Cluster Insider Buying: How to Tell Real Conviction from Share-Plan Noise

The cluster buy is the strongest variant of the strongest signal in markets. The research puts clustered insider purchases at roughly twice the abnormal return of solitary ones — multiple people with inside visibility independently reaching for their wallets in the same month is hard to fake. So when our 60-day scan of the Nordic register surfaced four companies with eight or more distinct insider buyers each, it looked like a feast: Skanska (13 buyers), Securitas (8), Tele2 (9), Bilia (8).

Then we read the filings, and three quarters of the feast evaporated. What is left is one genuinely exceptional conviction story, one decent one, and a free education in how Nordic disclosure data lies to screens. All four companies score well on our framework — this is not an exposé, it is a field guide.

Exhibit A: thirteen Skanska insiders pay 255.580857 kronor — all of them, to six decimals

On 14 April, thirteen Skanska insiders — the CEO, the CFO, board members, employee representatives — all filed purchases. Same day. And every single one of them paid exactly SEK 255.580857 per share. On 12 May it happened again: eleven of the same names, every one at exactly SEK 250.9979.

Thirteen people do not independently negotiate a price to six decimal places. That number is a programme VWAP — the averaged execution price of a company share-savings or incentive plan that buys for all participants in one block and allocates pro-rata. The monthly rhythm (mid-April, mid-May) confirms it. Securitas shows the same fingerprint: eight executives on 12 May, every one at exactly SEK 151.68. These filings are real, legal, and correctly disclosed as acquisitions (“Förvärv”). They are also not decisions. Nobody looked at the price that morning and chose to buy; payroll did.

Exhibit B: Tele2’s cluster, two-thirds grants

Tele2’s nine April-May buyers split cleanly when you look at the price column: six of the nine filings carry a price of zero — long-term-incentive allocations recorded as acquisitions. But the remaining three are the real thing: board member Nicholas Ove Högberg bought 12,000 shares at SEK 175.275 on 28 May (about SEK 2.1 million) and another 1,000 the next day, and Karin Wadström Sjöstedt bought 4,500 shares at SEK 185.35 on 22 May (about SEK 834,000). Two insiders, three purchases, real cash, at market prices, at a company carrying a StockRank of 91.7. A decent signal — once two-thirds of its apparent strength is correctly thrown away.

Exhibit C: the real thing at Bilia — a chairman averaging UP

Which brings us to the one that survives every filter. Bilia (BILI-A.ST) is the Nordic region’s largest car dealership and service group, a USD 1.4 billion company with a StockRank of 86 and a Strong Buy composite signal. Its chairman, Mats Qviberg — one of Sweden’s best-known investors — made three open-market purchases in four weeks:

6 May: 100,000 shares at SEK 132.20 — SEK 13.2 million.
20 May: 50,000 shares at SEK 133.97 — SEK 6.7 million.
2 June: 40,000 shares at SEK 139.35 — SEK 5.6 million.

That is roughly SEK 25.5 million — about USD 2.8 million — of personal capital, deployed in three tranches, each at a higher price than the last. Averaging up is the detail worth dwelling on: it means each successive purchase was made after the market had already moved his way, and he wanted more anyway. The sizes vary, the dates vary, the prices are market prices. This is what discretion looks like in filing data.

Around the chairman, six more Bilia insiders — the CFO and five board members — filed smaller purchases on 12 May (between 225 and 1,500 shares each). Honesty requires the caveat we have been applying all article: those six share an identical four-decimal price, SEK 135.8837, so they carry the programme fingerprint, not the conviction one. The Bilia story is Qviberg, and Qviberg is enough.

The field guide

Three tests, in order, before you believe any insider cluster:

1. The price-precision test. Multiple insiders at one price quoted to four-plus decimals on the same day = programme VWAP. Real buyers hit the market at different moments and get different fills.
2. The zero test. Price 0.00 = grant, vesting, or allocation. Delete the row.
3. The cadence test. Same names, same mid-month date, every month = savings plan on payroll rhythm. One-off dates that track news flow and weakness = humans deciding.

What survives all three — varied dates, varied sizes, market prices, ideally escalating — is the cluster the academic literature actually measured. It is rarer than the raw screens suggest, which is exactly why it is worth finding.

The caveats

One insider can be wrong, even a famous one — Qviberg’s SEK 25.5 million is conviction, not a guarantee, and as Bilia’s chairman his incentives to support the stock are not purely informational. Sixty days is a snapshot. Programme purchases, while informationally empty, are not bearish — they are simply silence. And as always: filed facts plus a framework are where the work starts, not investment advice.

Frequently asked questions

What is a cluster buy?

Multiple insiders at one company buying within a short window. Research finds clustered discretionary purchases produce roughly double the abnormal returns of solitary ones.

How common is fake clustering?

In our 60-day Nordic scan, three of the four largest apparent clusters were share-plan artefacts. Expect most raw clusters to fail the price-precision, zero, or cadence test.

Why does averaging up matter?

Buying more after the price has already risen means the insider is not bottom-fishing or supporting a falling stock — they expect the move to continue. It is the most expensive, and therefore most credible, way to add.