Analyst Upside

The book the strategy actually owns right now, by reporting currency, in % of book (no dollar amounts). It is signal-driven and rebalanced daily, but each name is held until its signal says exit (hold-until-exit, anti-churn) — so turnover stays low. A book can hold fewer names than the model ranks: positions are sized by rule — a volatility target, whole-share lots, per-name caps — with the remainder in cash (the Cash badge).

What the strategy would pick if it rebalanced today.

The model ranks these names by signal daily; we hold the names we own until a name's signal says exit — the gap between this ranking and the Holdings above is our hold-until-exit (anti-churn) discipline. The ranking itself is the same for both tranches — the Aggressive/Conservative toggle only changes which signals trigger a trade, so it drives the Holdings and curve above, not this list. So this is a research signal, not a confirmed buy list.

How Analyst Upside picks names.

Buys names where multiple independent signals agree the consensus is too low. EPS estimates trending up, insiders putting their own money in, sell-side targets sitting well above spot. Those signals rarely align without reason.

1. 01

   Universe filter

   US, European, and Polish listings with sufficient analyst coverage (target prices, ratings available) and a per-market size floor; for non-US listings, analyst data falls back to a secondary provider where the primary feed is blank. Hard floors run before scoring — at least 3 covering analysts and a consensus target at or above the current price — and they fail closed: a name with missing data is dropped. The V1.2 investability screen caps the share price at $600 (US) / €1,000 (EU) / 4,000 zł (PL), so every pick is buyable as a whole share within the strategy's fixed entry ticket. Under the current rules the cap also applies to held names: a position whose price outgrows it leaves the ranking and is sold.

2. 02

   EPS revisions (30d)

   Two separate signals. The dollar change in the consensus EPS estimate over the past 30 days (capped to filter corporate-action artifacts) is the highest-weighted factor in the composite. The balance of up- versus down-revisions is scored separately and also acts as a hard floor: names with net downgrades are dropped.

3. 03

   Insider activity (90d)

   The more abnormal of two signals, each measured as a z-score: total open-market insider purchase dollars over 90 days, or the number of distinct insider buyers over the same window. Only open-market purchases (SEC Form 4 code P) count — sales, option exercises, and grants are not part of the factor.

4. 04

   Analyst upside

   Spread between consensus 12-month target and current price. Premium for names with high spread AND recent EPS upgrades.

5. 05

   Momentum acceleration

   Difference between recent 3-month and 12-month momentum — a tilt toward names whose momentum is improving. Under review: interim measurements show this factor measuring adverse — negative, not just flat — on this strategy's own universe; its demotion is pre-registered for the ~Sep 2026 walk-forward.

6. 06

   Composite score + AI research

   All six factors — the five above plus a consensus analyst-rating score — are winsorised, z-scored, and weighted into a composite. The factor set and weights are frozen and pre-registered until the ~Sep 2026 walk-forward; they never change quietly. The top-ranked names are selected per market. Each pick gets the same AI research card as QM (Perplexity brief + 3-model council vote) before order placement. On the rare day the research service is unavailable, the paper book trades on the quant ranking alone and the missing verdict is visible on the pick page.

Rebalance cadence

Signal-driven, rebalanced daily, hold-until-exit (anti-churn). A credible backtest accumulates through ~Sept 2026.

Universe

US, European, and Polish mid-to-large-cap equity with analyst coverage