Do Insider Buying Alerts Beat the Market? We Tested Ours Against 4 Benchmarks
Key findings
- Over 60 trading days (about 90 calendar days), the 27 ClusterSignal alerts with full data gained a median +14.5%, and 85% were positive.
- That beat the S&P 500 by a median +11.1% (95% interval +3.7% to +20.5%), a clear result.
- Against the Russell 2000 the edge was +9.8%, but the interval (0.0% to +19.0%) touches zero.
- Against similar stocks with no insider buying the edge was +9.5%, and against plain insider buys from the same week it was +5.2%. The second one is not statistically proven (interval -0.2% to +23.0%).
- Peak returns look much bigger (+26.9% at 60 days) but flatter any single stock. The fair peak comparison is against similar stocks: +4.8%.
- This is a small sample from one six-month period. Treat it as an early result, not a proof.
Do insider buying alerts beat the market? It is the first question a sensible investor asks, and most services answer it with a flattering chart. We tried to answer it the hard way: by testing the alerts shown on our public Results page against four different benchmarks, and by reporting the cases where the edge is thin or unproven.
How we measured
- Which alerts: the BUY alerts shown on our Results page, 42 in all. Of these, 27 have complete 90-day data, and all 27 were sent between March and June 2026. For the peak comparison at about 30 days, 32 alerts have complete data.
- Entry: the next regular-session market open after the alert date, not the price stored at alert time. (The stored price was slightly different from the real next open: a median -0.5% after the close and -1.8% before the open, so we used the real open.)
- Horizons: 21 trading days (about 30 calendar days) and 60 trading days (about 90 calendar days).
- Two ways to measure: the price on day 60, which is what an investor would actually have held, and the peak, the highest close inside the window divided by the entry open.
- Four benchmarks: (1) the S&P 500, (2) the Russell 2000 small-cap index, (3) similar stocks with no insider buying, matched on price, trading volume and distance from the 52-week high, and (4) plain insider buys from the same week as listed on OpenInsider, with no scoring applied.
- Reporting: medians, with a 95% interval for each edge. If the interval includes zero, we say the edge is not proven.
The alerts and their returns are public on the Results page. The benchmark comparisons are our own analysis and are not shown there.
Result 1: what an investor would have held at day 60
This is the cleanest measure. We bought at the next open and looked at the price 60 trading days later, for the 27 alerts that have reached that point.
| Compared with | Median edge | 95% interval | Read |
|---|---|---|---|
| Alert return, raw | +14.5% | +9.2% to +22.3% | 85% of alerts positive |
| S&P 500 | +11.1% | +3.7% to +20.5% | Clear |
| Russell 2000 | +9.8% | 0.0% to +19.0% | Borderline, low end touches zero |
| Similar stocks, no insider buying | +9.5% | +1.7% to +30.4% | Positive, but a wide range |
| Plain insider buys, same week | +5.2% | -0.2% to +23.0% | Not proven |
| S&P 500, one alert per ticker | +9.9% | -1.5% to +21.1% | Not proven once repeat alerts on one stock are removed |
Two things stand out. First, the alerts beat the S&P 500 by a clear margin. Second, the edge shrinks as the benchmark gets closer to what the alerts actually are, a set of insider-bought stocks. Against plain insider buys the interval includes zero, so with 27 alerts we cannot say the scoring adds an edge over simply following insider purchases. We can only say the direction is positive.
Result 2: peak returns, and why they flatter
Peak is the best close in the window. Nobody can know it in advance, so it is not a return an investor can bank. We report it because it is how many alert services show results, and because it shows a trap.
| Window | Alert peak | S&P 500 | Russell 2000 | Edge vs S&P 500 | Edge vs Russell 2000 | Edge vs similar stocks | Edge vs plain insider buys |
|---|---|---|---|---|---|---|---|
| 21 trading days (~30 calendar), n = 32 | +9.3% | +1.7% | +4.4% | +8.3% [+4.7, +13.0] | +5.1% [+2.3, +11.1] | +1.2% [-1.7, +2.9] | +0.8% [-3.4, +5.7] |
| 60 trading days (~90 calendar), n = 27 | +26.9% | +4.2% | +6.5% | +21.5% [+17.1, +37.4] | +18.4% [+14.8, +34.4] | +4.8% [+0.5, +13.0] | +3.7% [-1.6, +19.5] |
Against the two indexes, the peak gap looks enormous: +21.5% over the S&P 500 at 60 days. Do not read that as skill. A single stock is far more volatile than an index, so its best close in any window is always much higher. Similar stocks with no insider buying also reach a peak of about +17.9% over 60 trading days. A peak versus an index mostly measures volatility.
The fair peak comparison is against similar stocks. There the alerts' edge falls to +4.8% at 60 days (interval +0.5% to +13.0%) and to +1.2% at 30 days, where the interval includes zero. Against plain insider buys, neither peak window is proven.
Where the evidence is solid, and where it is not
- Solid: a clear edge over the S&P 500 at day 60, and a positive edge over similar stocks at day 60 (interval above zero).
- Borderline: the edge over the Russell 2000 at day 60.
- Not proven: any edge over plain OpenInsider buys, at either horizon and on either measure.
Limits you should know about
- Small sample. 27 alerts at 90 days. With a sample this size, only large edges can be separated from chance.
- One six-month period. All 27 alerts were sent between March and June 2026, in one market regime.
- Recent alerts are behind. Alerts sent since July do not have 90 days of data yet. At about 20 trading days, the 6 alerts from July to September were negative, with a median edge against the S&P 500 between -2% and -7%.
- Repeat alerts. When we keep one alert per ticker, the edge over the S&P 500 is +9.9% with an interval of -1.5% to +21.1%, so part of the result depends on a few stocks.
- Re-check. We plan to repeat this test around January 2027, when the alerts from July to October have their 90-day data.
Frequently asked questions
Do insider buying alerts beat the market?
In our test, yes. The alerts on our Results page beat the S&P 500 by a median of 11.1 percentage points over 60 trading days (about 90 calendar days), with a 95% interval of +3.7% to +20.5%. Against more demanding benchmarks the edge is smaller: +9.5% vs similar stocks with no insider buying and +5.2% vs plain insider buys from the same week, and the latter is not statistically proven. The sample is 27 alerts from March to June 2026.
Do insider buying alerts beat the Russell 2000?
Probably, but it is borderline. At day 60 the median edge vs the Russell 2000 (IWM) was +9.8%, but the 95% interval runs from 0.0% to +19.0%, so the low end touches zero. Measured at each alert's peak, the edge vs the Russell 2000 was clearer, but a peak comparison against an index flatters any single stock.
Is the edge over plain insider buys statistically proven?
No. Against plain OpenInsider buys from the same week (no scoring), the median edge at day 60 was +5.2% with a 95% interval of -0.2% to +23.0%. The interval includes zero, so with 27 alerts we cannot rule out that the edge is zero.
Why does the peak return look so much better than the day-60 return?
A peak is the best close in the window, which nobody can know in advance, and any single stock has a much higher peak than an index because it is more volatile. Similar stocks with no insider buying also reach a peak of about +17.9% over 60 trading days. The fair peak comparison is against similar stocks, where the alerts' edge was +4.8%.
How reliable is this result?
It is early. The 90-day results cover 27 alerts from one six-month period (March to June 2026). Alerts sent since July do not have 90 days of data yet, and at about 20 trading days the 6 July to September alerts were behind the S&P 500. We plan to re-check around January 2027.
Is this investment advice?
No. ClusterSignal is an information service. This analysis describes past alerts and does not tell anyone to buy or sell a security. Past performance does not guarantee future results.
Related research
If you are wondering whether the simpler strategy of copying one insider works, we tested that too across 102,083 insider purchases: Should you copy a single insider's trades? For the mechanics behind these alerts, see what a Form 4 filing is.
⚠ ClusterSignal is a financial information service. This analysis and the alerts it describes are for informational purposes only and do not constitute investment advice. Past performance does not guarantee future results. Consult a licensed financial advisor before making investment decisions.
Check the alerts behind these numbers
Every resolved alert, including the losers, is public.
View the Results page →