Do Insider Buys Beat the Market? 63,000 European Deals, 120 Sessions Later
"Insiders sell for many reasons, but buy for only one." The line is usually credited to Peter Lynch, and it is the premise of most insider-trading strategies: when a director spends their own money on their company's shares, the shares should do well afterwards.
Most of the evidence for that comes from US Form 4 data. We looked at Europe instead, using every PDMR purchase in our archive with enough price history after it.
The data
- Deals: market purchases (
BUY) at a price above zero, made by the PDMR personally (closely associated persons excluded), from 17 national registers, not withdrawn. - Returns: the share's split- and dividend-adjusted return from the deal date to 20, 60 and 120 trading sessions later. 120 sessions is roughly six months.
- Sample: 63,525 purchase lines at 2,583 companies with a full 120-session window, dealt between November 2013 and April 2026. The headline table also counts deals whose shorter windows have closed, so its 20- and 60-session rows are slightly larger. Sells follow the same rules.
Headline result
| Sessions after | Median return | Share of deals followed by a rise | |
|---|---|---|---|
| Insider buys | 20 | +0.51% | 52.2% |
| 60 | +1.33% | 53.7% | |
| 120 | +1.92% | 54.1% | |
| Insider sells | 20 | 0.00% | 48.9% |
| 60 | +0.24% | 50.4% | |
| 120 | +1.18% | 52.4% |
Shares rose after insider buys more often than after insider sells, and by more: a median gap of about 0.7 percentage points after 120 sessions, and 1.1 points after 60.
But look at the sells row. Shares also rose after insiders sold. Markets went up over most of this period, so a raw return above zero is not an edge. The fair comparison is buys against sells, or against an index, not against zero.
The averages are higher than the medians (+5.56% mean after 120 sessions for buys): a few large winners pull the mean up. We quote medians because the typical deal is what you would actually get.
Not all buys are equal
The 63,525 purchases, split by who bought and when:
| Buys | Lines | Median, 60 sessions | Rose, 60 | Median, 120 sessions | Rose, 120 |
|---|---|---|---|---|---|
| Near the 52-week high (top 20% of range) | 12,265 | +2.73% | 59.5% | +3.47% | 58.8% |
| Part of a cluster (3+ insiders in 30 days) | 29,050 | +1.62% | 55.0% | +2.71% | 56.3% |
| Board members and other executives | 48,562 | +1.41% | 54.1% | +2.25% | 55.1% |
| All buys | 63,525 | +1.29% | 53.5% | +1.92% | 54.1% |
| Near the 52-week low (bottom 20% of range) | 21,872 | +0.70% | 51.6% | +0.04% | 50.1% |
| CEO or CFO | 14,963 | +0.68% | 51.5% | +0.47% | 50.7% |
| Only insider buying in 30 days | 22,457 | +0.66% | 51.5% | +0.64% | 51.0% |
Three things stand out.
Clusters beat lone buyers. When three or more insiders bought within 30 days, the median 120-session return was four times that of a lone buyer. This is the most consistent pattern in the table.
Buying near the high beat buying the dip. Insiders who bought near a 52-week high did best, and those who bought near the low were flat after six months. A director "buying the dip" is often catching a falling share, and the share keeps falling.
CEO and CFO buys did worse than other insiders' buys. This runs against intuition: the people closest to the numbers should know most. One possible reading is that chief executives often buy to send a signal, after bad news or a fall in the share price, while other insiders buy when they simply think the price is low. The data shows the outcome, not the motive.
What this does not show
- No market adjustment. These are raw returns, not returns relative to an index or sector, and the period was mostly a rising market.
- Lines, not decisions. A filing with ten executions counts ten times. Large, multi-execution deals weigh more.
- Overlapping groups. A CEO's purchase near the 52-week low inside a cluster sits in three rows.
- Survivorship. Companies that delisted have no price series and are missing, which flatters the results.
- Coverage. A little over half of all purchases have a usable price series with 120 elapsed sessions.
- No costs. Spreads and commissions aren't deducted, and an outsider only sees the deal after it is published, a business day or more later.
It is a description of the past across many markets and years, not a trading rule.
Reproduce it
Each transaction in the API carries a market block with the derived figures: premium to the close, position in the 52-week range, and the returns after 5, 20, 60 and 120 sessions, each null until its window has passed. coverage explains any gap.
curl "https://api.duceusapi.com/signals?preset=insider_buy&sort=recent&limit=100" \
-H "Authorization: Bearer YOUR_API_KEY"
"market": {
"coverage": "OK",
"pct_of_52w_range": 0.86,
"return_20d_pct": 2.4,
"return_60d_pct": 5.1,
"return_120d_pct": null
}
Per company, /issuers/{id}/insider-activity reports the average return after insiders' buys and sells there, and /persons/{id}/track-record does the same per person, both computed only from windows that have closed.
Summary
- Across 63,525 European insider purchases, shares rose a median 1.92% over the next 120 sessions, against 1.18% after insider sales.
- Cluster buys and buys near the 52-week high did best; dip buys and CEO/CFO buys did worst.
- None of it is market-adjusted, and all of it is history.