One insider buying is a data point. Several insiders buying the same company within a few weeks, a "cluster," is the pattern research has long tied to above-average returns. We tested it head-on: across 45,712 open-market SEC Form 4 purchases over five years, we grouped buys into clusters of 2, 3, 4 and 5 or more different insiders within 30 days, and measured what each stock did 30, 60 and 90 days later against the Russell 2000.
The win rate and the average return both climbed with each additional insider, monotonically. That kind of clean, ordered relationship is uncommon in market data, and it is the main reason cluster buying is one of the few insider signals we act on directly.
| Signal | Sample | Avg. 30d | Avg. 60d | Avg. 90d | % up (90d) | % beat Russell (90d) |
|---|---|---|---|---|---|---|
| 2+ insiders within 30 days | 4,860 | +4.1% | +4.3% | +5.3% | 53.2% | 47.2% |
| 3+ insiders | 129 | +0.2% | +4.0% | +5.9% | 60.5% | 51.9% |
| 4+ insiders | 57 | +0.9% | +7.8% | +8.7% | 70.2% | 59.6% |
| 5+ insiders small sample | 25 | +1.4% | +9.5% | +8.6% | 80.0% | 60.0% |
Stacking a cluster with one more condition sharpened it further. A CFO buying inside a cluster, or a 3-plus cluster forming near a 52-week low, both improved on the plain cluster.
| Signal | Sample | Avg. 30d | Avg. 60d | Avg. 90d | % up (90d) | % beat Russell (90d) |
|---|---|---|---|---|---|---|
| 3+ insider cluster | 129 | +0.2% | +4.0% | +5.9% | 60.5% | 51.9% |
| CFO buying in a cluster | 480 | +6.0% | +6.6% | +8.2% | 59.2% | 51.0% |
| 3+ cluster near the 52-wk low | 4,156 | +3.9% | +3.8% | +4.9% | 58.3% | 48.5% |
As in every slice of this data, the advantage is strongest in the first 30 to 90 days and fades after that. Insider buying reads like a short-to-medium-term catalyst, not a set-and-forget signal, which is why we track it live and mark every buy on the price chart.
We start from every open-market purchase (SEC Form 4, code P) of $10K or more filed over the last five years across the US market, enter at the closing price on or just after the transaction date, and measure the return 30, 60 and 90 days later against the Russell 2000 over the same dates. "Cluster" means that many different insiders bought the same company within 30 days. Option exercises, grants, and obvious price-data errors are excluded, as are tickers with no daily price history. Returns shown are averages (means); because a few big winners can lift an average, we also show how often each signal was simply positive and how often it beat the market. The 4-plus and especially 5-plus clusters are rare, so those rows are the strongest but the least certain. Caveats: this covers 2021-2026 only, one bull market plus the 2022 selloff, not a full range of cycles. This is analysis of past filings, not a prediction or investment advice. Past performance does not predict future results.
InsiderTape flags CFO buys, cluster buying, first buys in years, and buying near lows in real time, plotted on the price chart. Start a free 7-day trial, cancel anytime.
START FREE TRIAL →