Do insider cluster buys beat the market?
We replay every past insider cluster buy in our data and track what the stock did over the next 3, 6 and 12 months against the S&P 500. No hindsight: each signal is dated on the day its last Form 4 was filed, and returns start at the next close.
All cluster buys
| Signal | vs S&P · 3 months | vs S&P · 6 months | vs S&P · 12 months |
|---|---|---|---|
| Every cluster buy | +0.6% median −1.8% · 45% beat · n=926 | −1.9% median −5.2% · 44% beat · n=837 | −4.9% median −11.7% · 40% beat · n=651 |
By size of the buying
Total open-market $ from the cluster's buyers at the time of the signal.
| Signal | vs S&P · 3 months | vs S&P · 6 months | vs S&P · 12 months |
|---|---|---|---|
| Under $1M | +1.5% median −1.1% · 46% beat · n=686 | −1.8% median −4.4% · 45% beat · n=629 | −5.1% median −10.1% · 40% beat · n=499 |
| $1M–$5M | −2.7% median −4.7% · 41% beat · n=159 | −4.5% median −10.4% · 36% beat · n=138 | −5.1% median −15.3% · 34% beat · n=104 |
| $5M+ | −0.8% median −5.7% · 41% beat · n=81 | +3.2% median −1.0% · 47% beat · n=70 | −2.6% median −19.0% · 46% beat · n=48 |
By number of insiders
Distinct insiders buying in the 60 days up to the signal.
| Signal | vs S&P · 3 months | vs S&P · 6 months | vs S&P · 12 months |
|---|---|---|---|
| 3 buyers | +1.0% median −2.5% · 45% beat · n=611 | −2.1% median −7.8% · 40% beat · n=564 | −5.4% median −14.9% · 37% beat · n=423 |
| 4+ buyers | −0.2% median −1.6% · 44% beat · n=315 | −1.4% median +0.3% · 51% beat · n=273 | −3.9% median −3.7% · 45% beat · n=228 |
By the buyers' earlier record
Whether these insiders' earlier buys of the same stock (at least 6 months before) had beaten the S&P — known at the time, so no hindsight.
| Signal | vs S&P · 3 months | vs S&P · 6 months | vs S&P · 12 months |
|---|---|---|---|
| Buyers’ past buys beat the S&P | +1.3% median −0.2% · 49% beat · n=178 | +0.4% median +1.6% · 54% beat · n=155 | +8.1% median −1.3% · 48% beat · n=104 |
| Buyers’ past buys lagged the S&P | +2.2% median −0.3% · 48% beat · n=266 | −2.5% median −5.8% · 42% beat · n=237 | +0.8% median −6.9% · 42% beat · n=166 |
| No earlier buys on record | −0.5% median −4.1% · 42% beat · n=482 | −2.3% median −7.9% · 41% beat · n=445 | −11.0% median −17.3% · 36% beat · n=381 |
Live results: credibility scores and sell flags
Since September 28, 2026 we've logged 288 signals as the site showed them. None has reached its first 3-month mark yet. The first results are due around December 2026.
Every signal, one by one
The 300 most recent measured cluster buys with their returns against the S&P 500.
How we measure it
The signal. Three or more different insiders (officers, directors or 10% owners) buying their company’s stock on the open market within 60 days. Awards, option exercises and gifts don’t count. We count at most one signal per stock per quarter, so a long buying spree isn’t counted several times.
The return. We buy at the first close after the Form 4 that completed the cluster, which is the earliest an outside investor could have acted. We then compare the stock’s total return with SPY (the S&P 500, dividends included) over 3, 6 and 12 months. A signal counts only once its full horizon has passed. Averages cap each result at ±200% so a single micro-cap spike can’t carry the number. Medians are shown next to them for the same reason.
Live tracking. The backtest can’t know what our credibility score or unusual-selling flag would have said at the time, because both depend on SEC lookups we only started doing recently. So we also log each signal the day the site first shows it, including its credibility label, and measure it as it matures. Those results fill in over the coming months.
Limits. The history covers stocks listed today, so companies that were delisted or acquired are missing. That can flatter the results. Past performance doesn’t guarantee future returns, and none of this is investment advice.