Prediction first

42 questions, 90 minutes — about two minutes each, and the clock is the real opponent. A working MCQ protocol:

1. Read the code first, the question second. Most questions show a program and ask "what is printed / what is true / what is returned". The code is the ground truth — read it top to bottom before reading the stem.
2. Predict BEFORE looking at the choices. Write your answer on scratch paper (trace table!). Then find it among the choices. If it is not there, re-trace — do not pick the closest-looking option. This is the single highest-yield habit: the wrong choices are built from your predictable mistakes.
3. Eliminate with probes, not preference. Empty/single/all-equal/ order probes (Module 12) kill two or three choices in seconds.
4. The 90-second rule. Two minutes per question average; if you are past 90 seconds without a prediction, mark, move on, return. Unanswered questions cost the same as wrong ones on the digital exam (no penalty for guessing) — so NEVER leave blanks; guess at the end if needed.

Why prediction-first matters psychologically: the choices include answers built from off-by-one errors, reversed conditions, and reference confusion — exactly the states YOUR trace would visit if done sloppily. Finding your predicted value among the choices confirms the trace; not finding it is a signal to re-trace, not to pick.