Insiders EU
About this strategy: the rule, its costs and its limits
Every Monday this strategy looks at open-market share purchases reported by directors and other senior insiders in Sweden, the Netherlands, Spain and Belgium. A purchase counts only if it is worth at least €25,000 and the same insider had not bought shares in that company in the two years before. For every company it adds up those purchases filed in the preceding 180 days, and it holds the ten with the largest total in equal weights — among companies whose shares had a median daily turnover of at least €100,000 over the 20 trading days before, and a last price of at least €1. A company that enters is bought on the next trading day; one that drops out is sold; one that stays is left alone. Filings count from the day they were published, and only if they were public before that Monday.
It starts at USD 100,000 on Jul 4, 2022 — the day the rule first had something to act on — and the S&P 500 stands next to it from the same amount on the same day. Each trade costs 0.1% one way, 0.2% round trip.
What it does not do
- It is long-only. There is no short leg and no leverage; a company leaving the ten is sold, never sold short.
- The backtest does not prove it. This rule was one of ten variants tried after a first test found no winner; it was the only one that held up in a later control period. A further check on 2018–2022 data, with its test fixed in advance, came out in its favor, but in those years the Swedish register did not record what kind of instrument was bought, so the two-year silence could not be checked there before September 2020 — that part of the check measured a looser rule.
- Not every country counts from the start. A purchase only counts once the register behind it covers the two years before it: Swedish and Dutch purchases count throughout, Spanish ones from 2025, and Belgian ones not before November 2026.
- Size ranks, it does not weight. A larger purchase can push a company into the ten, but inside the ten every position is the same size.
- The universe is today's survivors. Companies that disappeared without ever getting a ticker are missing, which flatters the curve.
- The Dutch notification date is mostly an assumption: the regulator does not publish when a filing arrived, so we use the transaction date plus three working days, the latest moment the law allows.
- Prices are converted. The portfolio is measured in dollars, so a krona or euro price is converted at the ECB rate of that day — part of the return is the currency, not the company.
- The portfolio uses adjusted closes, so dividends are in it; the Stoxx Europe 600 line beside it is a price index without dividends.
Every figure here is a calculation over filed forms, made afterwards, under assumptions that are written down. A backtest says what a rule would have produced on filings that were already public. The definitions are on the methodology page.
The numbers
Return
Change in portfolio value over the last trading day.
Change in portfolio value over the last 30 calendar days.
Change in portfolio value over the last 12 months.
The yearly growth rate that would produce the total return over the whole period.
Mean result of positions that gained, and of those that lost.
Share of all positions that ended above what they cost. A position at exactly zero counts in the denominator, not in the numerator.
Every buy and sell the rule placed, entries plus exits.
Risk
How much daily returns swing, scaled to a year.
Largest fall from a peak to a later low.
How strongly the portfolio moves with the benchmark; 1 means one-for-one.
Relative
Yearly return beyond what the portfolio's beta to the benchmark explains.
Return per unit of volatility, with a risk-free rate of 0.
Excess return over the benchmark per unit of tracking error.
Return per unit of beta.
- Every position
- Which insiders bought or sold it
- Rebalance log