Original paper
Abstract
Signaling undervaluation is often considered a primary motive for repurchasing stock, but insider trading activity by repurchasing firms is not always consistent with undervaluation. Net insider buying and selling are both more frequent in quarters when firms are repurchasing non-trivial amounts of stock, with the odds of observing a repurchase the highest in quarters with net insider selling. In multinomial logit models, shares repurchases associated with net insider selling are positively related to illiquidity, option exercises by insiders, and pre-repurchase returns and negatively correlated with industry-adjusted book to market ratios when compared to other repurchases. Hence, repurchases when insiders are selling stock are more likely done to support share prices or avoid dilution and are less likely undervaluation signals. We find that insider trades either validate or mitigate the undervaluation signal of the repurchase. Abnormal returns of repurchasing firms with net insider buying versus net insider selling in a given quarter are significantly higher for the quarter immediately after the repurchase and the three subsequent years. For repurchases accompanied by net insider selling, abnormal returns are negligible after only one year.
Keywords:Â Repurchases, buybacks, insider trading, long-run returns, signalling
Trading rules
- Scope of investment: Focus on stocks from NYSE, AMEX, and NASDAQ using data sourced from Compustat.
- Stock buybacks: Determine by subtracting any decrease in redeemable preferred stock from the expenditure on both common and preferred stock repurchases, then adjust this value by the fraction of shares in circulation.
- Non-trivial buybacks: Recognized when they amount to 0.5% of the firm's market value.
- Insider transactions: Calculate total dollar values of insider open market buybacks and sales for each fiscal quarter
- Net buying insiders: Identified when insider purchases exceed insider sales by at least 0.005% of the firm’s market capitalization
- Stock selection: Buy stocks with non-trivial buybacks and net buying insiders in the previous quarter
- Holding period: Stocks held for one year
- Rebalancing: 1/4 of the portfolio rebalanced each quarter
- Portfolio weighting: Equally weighted stocks
Python code
Backtrader
import backtrader as bt
import datetime
class InsiderRepurchaseStrategy(bt.Strategy):
def __init__(self):
self.share_repurchase = None # To be calculated based on input data
self.insider_transactions = None # To be calculated based on input data
self.non_trivial_repurchase = 0.005
self.net_buying_insiders = 0.00005
def next(self):
to_buy = []
for data in self.datas:
if self.share_repurchase[data] > self.non_trivial_repurchase * data.close[0]:
if self.insider_transactions[data]['purchases'] - self.insider_transactions[data]['sales'] > self.net_buying_insiders * data.close[0] * data.open_interest[0]:
to_buy.append(data)
if to_buy:
self.rebalance_portfolio(to_buy)
def rebalance_portfolio(self, to_buy):
target_weight = 1.0 / len(to_buy)
for data in to_buy:
position_size = target_weight * self.broker.get_cash()
self.order_target_value(data, position_size)
if __name__ == '__main__':
cerebro = bt.Cerebro()
# Add stocks data from NYSE, AMEX, and NASDAQ here
# cerebro.adddata(...)
# cerebro.adddata(...)
# ...
cerebro.addstrategy(InsiderRepurchaseStrategy)
cerebro.broker.set_cash(100000)
cerebro.addanalyzer(bt.analyzers.SharpeRatio, _name='sharpe_ratio')
cerebro.addanalyzer(bt.analyzers.DrawDown, _name='draw_down')
cerebro.addsizer(bt.sizers.PercentSizer, percents=25)
results = cerebro.run()
cerebro.plot()
print(f"Sharpe Ratio: {results[0].analyzers.sharpe_ratio.get_analysis()['sharperatio']}")
print(f"Maximum Drawdown: {results[0].analyzers.draw_down.get_analysis()['max']['drawdown']}")Please note that this code is a basic outline of the strategy and assumes the necessary data is available for share repurchases and insider transactions. You’ll need to provide the data for each stock from the investment universe and also calculate the share_repurchase and insider_transactions variables using the provided data.