IMC Cognitive Assessment: both item types, interactive
The 2026 IMC trader funnel opens with an SHL-style cognitive screen built from exactly two item types: tap-to-fill figure sequences (paint the next grid state in solid green or purple dots) and drag-the-line node chains (extend a green line through labelled circles in rule order). This is a screen-for-screen replica of both, with a timed 12-item run and an untimed practice mode that states every rule in full so you can check each answer yourself. Free, no signup.
What the real screen looks like
Every question sits on a plain white card with a small Question label and one instruction line. The tap-to-fill items show three or four small panels of the same figure across the top (a hexagram, a 4×4 pinwheel grid, a triangular star, nested squares, a quartered circle, or a pentagram) with cells filled in solid green or purple polka-dot. Below sits a large empty copy: you tap cells to cycle them empty → green → dotted → empty until the big figure shows the sequence's next state, then hit the dark-blue Next pill. A small blue reset square sits top-left.
The drag-the-line items scatter six to nine circles labelled with short codes such as LNF, 3L, ^.a and T*P, and pre-connect the first few with a solid green line. You drag from the last green node toward the code that continues the progression; while dragging you see a dashed green line with an arrowhead, and a blue undo square sits top-right. The rule is always an alphanumeric progression: letters advancing by a fixed stride, digits counting up or down, or a symbol rotating through the positions.
How this replica grades you
- Tap-to-fill items are graded on an exact cell-state match: every cell must have the right fill (empty, green, or dotted), exactly like the real scoring.
- Drag-the-line items are graded on the full chain: every node visited, in exactly the rule's order. Undo steps you back one link at a time.
- Timed mode is the real thing: 12 items, 60 seconds each, feedback only at the end. Practice mode is untimed and reveals the rule after every item.
- Every graded node chain is generated from a strict rule and checked against it before release: 16 chains and 10 figure sequences in the pool, 6 of each type drawn per run, so the mix reshuffles every run.
- The exact chains candidates reported from the real screen are listed on the intro screen for familiarity only. The reports give the order but not a rule that reproduces it, so those chains are never graded.
How to beat it
- Name the rule family first. Rotation, stepping, alternation: on figure items, check whether the green and dotted layers move together or independently before touching a cell.
- On node chains, diff two adjacent codes. The stride between the first two connected nodes almost always generalises; check it once against the third node, then run the chain without re-deriving.
- Watch for interleaved families. Chains like
P1 → Z9 → Q2 → Y8alternate between two progressions, the single most-missed pattern in candidate reports. - Use reset early, not late. On tap-to-fill items a mis-painted cell costs the whole item; if the picture looks off, reset and repaint rather than hunting the one wrong cell.
The cognitive screen is only the gate; the math screens behind it carry the weight. Keep those warm:
Full IMC funnel → · IMC online assessment guide · IMC interview questions · Optiver Number Logic (same pattern muscle)
How we keep this accurate
Rebuilt from 2026 candidate-reported IMC assessments: the two instruction lines, the figure families, the green/purple-dot fills, the blue reset and undo squares, and the node label styles are reproduced from screenshots of the real screens. The exact chains candidates reported are shown on the intro screen but not graded, because the reports give the order without a rule that reproduces it; every graded item is generated from a stated rule and checked against it before release. Formats rotate between cycles and QuantVault is not affiliated with IMC. Sat it recently and something's off? [email protected].