Insiders Trades
We measure the ranking on a corrected backtest over 2022-2026Q1, against the STOXX Europe 600 on the same dates, dividends included on both sides and round-trip costs deducted. The live track record started on May 27, 2026. Its results are published below, month by month.
Each month, the simulated portfolio buys the 10 highest-scored insider BUY signals (EU markets) in equal weights, then sells each position 90 days later. One company often files several times, so these 10 signals cover 5.0 distinct companies on average. Result is net of 0.6% round-trip fees.
Average gap with the index: +2.7 points a year, 95% confidence interval -3.0 to +8.5 points. Without the ranking, all eligible insider purchases returned 0.1%/yr.
Simulation over 2022-2026Q1, each purchase made the day after the filing and sold 90 days later (the simulation ties up the capital for 91 days), dividends included for the ranking and the index alike, 0.6% round-trip cost, capital split across overlapping baskets. A monthly basket holds 5.0 distinct companies on average.
Backtest last measured on October 13, 2026 (code 4d9a1772) on 24,701 eligible purchases: entries from 2022-2026Q1, portfolio value followed over 54 months.
Every night, the top 10 BUY list served on the site is frozen with the last known price. Each company counts once per holding window, from the first day it was served, and its return is compared with its regional index over the same dates.
Under construction. The current construction (composite PIT frame) has been served since 16 Aug 2026; its first 90-day result is expected around 14 Nov 2026.
Same measurement as the track record table, restricted to companies that reached the first ranks. A company joins the group on the first day it appears in it, then its return is compared with its regional index over the same dates.
The interval contains zero: the gap cannot be told apart from chance.
Raw returns in local currency, before costs and without dividends, for the position and for the index. Entry: the last close known on the day the company joins the group. A company counts once per holding window. The top 10 is the full served list: its figures are the "All periods" row of the table.
Only months whose positions have all reached 30 days, with at least 10 positions, are plotted. The running month is in the monthly detail.
| Serving period | Horizon | Positions measured | Coverage | Beat the index (95% CI) | Mean gap vs index (95% CI) | Index, mean |
|---|---|---|---|---|---|---|
| V14e ladder, before the composite, until 3 Jul 2026provisional, fewer than 30 positions · 1 position excluded: symbol of another company | 30 days | 28 / 28 | 100.0% | 42.9% [26.5, 60.9] | -0.8 pt [-2.7, +1.3] · includes 0 | +1.8% |
| Composite, 4 Jul 2026 to 15 Aug 2026, 27 serve days | 30 days | 57 / 57 | 100.0% | 50.9% [38.3, 63.4] | +3.0 pt [-1.7, +7.7] · includes 0 | +1.7% |
| Composite PIT frame, 16 Aug 2026 to 17 Oct 2026, 24 serve daysprovisional, fewer than 30 positions | 30 days | 22 / 22 | 100.0% | 50.0% [30.7, 69.3] | -3.8 pt [-8.5, +0.8] · includes 0 | -1.6% |
| V14e fallback (composite unavailable), 19 Jul 2026 to 29 Sep 2026, 50 serve daysprovisional, fewer than 30 positions · 5 positions excluded: symbol of another company | 30 days | 23 / 25 | 92.0% | 52.2% [33.0, 70.8] | +0.5 pt [-7.5, +10.2] · includes 0 | -0.6% |
| All periods6 positions excluded: symbol of another company | 30 days | 130 / 132 | 98.5% | 49.2% [40.8, 57.7] | +0.6 pt [-2.2, +3.7] · includes 0 | +0.8% |
| V14e ladder, before the composite, until 3 Jul 2026provisional, fewer than 30 positions · 1 position excluded: symbol of another company | 90 days | 22 / 22 | 100.0% | 36.4% [19.7, 57.0] | -2.1 pt [-14.5, +7.1] · includes 0 | +3.3% |
| Composite, 4 Jul 2026 to 15 Aug 2026, 27 serve daysprovisional, fewer than 30 positions | 90 days | 11 / 12 | 91.7% | 45.5% [21.3, 72.0] | +1.3 pt [-10.6, +13.2] · includes 0 | -2.6% |
| Composite PIT frame, 16 Aug 2026 to 17 Oct 2026, 24 serve days | 90 days | Pending, first result around 14 Nov 2026 | ||||
| V14e fallback (composite unavailable), 19 Jul 2026 to 29 Sep 2026, 50 serve days4 positions excluded: symbol of another company | 90 days | Pending, first result around 17 Oct 2026 | ||||
| All periods5 positions excluded: symbol of another company | 90 days | 33 / 34 | 97.1% | 39.4% [24.7, 56.3] | -1.0 pt [-9.8, +5.1] · includes 0 | +1.3% |
Small samples: these figures will move. When the confidence interval of the gap contains zero, the difference with the index cannot be told apart from chance. Raw returns in local currency, before fees and dividends; entry is the last close known when the pick was served. A position with no usable price is excluded and counted in the coverage.
Interval of the mean gap: bootstrap by entry day (5000 resamples, seed 42). Positions entered on the same day ride the same market, so they are resampled together, as one block. With fewer than 10 entry days, the wider of this interval and the position-by-position bootstrap is shown.
58 snapshot rows excluded from the statistics: symbol of another company.
Comparison indices: United States: S&P 500 (^GSPC) · Europe: STOXX Europe 600 (^STOXX) · Hong Kong: Hang Seng (^HSI) · Canada: S&P/TSX Composite (^GSPTSE)
| Month | Horizon | Positions measured | Coverage | Beat the index (95% CI) | Mean gap vs index (95% CI) | Index, mean |
|---|---|---|---|---|---|---|
| May 2026provisional, fewer than 30 positions · 1 position excluded: symbol of another company | 30 days | 13 / 13 | 100.0% | 15.4% [4.3, 42.2] | -2.5 pt [-6.5, +1.0] · includes 0 | +1.4% |
| June 2026provisional, fewer than 30 positions | 30 days | 12 / 12 | 100.0% | 58.3% [31.9, 80.7] | -0.4 pt [-4.0, +2.7] · includes 0 | +2.4% |
| July 20263 positions excluded: symbol of another company | 30 days | 58 / 59 | 98.3% | 51.7% [39.2, 64.1] | +2.4 pt [-1.9, +7.1] · includes 0 | +2.2% |
| August 2026 | 30 days | 43 / 43 | 100.0% | 53.5% [38.9, 67.5] | +0.7 pt [-4.1, +6.8] · includes 0 | -1.6% |
| September 202613 positions still running · 2 positions excluded: symbol of another company | 30 days | 4 / 5 | 80.0% | Fewer than 10 positions measured: no statistic published | ||
Computed on 17 Oct 2026 from 2576 daily snapshot rows (150 distinct companies on the BUY side). Raw data (API)
An executive who buys shares in their own company is putting personal money on it. Every day we read the insider filings published in 37 markets, convert the amounts to euros and score each eligible open-market purchase from 0 to 100 against purchases in the same market. A position size is suggested only on the venues of the backtested strategy universe. Elsewhere, signals stay on watch with no suggested size.
How the returns are produced
Historical backtest, not a live return or investment advice. Detailed methodology ↗ · Backtest explorer
Every on-market purchase filed by an executive (at least €1,000, filing parsed) published from 2 Jan 2023 to 9 Oct 2025 whose one-year price is known: 33,803 filings across 4,879 companies. No case is hand-picked.
Return of the stock alone, in local currency, from the close of the day after publication to one year later, without dividends or costs and without comparison to an index. 291 of 34,094 rows are excluded because they show more than +300% or a total loss, a likely price error. Computed on 17 Oct 2026.
This is neither the ranking nor a portfolio: it is the unsorted starting point. The ranking itself is measured above, against the index over the same dates.
The rules set a short deadline: 3 business days in the EU (MAR, article 19) and 2 business days in the US (Form 4). We measure the gap in calendar days between the transaction date and the publication date, on 93,074 filed purchases with a computed 90-day return (gaps between 0 and 90 days, computed on 17 Oct 2026).
Half of the purchases are published less than 1.5 days after the trade. A signal shown on the site is therefore a few days old: the earliest price a reader can get is the close after publication, not the insider's price. Calendar days count weekends, so this share is not a compliance rate.
Insiders Trades reduces a global public-filings dataset to a watchlist worth examining. The ranking prioritises research. It does not give an expected return.