The most expensive SAT math mistakes happen on problems you know how to solve. After years of coaching students through score reports, I can tell you the pattern that repeats: the algebra was fine, and the point still vanished, because the error lived in the reading, the entry box, the clock, or the calculator, not in the math.
That is good news, strangely. Knowledge gaps take months to close. The mistakes below are habits, and habits can be fixed in a week of deliberate practice. Here they are, ranked by how many points I watch them cost.
1. Losing points to the answer box, not the answer
About a quarter of SAT math questions are fill-ins, where you type the answer instead of choosing it, and the entry rules are unforgiving in ways students discover live on test day. The rules come straight from College Board's official test directions:
- Mixed numbers do not exist to the grader. If your answer is three and a half, entering it the way you would write it, 3 1/2, gets read as the fraction . The directions require or 3.5.
- Long decimals must fill the box. A positive answer gets 5 characters, a negative one 6, and a long decimal must be truncated or rounded at the fourth digit. For , entries like .6667 or .666 are fine; .66 or .67 can be marked wrong. Entering the fraction 2/3 itself sidesteps the whole issue, which is why fraction answers are usually safer than decimal conversions.
The fix costs one evening: do a set of fill-in practice questions and enter every answer as a fraction unless the question demands otherwise. This is the purest free-points repair on the list, because no math ability is involved at all.
2. Solving for x when the question asked for 6x
The SAT writes questions specifically to punish autopilot. Try this shape:
Your hands solve for , get 4, and 4 is sitting right there among the answer choices. The actual answer is 24, since makes . The wrong option is not an accident; distractors are built from the predictable half-finished answer. The same trap wears other outfits: solving for the wrong variable in a system, answering in minutes when the question said hours, giving the discount when the question asked the final price.
The repair is mechanical, not intellectual: before entering anything, reread only the final sentence of the question and confirm your number answers it. Two seconds, every question, non-negotiable. Of all the habits on this list, this one shows up on the score report fastest.
3. Using Desmos for everything, or for nothing
Every question on the digital SAT allows a calculator, with Desmos built into the testing app, and both extremes lose points. Students who distrust it grind algebra on problems the graphing tool solves in fifteen seconds; students who worship it type simple equations into a graph while the clock runs. College Board itself says the section includes questions where not using a calculator is the better move. I wrote a full playbook on when Desmos is faster and when it costs you time; the one-line version is that Desmos wins when the question hands you an equation to graph or solve, and loses when setup is the actual task. Decide before the pencil moves, not after two minutes of drift.
4. Letting one question eat four minutes
The math section gives you 44 questions in 70 minutes, about 95 seconds each, and the single most visible pacing failure is one stubborn question consuming three or four shares of that budget while easier points sit unread further down the module. Nothing about your score rewards solving the hardest question; the last question is worth the same as the first.
Set a personal bail-out rule: if a question has no visible path after roughly 90 seconds, mark it, guess, and move. The SAT has no wrong-answer penalty, so a guess is strictly better than a blank, and the flag lets you return with whatever time survives. This failure mode is not SAT-specific, by the way: the enhanced ACT punishes it even harder, and the evidence there is worth reading even if you never sit that test, in my breakdown of why the one-minute ACT pacing rule is now wrong.
5. Misreading the "how many solutions" questions
The digital SAT loves a question shape most classes underpractice: not "solve this equation" but "how many solutions does this equation have". Students trained only to produce values panic when the variable cancels entirely. The reading rule takes ten seconds to learn: if the equation collapses to a true statement like , it has infinitely many solutions; if it collapses to something false like , it has none. I wrote out the whole decision, including the traps around , in how to tell "no solution" from "all real numbers". On the SAT this is a full point for reading one line correctly.
6. Percent chains that don't undo
In the problem-solving and data questions, the recurring killer is treating percent changes as reversible. A price increased 20 percent and then decreased 20 percent is not back where it started:
a net 4 percent decrease, because the second percentage acts on a bigger base. Same mechanism behind "20 percent off, then 10 percent off" not equaling 30 percent off. If percent stacking is shaky, drill it directly for an evening on the percentage calculator, which shows the base each step acts on; watching the base shift is what makes the trap stop working on you.
7. Never sanity-checking the number
The cheapest safety net on the whole test is asking whether the answer is a plausible size, and almost nobody does it under time pressure. A discounted price that came out higher than the original, a probability of 1.4, an average that is bigger than every value in the table, a side length of negative 12: each of these is a finished mistake advertising itself, and each gets entered anyway by students who stopped reading the number the moment it appeared. You do not need to re-solve anything. One glance, "could this number be the answer to this story", catches the sign flips and base mix-ups that survive all your other habits. It costs three seconds and it is the only check that works even when you cannot find your own error.
One more entry-box detail belongs in your last week too: a negative answer gets six characters and the minus sign spends one of them. A student who computes , converts it to the decimal , and tries to type every digit discovers the box will not hold it, panics, and burns a minute. The directions already told you the move: truncate or round at the fourth digit, or better, skip the conversion entirely, because -5/11 fits the box exactly as it is.
The week-before checklist
None of these six require learning new math, which is exactly why they are worth your final week while new content is not. Concretely: one fill-in set entered fraction-first, the final-sentence reread drilled until automatic, a Desmos decision rule, a 90-second bail-out, one pass through the solution-counting logic, one percent-chain drill, and the three-second plausibility glance on everything. In my experience, students bleeding 40 to 60 points to this list claw most of it back in days, because they already knew the math. They were just donating points to habits, and habits take instruction better than algebra does.