How insider buys predict returns
The academic research behind why insider purchases tend to outperform, and the nuances.
The canonical paper is Lakonishok & Lee (2001), who found that US insider purchases predicted excess returns of about 4-6% per year over the following 12 months, with the signal strongest in small- and mid-cap stocks.
The logic is simple: insiders know the business better than any analyst. They see the pipeline, the hiring trends, the customer renewals. When they voluntarily convert salary or savings into their own company's stock, they have skin in the game.
But it's not symmetric. Insider SELLS are a much weaker signal because they have many non-directional motivations: tax, diversification, option-vesting schedules, divorce settlements, charity. We don't rank insiders on sells.
The catch: individual variation is huge. Research by Cohen, Malloy and Pomorski (2012) showed that about half of insiders are noise — their buys don't beat the market. The other half ("opportunistic" insiders) outperformed by 11% per year. The job is finding which half each insider is in.
How we separate them: for every insider with at least 3 disclosed buys, we compute their historical 3-month and 12-month forward returns. We apply Bayesian shrinkage so a 3-out-of-3 lucky streak doesn't top the leaderboard. Over time, the insiders who consistently time their buys rise to the top.
What could go wrong? Survivorship bias: we only see the buys that are already disclosed. A CEO at a company that went bankrupt is noise. Also: markets are more efficient than in 2001, so the raw-signal edge may have shrunk. Our honest view: it's still worth tracking as one input among many.