'] = (self.inds[d]['rising'] - self.inds[d]['falling']) / (self.inds[d]['rising'] + self.inds[d]['falling'])
def rebalance(self):
mbr_values = [(d, self.inds[d]['mbr'][0]) for d in self.datas]
long_countries = sorted(mbr_values, key=lambda x: x[1], reverse=True)[:int(self.params.n_countries * self.params.long_pct)]
short_countries = sorted(mbr_values, key=lambda x: x[1])[:int(self.params.n_countries * self.params.short_pct)]
for d, _ in long_countries:
self.order_target_percent(d, target=1 / len(long_countries))
for d, _ in short_countries:
self.order_target_percent(d, target=-1 / len(short_countries))
# Flatten positions for other countries
remaining_countries = [d for d in self.datas if d not in [x[0] for x in long_countries] and d not in [x[0] for x in short_countries]]
for d in remaining_countries:
self.order_target_percent(d, target=0)
def next(self):
if self.datetime.date(0).day == 1: # Rebalance monthly
self.rebalance()