The backtest establishes that a handful of relationship types — shared litigation, shared directors, ownership, procurement — anticipate designations with measurable recall and a multi-year median lead. The obvious next move is to run the same link types against the entities that are on no list today, and see who they surface. That is the forward watchlist.
The construction
We take the full designated cohort — 9,837 sanctioned parties across OFAC and the allied lists — and find every currently-unsanctioned entity our data links to one of them. The raw count is enormous: 706,709 links. Most of it is noise. A single shared court case is, far more often than not, ordinary SME debt collection — two companies in a payment dispute, no signal. So the raw number is not the product; the filtered queue is.
We qualify a candidate only when it clears one of two bars: two or more independent channels, or three or more distinct sanctioned ties. That leaves 35,632 qualified candidates. Of those, 12,096 are linked through two or more independent channel types, and 1,012 through three or more — the cases where several unrelated mechanisms point at the same entity, not a single thread. The list is ranked by channel breadth first, then by the count of distinct sanctioned counterparties, then by recency of the most recent link — so the top of the queue is entities tied to many sanctioned parties, through several independent mechanisms, recently.
What it adds over ownership screening
Most sanctions screening leans on the 50% Rule: if a sanctioned party owns a majority, the subsidiary inherits the exposure. That's necessary, but it's blind to everything that isn't an equity stake. 35,628 of our qualified candidates are behavioral — surfaced by litigation or directorship ties the 50% Rule alone would never catch. A company that shares a director with three sanctioned defense suppliers, or repeatedly litigates alongside them, is in the pre-designation relationship space whether or not anyone owns it. That behavioral layer is the part of the watchlist you can't get from a cap-table.
How to read it — and how not to
This is a recall-oriented triage queue, not a prediction. The backtest's median lead is roughly five years; many entities in the relationship space are never designated at all, and others only years out. We are not claiming these companies will be sanctioned. We are claiming something narrower and more defensible: these are the entities sitting in exactly the relationship structure that has, again and again, preceded designation — ranked so the densest cases are at the top.
Used the right way, that's an enhanced-due-diligence accelerant. If a counterparty turns up here with four channels and a dozen distinct sanctioned ties, that is not proof of anything — but it is a reason to look harder before the list catches up, not after. The backtest tells you how often "before" turns out to be real, and how much runway it has historically bought.
Cadence
The watchlist rebuilds with the pipeline. Entities enter as new links accrue and leave when they're designated (at which point they graduate into the backtest as a scored outcome). When the composition of the top of the queue shifts in a way worth noting — a cohort of newly multi-channel entities, a cluster around a single sanctioned holding — we'll write it up here.