The Science Behind Insider Tracking

Corporate insiders — CEOs, CFOs, board members — know their companies better than any analyst. When they spend their own money buying shares on the open market, they're making a bet with real skin in the game. Decades of academic research confirm that these purchases consistently predict future stock returns.

A peer-reviewed study in 2012 analyzed over 65,000 insider transactions across 17 Western European countries. The headline finding: high conviction insider purchases generated an average excess return of +20.94% annually.

But not all insider trades are equal. The same study showed that unfocused tracking of all insider activity produces mediocre results. The edge comes entirely from filtering. High conviction trades outperform dramatically. Low conviction trades actually lose money.

What makes a trade "high conviction"?

The research points to a few consistent patterns. Cluster buying — multiple insiders at the same company buying within a short window. Infrequent traders making rare, deliberate purchases rather than routine compensation-related activity. Contrarian timing — insiders buying after their stock has dropped, not after it's run up. And undervalued companies where the market price hasn't caught up with what insiders already know.

Sells, on the other hand, carry almost no signal. Insiders sell for many reasons — taxes, diversification, buying a house. A purchase has essentially one motive: they believe the stock is going up.

How the Insider Transaction Terminal helps

We aggregate insider filings from 14 European financial regulators into a single dashboard, updated daily. We filter out options exercises, stock awards, and other compensation noise — showing only genuine open-market purchases and sales. Our cluster detection highlights when multiple insiders are moving in the same direction. And company profiles with price overlays let you see exactly when insiders bought relative to the stock's performance.

The research is clear: the edge exists. The challenge is accessing the data fast enough and filtering it well enough to act on it. That's what we built this for.