CAT inference report V2
Strategy
Deep CAT 2020-2025 inference with compact evidence snippets, linked examples, recurrence patterns, source reasoning, and decision playbooks.
Decision-based strategy
Percentile-specific playbooks
| Percentile | Target attempts | Accuracy | First pass | Second pass | Practise | Ignore |
|---|---|---|---|---|---|---|
| 90 percentile | VARC 12-15, DILR 6-8, Quant 8-10 | 70-75% | Harvest must-attempts first. Skip unstable DILR and hidden-constraint Quant. Use passage evidence, not outside plausibility, in VARC. | Return to marked questions with clear setup. Do not reopen abandoned DILR unless representation is now visible. Recheck MCQ traps before final submission. | RC paragraph-function notes, DILR scan logs, Quant concept-family clusters, Trap-option elimination drills | Exact source memorisation, Story-template memorisation, Long calculations without a constraint check |
| 95 percentile | VARC 15-18, DILR 8-10, Quant 10-13 | 75-80% | Harvest must-attempts first. Skip unstable DILR and hidden-constraint Quant. Use passage evidence, not outside plausibility, in VARC. | Return to marked questions with clear setup. Do not reopen abandoned DILR unless representation is now visible. Recheck MCQ traps before final submission. | RC paragraph-function notes, DILR scan logs, Quant concept-family clusters, Trap-option elimination drills | Exact source memorisation, Story-template memorisation, Long calculations without a constraint check |
| 99 percentile | VARC 18-21, DILR 10-14, Quant 14-17 | 82-88% | Harvest must-attempts first. Skip unstable DILR and hidden-constraint Quant. Use passage evidence, not outside plausibility, in VARC. | Return to marked questions with clear setup. Do not reopen abandoned DILR unless representation is now visible. Recheck MCQ traps before final submission. | RC paragraph-function notes, DILR scan logs, Quant concept-family clusters, Trap-option elimination drills | Exact source memorisation, Story-template memorisation, Long calculations without a constraint check |
| 99.5+ percentile | VARC 21+, DILR 14+, Quant 17+ | 88%+ | Harvest must-attempts first. Skip unstable DILR and hidden-constraint Quant. Use passage evidence, not outside plausibility, in VARC. | Return to marked questions with clear setup. Do not reopen abandoned DILR unless representation is now visible. Recheck MCQ traps before final submission. | RC paragraph-function notes, DILR scan logs, Quant concept-family clusters, Trap-option elimination drills | Exact source memorisation, Story-template memorisation, Long calculations without a constraint check |
Decision cards
What to do in the exam
When to leave a DILR set
Trigger: No stable representation or cases keep branching after 3-5 minutes.
Action: Mark and leave; return only after one cleaner set is banked.
When to skip a Quant question
Trigger: The stem has hidden integrality/domain/boundary constraints and no quick setup.
Action: Classify it as revisit, solve lower-load arithmetic/algebra first.
When not to overthink VARC
Trigger: Two options differ mainly by scope or force, and one is directly passage-backed.
Action: Choose the supported option; do not import outside nuance.
How to identify trap options
Trigger: Option repeats keywords but changes direction, scale, or certainty.
Action: Test the option's claim against paragraph function.
How to choose the first DILR set
Trigger: A table/grid/chart or explicit finite list is visible, and the first clue fixes a row/column/order.
Action: Start there; avoid pure prose conditionals until scan is complete.