Express Entry draws & predictions

Cut-off model

For each category, nine candidate models compete: carry-forward, medians and exponentially weighted averages of recent cut-offs, robust (Theil–Sen) trend extrapolations, a pool-inversion model — IRCC's published score distribution tells us how many candidates sit above any score; past draws tell us what share of them belongs to this category; inverting that against the planned draw size yields a cut-off — and blends of the two families.

Every candidate is walk-forward backtested: for each 2025–26 draw we predict it using only draws and pool snapshots published before it (draw size estimated from prior draws, never the actual). The model shown is whichever had the smallest average miss, and the range comes from its real past errors — sized so roughly 8 in 10 draws landed inside.

Pool model

Pool size follows a drift model: the recent net growth rate (weighted toward the latest snapshots, half-life 30 days) extrapolated from the last published count. Backtested the same walk-forward way against IRCC's own snapshots: average 28-day error 1.4% over 48 tests.

Scenario explorer

Your sliders re-solve the pool-inversion cut-off against a pool scaled by your growth rate, with a feedback term: fewer or smaller draws in your category leave more candidates in the pool. Score bands scale proportionally with the total, and projections are anchored so default sliders reproduce the backtested prediction. The further out a draw, the wider its range.

What this can't see

Policy. A ministerial instruction, a new category, or a paused program changes the game overnight, and no model of past draws will see it coming. Treat everything here as a well-calibrated reading of current patterns — not a promise. Source data: IRCC rounds of invitations, refreshed daily.