Original paper
Abstract
This article studies the relationship between initial market response to earnings surprise and subsequent stock price movement. We first develop a new measure – the earnings response elasticity (ERE) – to capture initial market response. It is defined as the absolute value of earnings announcement abnormal returns (EAARs) divided by the earnings surprise. The ERE is then examined under various categories contingent on the signs of earnings surprises (+/-/0) and EAARs (+/-). We find that a weaker initial market reaction to earnings surprises, or lower ERE, leads to a larger post-announcement drift. A trading strategy of taking a long position in stocks in the lowest ERE quintile when both earnings surprises and EAARs are positive and a short position when both are negative can generate an average abnormal return of 5.11 percent per quarter.
Keywords:Â earnings surprise; post earnings announcement drifts
Trading rules
- Target assets: Stocks on NYSE.
- Determine quarterly Earnings Response Elasticity (ERE) for each stock:
- ERE = |Abnormal Return (3-day window centered on earnings announcement date)| / Accounting Earnings Surprise
- Classify stocks into five groups based on ERE.
- For buying: Select the stocks in the lowest ERE group that have positive earnings surprises and positive abnormal returns.
- For short-selling: Choose from the highest ERS group that have negative earnings surprises and negative abnormal returns.
- Retain chosen stocks until the next quarter.
- Rebalance strategy daily given the different dates of earnings announcements.
Python code
Backtrader
import backtrader as bt
import math
import numpy as np
class EREStrategy(bt.Strategy):
params = (
('quintile_size', 0.2),
)
def __init__(self):
self.ere = dict()
self.next_quarter = dict()
def next(self):
# Calculate ERE for each stock in a quarter
for d in self.getdatanames():
data = self.getdatabyname(d)
if data.datetime.date() >= self.next_quarter[d]:
accounting_earnings_surprise = data.close[0] - data.close[-1]
abnormal_return = abs(data.close[0] - data.close[-1]) / data.close[-1]
self.ere[d] = abnormal_return / accounting_earnings_surprise if accounting_earnings_surprise != 0 else 0
self.next_quarter[d] = data.datetime.date() + bt.DateOffset(months=3)
# Sort stocks into quintiles based on ERE
sorted_stocks = sorted(self.ere.items(), key=lambda x: x[1])
quintile_size = int(len(sorted_stocks) * self.params.quintile_size)
# Enter long and short positions
for i, (d, ere_value) in enumerate(sorted_stocks):
data = self.getdatabyname(d)
position = self.getpositionbyname(d)
if i < quintile_size:
if ere_value > 0 and data.close[0] > data.close[-1] and not position:
self.buy(data=data)
elif i >= len(sorted_stocks) - quintile_size:
if ere_value < 0 and data.close[0] < data.close[-1] and not position:
self.sell(data=data)
# Close positions at the end of the quarter
if data.datetime.date() >= self.next_quarter[d]:
if position:
self.close(data=data)
if __name__ == '__main__':
cerebro = bt.Cerebro()
# Add data for NYSE stocks
for ticker in nyse_stock_list:
data = bt.feeds.YahooFinanceData(dataname=ticker, fromdate=start_date, todate=end_date)
cerebro.adddata(data, name=ticker)
cerebro.addstrategy(EREStrategy)
cerebro.run()Please note that this code assumes you have a list of NYSE stock tickers called nyse_stock_list and appropriate start_date and end_date variables. Replace them with your actual data sources and date range.