Three phases of the endgame
- Finish in-progress work. Complete any problem you’re already partway through before starting anything new.
- Scan for quick wins. Skim any remaining unattempted or skipped problems for one that looks fast, even out of order.
- Final bubble audit. Confirm every problem has an answer marked, and that the marked answer matches what you actually intended.
Priority order
If time is tight, your priority should be securing points where they’re most reliably available: problems 1–10 first (aim for near-complete accuracy there), then 11–15, then 16–20 only if you’re genuinely confident, and 21–25 last, since those are the least likely to be worth the remaining time relative to double-checking earlier work.
Unanswered problems at the buzzer
Apply the elimination-and-guessing rule from earlier in this section to anything still blank: even a single eliminated choice makes guessing better than leaving it empty, and with the clock nearly out, a fast guess is strictly better than continuing to work a problem you won’t finish anyway.
A quick emergency checklist
With under a minute left: bubble a guess for every blank problem first (this guarantees you’re not leaving free expected value on the table), then use any remaining seconds to double check that your answer sheet actually matches your work.
Applying it
Scenario: With 3 minutes left, you have problem 18 half-solved, and problems 22 through 25 completely blank.
Response: Spend up to a minute finishing problem 18 if it’s genuinely close to done. Then, rather than attempting a fresh hard problem, apply quick elimination heuristics to the blank problems and guess — four quick, reasoned guesses are worth more in expectation than one rushed, incomplete attempt at a new problem.
Practice problems
1.With 2 minutes left, you have one problem half-worked and three completely blank. What's the best use of your time?
2.Why does the priority order recommend focusing on problems 1-10 over 21-25 when time is short?
3.At the very end, why bubble in guesses for blank problems before doing anything else?
4.What's the risk of attempting a brand-new hard problem with only 90 seconds left, instead of following the endgame checklist?