Pacing the ACT Math Section

Math timing problems that show up as a crisis at question 43 usually started ten questions earlier. Most students do not get uniformly slower across the section. They lose time in a small number of specific places, and the effect of that loss doesn't become visible until much later, when the reserve it was quietly spending is gone. The question this chapter answers is not “how do I go faster.” It's “where does time stop buying useful information, and what do I do about it.”

The current enhanced section gives 45 questions and 50 minutes, an average of about 67 seconds per question. That average is a warning against sustained overspending, not a per-question law. Some questions are genuinely 15 to 30 second items; others need several linked steps and legitimately take longer. Five consecutive two-minute questions is a problem worth noticing. One 90-second question that is nearly solved usually is not.

When a question is taking too long, the first useful move is classifying why, not just noticing that it is. Time disappears for one of four reasons: recognition delay, when no first step has appeared yet; execution delay, when the route is clear but the algebra, arithmetic, or tool sequence is slow; indecision, when the route and the execution are both fine but the question keeps getting rechecked anyway; and tool friction, when a calculator or Desmos stops removing work and starts creating it. These four causes call for different fixes, which is why naming the right one matters more than simply working faster.

▸ Eight patterns worth naming

Mechanism	What it looks like	Required response
Recognition delay	No first step appears; rereading the question without any new information emerging.	Classify the question fast: equation, function, ratio, geometry, data, or parameter. If no route appears after one more read, mark and move.
Correct-but-slow method	A valid solution path is reached, but it took longer than a faster available route would have.	Review correct-but-slow results as pacing errors during practice, not just as correct answers, and learn the faster alternate route.
Sunk-cost persistence	Staying on a question because leaving would make the time already spent feel wasted.	Three minutes already spent is not a reason to spend the fourth. The time is already gone either way.
Overchecking	A question already answered gets recalculated again with no specific new concern driving the recheck.	A second look must have one named job. Recalculating without a reason consumes reserve without adding accuracy.
Calculator or Desmos friction	Simple arithmetic entered slowly; a wrong mode, window, or re-entry costs more time than the calculation itself would take by hand.	Apply the hand/handheld/Desmos/hybrid framework from Part VII in the moment, not just as a study concept.
Reserve failure	The section's more expensive cluster of questions is reached with too little time left to handle it.	Use section landmarks from practice to monitor whether reserve is on track, instead of pacing every question equally.
Panic-speed cascade	After noticing the section is behind schedule, subsequent easy questions get rushed and start producing new errors.	Recovery means returning to normal solving quality and tightening future exits, not compounding the loss by rushing everything else.
Format friction	Bubbling, page annotation, or digital scratch-work and Desmos navigation costs time the student didn't practice for in that format.	Practice complete, timed sections in the actual format you will test in, not just in whichever format is more convenient to practice in.

Recognizing these eight patterns matters more than memorizing their names. A student who can say “this is recognition delay, not execution delay” has already done the hard part of the diagnosis, because the fix for each one is different and applying the wrong fix wastes the same time it was meant to save.

▸ Return priority for skipped questions

A skipped question needs a return priority, not just a mark. Returning to marked questions in the order you happened to skip them treats a nearly-finished problem the same as one where no route ever appeared, which is not an efficient use of remaining time.

Priority	Question state	Return action
1	One named step or check remains	Finish it first. This is the cheapest remaining point on the section.
2	The route is known but execution was long	Resume from the work already written down. Do not restart from scratch.
3	Two answer choices remain plausible	Use one targeted test or backsolve to distinguish between them, instead of re-deriving the full solution.
4	Skipped because a calculator or graph route might be faster later	Return once that tool is genuinely available and the setup is already understood.
5	No route ever appeared	Lowest priority. Keep the current best guess unless a new route appears immediately upon return.

▸ What a second look is actually for

A specific second look has one job, not a general recheck of everything. The most useful named checks on this section are: the requested object (x versus 2x, input versus output, radius versus diameter); a single sign or exponent; units (linear versus square or cubic, percent versus decimal); a calculator entry compared once against the printed problem; a restriction (an extraneous root, a zero denominator, a positive-length requirement); and a diagram assumption (was the right angle or the similarity relationship actually given, or only assumed). None of these is a full redo of the solution. Restarting a settled answer only because the clock made it feel suspicious is its own kind of time loss.

▸ Recovering after a slow question

Recovery after a slow question follows a different rule than instinct suggests. The natural response to losing three minutes on one problem is to rush the next three; the better response is to return to normal solving quality and tighten how quickly you decide to exit future low-value questions. The lost time is already spent either way, and rushing compounds the damage by introducing new errors on top of the original one.

Situation	Instinct to avoid	Better response
One question consumed three minutes	Rush the next three questions to make up time.	Return to normal solving quality; look for the next opportunity to exit a low-value question sooner.
Behind at a section landmark	Do every remaining step of every problem, just faster.	Stop unnecessary second looks; preserve accuracy on the cheap, fast questions still ahead.
Several questions are marked	Return to them in the order they were skipped.	Return by the priority hierarchy above, closest-to-solved first.
A calculator route keeps failing	Keep retrying the same entry.	Switch to a different route, or mark and move.
Very little time remains	Fully solve one final difficult question.	Make sure every remaining question has an answer selected first.

▸ Matching the fix to the actual profile

Different students hit different pacing failures, and the fix depends on which profile actually describes what happened. Applying the wrong profile's fix, such as forcing more speed onto a student whose real problem is content weakness, wastes practice time without solving anything.

Profile	Diagnosis	Fix
High accuracy untimed, but incomplete when timed	An execution or recognition bottleneck, not a content gap.	Timed microsets on individual skills, practice with faster routes, full timed sections, and skip-and-return discipline.
Finishes the section but accuracy collapses late	Earlier overspending creates rushing and fatigue by the end.	Find the first place time leaked, then practice recovering from it without a late-section quality drop.
Slow and inaccurate even untimed	A genuine content or method weakness, not primarily a pacing problem.	Return to the relevant content chapter first. Forcing aggressive timing onto a content gap doesn't close it.
Finishes early but with careless errors	Over-speeding or insufficient review, not a timing shortage.	Use the named second-look checks above and this book's error-log habits, not simply slowing down uniformly.

▸ Auditing a timed practice section

After any serious timed practice section, a short audit is worth more than the practice itself if the results are never examined. Useful questions to ask: Did every question get an answer, and if not, where did coverage first become endangered? Which question first took much longer than the normal solving pace? Was that delay recognition, execution, tool friction, or indecision? Which correct answers took too long, even though they were right? Where did sunk cost keep a question going after the next step had stopped being clear? Which item had a faster alternate route available, such as backsolving, estimating, or a property shortcut? Were any calculator entries unnecessary? Did checking the timer improve decisions, or just fragment attention? And, honestly: was accuracy also a problem untimed? If it was, this is not purely a timing problem, and no amount of pacing practice will fix it alone.

The system this chapter is built around reduces to a short list: recognize the route before starting to calculate; choose the fastest reliable engine for that specific question, using the framework from Part VII; stop a second look once it no longer has a specific job; mark a skipped question with a reason; return to marked questions by expected payoff, not by the order they were skipped; keep full coverage as the priority once time is genuinely short; and audit correct-but-slow work after every serious timed section, not just outright wrong answers. None of this replaces the content taught in Parts I through VI. It is what makes that content actually usable at full speed, under the current 45-question, 50-minute format, on test day.