How the signal score works
Every insider buy gets a score from 0 to 100. It is a summary of what the filing says, built from published research on which insider buys have tended to matter. It is not a prediction, and a score is never rewritten once it is on the record.
The factors
The positive factors add up to 100 at most; the penalties take points off. Each one is shown on the buy's card, with its points, so you can see why a buy scored what it did.
| Factor | Points | Why | Evidence |
|---|---|---|---|
| Who bought | +30 | CFO buys have beaten CEO buys by about 5 points over 12 months; officers beat directors; a 10% owner says the least. | Wang, Shin & Francis, JFQA 2012 |
| Change in their stake | +18 | A brand-new position, or a buy that grows the stake a lot, says more than a small top-up. | Lakonishok & Lee, RFS 2001 |
| Size of the buy | +12 | Bigger buys carry more conviction, with diminishing returns past $1M. | Jeng, Metrick & Zeckhauser, NBER |
| Routine or opportunistic | -12 to +10 | Buys made in the same calendar month as an earlier year's buy are routine and earn about nothing. An insider's first buy in a long while is the strongest kind. | Cohen, Malloy & Pomorski, Journal of Finance 2012 |
| Other insiders buying | +22 | Two or more insiders buying the same stock within days is followed by more than 2% abnormal return the next month. | Alldredge & Blank, Journal of Financial Research 2019 |
| Where the price sits in its 52-week range | +8 | Buys made far above the 52-week low have beaten buys near it by 3.3 points over the next 30 days: insiders buying a stock that isn't on sale know something. | Li, Wang, Yan & Zhang, Journal of Portfolio Management 2019 |
| Pre-arranged plan | -12 to 0 | A Rule 10b5-1 plan set the timing in advance (the box on the Form 4 since April 2023), so the buy says little about what the insider thinks now. | SEC, Rule 10b5-1 amendments 2023 |
| Filed late | -15 to 0 | A filing after the two-business-day deadline is a hidden deal: worth hearing about, but the market may have moved and a late filer is a weaker tell. | Cheng, Nagar & Rajan, RFS 2007 |
| Fund or investment firm | -8 to 0 | A fund buys for its mandate, not from what it knows about the company. | InsiderBeacon's own rule |
| Hard to trade | -6 to 0 | Under $2 a share, spreads and price impact eat most of the edge, and a tiny company's moves are mostly noise. Flagged and a few points off, never hidden. | Zhao, arXiv 2026 |
What the score does not know
- Market cap and trading volume. SEC filings don't carry them and the app doesn't buy that data yet, so "hard to trade" is judged from the share price alone (under $2). Very small companies with a higher share price aren't flagged.
- Routine buyers beyond the filings on record. The research defines a routine insider by three years of same-month buys. The app checks the filings it has stored (the method page says how far back that goes), so a buyer is called routine as soon as one earlier same-month buy is on record, and "first buy we've seen" means first in the stored window.
- The 52-week range needs a year of prices for the stock. When they aren't stored yet, that factor scores 0 and the card says so.
- What happens next. The published edges are tens of basis points a month before costs, concentrated in small companies, and about half of the price reaction comes on the first trading session after the filing. Nothing here is investment advice.
What happened after each kind of buy
Measured from InsiderBeacon's own track record: for every buy scored with this method, the price change 1 week and 1 month after the first close that followed the filing, split by whether the factor counted. 1,355 buys have this score so far. Buckets marked small have fewer than 20 measured buys, so their averages mean little yet.
| Factor | When it counted: buys · 1 week · 1 month · up after a month | When it didn't: buys · 1 week · 1 month · up after a month | ||||||
|---|---|---|---|---|---|---|---|---|
| Who bought | 315 | −1.0% | −5.4% | 36% | 0 | — | — | — |
| Change in their stake | 210 | −1.2% | −6.2% | 35% | 105 | −0.7% | −3.8% | 38% |
| Size of the buy | 315 | −1.0% | −5.4% | 36% | 0 | — | — | — |
| Routine or opportunistic | 315 | −1.0% | −5.4% | 36% | 0 | — | — | — |
| Other insiders buying | 153 | −1.4% | −4.3% | 41% | 162 | −0.8% | −6.5% | 31% |
| Where the price sits in its 52-week range | 126 | −1.9% | −3.7% | 39% | 189 | −0.5% | −6.5% | 34% |
| Pre-arranged plan | 11 small | +0.6% | +1.7% | 64% | 304 | −1.1% | −5.7% | 35% |
| Filed late | 22 | −0.2% | −3.9% | 36% | 293 | −1.1% | −5.5% | 36% |
| Fund or investment firm | 60 | −0.1% | −3.3% | 40% | 255 | −1.3% | −5.9% | 35% |
| Hard to trade | 31 | −0.7% | −13.4% | 13% | 284 | −1.1% | −4.6% | 38% |
Past results don't predict future ones. Averages are simple means of the price change in the buy's direction, before costs; a few big movers can swing a small bucket.