"Is this insider buy a big deal?" is usually asked against the wrong reference point — the stock chart, or other companies. The comparison the data actually supports is narrower and more useful: this purchase versus that insider's own history in that same stock.
Put in the numbers and see which tests fire:
- First open-market purchase on record. Not marked as a first purchase.
- Much larger than their own previous biggest. This buy is 7.5× their previous largest. Threshold is 3×.
- Buying below their own last purchase price. -22.1% vs their last buy. Flagged at -15% or lower.
- Materially increases their own stake. Raises their holding by 333%. The distribution feed uses 50%+.
“Unusual” here means unusual for this person at this company — not a prediction, and not a comparison to the market. All four tests can fire on a purchase that still goes nowhere.
The four tests, and why each one
First buy on record. An insider who has only ever received grants, suddenly buying on the open market, has changed behaviour. The caveat is history depth — "first buy" only means something if there is enough history behind it, which is why a thin record stays silent rather than claiming a first.
Size against their own previous largest. Insiders vary their sizing constantly, so the threshold is deliberately high: three times their previous biggest purchase before it counts as a change rather than a routine top-up. A test that fired on half of all buys would train you to ignore it.
Price against their own last purchase. Buying more than 15% below the price they last paid — buying into their own drawdown — is a distinct posture from averaging up. Under 15% is inside the ordinary volatility of the weeks between two filings.
Change in their own stake. A purchase that lifts an insider's holding by 50% or more is the difference between "a director bought some stock" and "a director materially increased their bet". This is the cut the site's distribution feed uses to decide an event is worth posting.
What the framework does not do
It does not predict returns. It does not account for what the insider knows. It does not weigh the buy against the company's fundamentals. All four tests can fire on a purchase that precedes a further 40% fall — insiders have an information edge on the business, not on the price.
What it does is separate the routine from the notable, so that the purchases worth an hour of your research are not buried under the ones that are just an executive topping up.
The short version
Judge an insider buy against that insider's own record in that stock: first buy, unusually large, below their last price, materially bigger stake. The more tests that fire, the more it stands out — for that person. Standing out is not the same as being right.
Related: is insider buying a bullish signal? · what is a cluster buy? · test your signal-reading