Buys divided by buys plus sells, by filing count. Ratios survive coverage changes: when a pipeline suddenly ingests more filings, both sides scale together.
35% of the compositeInsider Confidence Index
A monthly 0-100 reading of insider buying pressure, per market and aggregated for Europe. Built from regulatory filings only: open-market purchases and sales by executives and directors, measured against each market's own history. It describes behaviour; it does not predict returns.
Europe, latest reading
Equal-weighted across European markets, June 2026
10 markets in the aggregate
-3.6 pts vs previous month
50 = in line with history. Above 50, insiders lean toward buying; below 50, toward selling.
Latest published month per market, ranked by index value. A market enters only when its filing feed is deep and continuous enough; nothing is interpolated.
| Market | Index | Percentile | Buys | Sells | Net flow | Month |
|---|---|---|---|---|---|---|
| KR | 83.8 | 99th | 1,849 | 403 | €106.9m | June 2026 |
| INfeed paused | 72.6 | 96th | 141 | 95 | €54.2m | April 2026 |
| EEfeed paused | 67.5 | 92th | 19 | 5 | €2.6m | September 2025 |
| CH | 64.0 | 83th | 102 | 73 | €35.4m | June 2026 |
| UK | 57.5 | 74th | 885 | 150 | €71.0m | June 2026 |
| FI | 56.6 | 55th | 1,416 | 293 | -€341.7m | June 2026 |
| CN | 55.2 | 73th | 151 | 198 | -€545.8m | June 2026 |
| NO | 54.7 | 66th | 33 | 14 | €12.7m | June 2026 |
| IT | 54.2 | 67th | 35 | 74 | -€54.5m | June 2026 |
| BE | 49.0 | 36th | 75 | 21 | €21.3m | June 2026 |
| BR | 47.0 | 47th | 517 | 446 | -€1.4bn | June 2026 |
| FR | 46.9 | 39th | 233 | 122 | €290.7m | June 2026 |
| ESfeed paused | 46.8 | 34th | 82 | 13 | -€1.3bn | May 2026 |
| NL | 43.4 | 27th | 58 | 16 | -€43.1m | June 2026 |
| DE | 37.9 | 18th | 136 | 48 | €19.3m | June 2026 |
| DK | 32.7 | 11th | 5 | 27 | -€349.3m | June 2026 |
| US | 7.4 | 1th | 1,048 | 6,933 | -€10.4bn | June 2026 |
Methodology, in full
Buys divided by buys plus sells, by filing count. Ratios survive coverage changes: when a pipeline suddenly ingests more filings, both sides scale together.
35% of the compositeBuys divided by buys plus sells, in reported EUR volume. This money-weighted view is sensitive to very large transactions. It is dropped for a month when too few filings carry a usable, non-estimated amount.
25% of the compositeCompanies with at least one insider buy, divided by companies with any classified insider trade that month. Separates broad-based buying from one insider loading up.
25% of the compositeShare of buys where two or more distinct insiders bought the same company within 30 days. The strongest conviction pattern in the literature; weighted lowest here because the historical flag has known gaps.
15% of the compositeEach component is converted to a z-score against the market's own published history, winsorized at ±2.5 standard deviations. The composite z is the weighted mean of available components, and the index is 50 + 20 × z, clamped to 0-100.
A month counts only if it carries at least 10 direction-classified trades. A market publishes only over runs of at least 12 consecutive qualifying months. Young feeds, gappy feeds and dead feeds simply produce missing months.
Real markets never see one side of the ledger vanish overnight. When a market's buy or sell count collapses to near zero against its own trailing median, that month is withheld and flagged as a data issue rather than published as a fake extreme. Held months are listed in the dataset.
Europe is the equal-weighted mean of its member markets' scores that month, published only when at least 5 European markets qualify. Equal weighting is deliberate: a volume-weighted version would be a UK-and-France index in disguise, and a raw-sum version would drift upward every time coverage improves.
Only explicit open-market purchase and sale labels, in each regulator's own vocabulary, classify a filing. Grants, option exercises, tax withholdings, subscriptions and holdings updates are excluded: they are compensation mechanics, not conviction. Markets whose filings carry no clean direction, such as Japan's holdings-based reports, stay out entirely.
Scores are normalized against the market's full published history and recomputed monthly. Past values can move as history grows, filings are corrected, feeds are backfilled or quality quarantines change. Every download carries its generation date.
The full series, every market, every component, one file. Updated monthly. Free for any use with attribution.
Reference it as "Insiders Trades Sigma, Insider Confidence Index" and link to this page. The dataset carries a CC BY 4.0 license: reuse it, chart it, publish it, with attribution.
Browse citable facts