SAT Score Predictor

Your starting point. Where your last score came from matters more than anything else you enter. College Board's Student-Level SAT Suite Growth Estimates (2018) tabulates, for every starting score, the average SAT score students went on to earn — about 983,000 students from a fall PSAT/NMSQT to the spring SAT, 486,000 from one SAT to a retake. A 500 on a section of the fall PSAT, for example, averaged about 520 on the spring SAT; a 500 on a real SAT averaged about 510 on the retake, and above 650 retakes barely move on their own. This tool reads those tables directly.

Practice tests run hot. Among my own students, official Bluebook practice scores ran about 50 points above the real test that followed (a small sample: 16 of my students' score pairs), so a Bluebook baseline is shifted down 25 per section before projecting. Prep-company practice tests showed no consistent gap in my data, so they keep their score but get a wider range. If you're not sure where a score came from, the tool uses the real-SAT table, shifts it down 10 per section and widens the range by 15 per section. There's no PSAT 8/9 option — the growth tables don't cover it; if that's your only score, pick PSAT 10 and treat the result as rough.

Prep adds on top. The study-hours curve is calibrated to the published causal effects of SAT prep — Weatherholtz et al. (2020, College Board technical report, n=545,640): +21 points composite at 6 hours, +39 for the "best practices" group; Powers & Rock (1999); Briggs (2001–2009); Appelrouth et al. (2015, about 2.3 points per tutoring hour). Returns diminish but don't stop: more hours keep helping, just less per hour; reviewed practice tests count as a modest bonus; and the closer a section already is to 800, the less room prep has to add. The often-quoted "115 points from 20 hours of Khan Academy" was a raw association the same authors later corrected to roughly +21–39 points. "With a tutor" assumes you hit that study's "best practices" pattern every week (the +39 group). The study measured self-study, not tutoring; a tutor's job is to make that pattern actually happen, and Appelrouth's ~2.3 points per tutoring hour lands in the same neighborhood. In this tool's units: a 540 Math / 560 Reading & Writing real SAT (1100) projects to about 1120 on a retake with no prep; 6 hours of focused prep takes that to about 1150, 20 hours to about 1180, 100 hours to about 1260, and 300 hours to about 1360. Those run above the +21/+39 study figures because the curve keeps paying (less per hour) past the 6–20 hours the studies measured, and because lower starting scores rebound more: the same 100 hours adds about 130 from an 800 start and about 90 from a 1400 start. A PSAT 10 baseline projects higher still (an 1100 PSAT 10 sophomore reaches about 1320 at 100 hours) — that is the growth table, not the prep: sophomores get a year of school before the SAT.

The range. The College Board tables also give the spread of outcomes at each starting score; each section's range is ±1.28 of that spread (an 80% band per section). The total range is the sum of the two section bands, so it is wider than an 80% band of the total — which is why it catches more like 9 in 10. Checked against my own students' before-and-after scores: for the 52 pairs whose "after" was a real test day, 47 (90%) landed inside their range. Counting the 42 pairs whose "after" was a practice test as well, 78 of 94 (83%) did — practice tests run hot, which is why they are shifted before projecting. Hours of study explain only 5–15% of the variation in gains, so the range does not narrow just because you plan to study more. Midpoints are capped at 1600; a flat line on the chart means the tables can't go higher, not that 1600 is certain.

The chance of reaching a target treats the two sections as independent and combines their spreads (√ of the sum of squares), then asks how far your predicted total sits above or below the target in those units. It is a normal-curve estimate, not a promise.

What it can't see: motivation, test anxiety, how you actually use the time, and luck on test day. Use it as a sanity check on your plan, not a guarantee.