BetaInformation only, not investment adviceData may be resetSend feedbackAbout the beta
Two locked filter bundles that narrow the feed to insider buys the published research says are worth a second look. Each is applied in one click in the Scanner and the Backtester and cannot be edited piece by piece — the bundle is the filter. They change nothing about how a signal is scored or tiered; they only decide what you see.
Insider buying is already a value screen: insiders buy when their own stock is cheap and beaten down (Rozeff & Zaman 1998; Jenter 2005; Piotroski & Roulstone 2005). So a “cheap” filter barely narrows an insider-buy feed — what separates the signals is why the stock is cheap.
The research is clear on which case carries more return: purchases in firms whose reported numbers are ambiguous or bad (low accruals quality, distress, deep drawdown) move prices more and predict better returns on average (Veenman 2012; Hill, Korczak & Wang 2026; Zhao 2026) than purchases in companies the market already agrees are solid. The two bundles are the two risk postures: Contrarian for the research-backed alpha case, Quality for the steadier, lower-dispersion case.
Basis: Seyhun (1986) on senior roles; Alldredge & Blank (2019) and Kang, Kim & Wang (2018) on clusters; Cohen, Malloy & Pomorski (2012) on opportunistic versus routine trades.
Top-role or cluster buys ≥30% below the 52-week high in small caps whose reported numbers look bad or ambiguous — where the literature finds the most insider-buying alpha. Higher dispersion.
Basis: Rozeff & Zaman (1998); Jenter (2005); Piotroski & Roulstone (2005) — insiders are contrarian value buyers; Veenman (2012) — purchase filings move prices most where earnings are least reliable; Hill, Korczak & Wang (2026) — insider buying in distressed firms predicts better returns on average; Zhao (2026) — distance from the 52-week high dominates microcap insider-purchase models.
Top-role or cluster buys ≥20% below the 52-week high in companies rated STRONG with a positive P/E under 25 and positive free cash flow — solid and cheap, insiders buying anyway. Steadier, less edge.
Basis: Piotroski (2000) — financial strength separates winners inside a cheap population; Cohen, Malloy & Pomorski (2012); Alldredge & Blank (2019) — opportunistic and clustered buys carry the signal; Lakonishok & Lee (2001); Jeng, Metrick & Zeckhauser (2003) — purchases are informative, sales are not.
Of 648 live open-market buys between 1 July and 25 August 2026, 45 fit Contrarian and 22 fit Quality. Over windows of 0–55 days, unadjusted, Contrarian averaged +6% with a +5% median and 58% positive but the subset with a closed 30-day window was −14% (three Polestar rows, Wallbox, BGDE), while Quality averaged +2% with 68% positive and no blow-ups. That is the dispersion the literature predicts. It is a handful of weeks of data and proves nothing; the bundles will be measured on the frozen corpus before either earns any weight in the score.
Nothing here is investment advice. Past performance — including anything measured on a research bundle — does not guarantee future returns.