How LSD.Law Ranks Law Schools

The data comes from self-reported admissions cycles on LSD.Law. Users track every school they apply to and mark the one they end up attending. That attending choice generates a head-to-head win for the chosen school against every other school the same applicant was admitted to.

An applicant admitted to five schools who enrolls at one produces four head-to-head results, not one. An applicant who doesn't enroll anywhere they reported an acceptance contributes nothing. Waitlists, rejections, and withdrawn applications are never counted — the only thing that moves the ranking is a choice between real offers.

Turning head-to-head results into an ordered list is a standard statistical problem. We use the Bradley-Terry model, the same method used to rate chess players and tennis professionals from match histories. Given a record of who beat whom, it assigns each competitor a strength score that best explains the observed outcomes. Schools sorted by strength produce the ranking.

The model handles indirect comparisons. If A consistently wins head-to-head against B, and B consistently wins against C, A will rank above C even when few applicants were directly admitted to both A and C. That chain places schools that rarely share applicant pools — a school ranked in the 40s and one in the top 10 almost never see each other's offers directly, but the schools in between fix their order.

A single admissions cycle doesn't give most schools enough head-to-head data to place them with confidence. So the model is fit on several completed cycles at once. More recent cycles count more heavily: a matchup from two and a half years ago counts half as much as one from the most recent cycle, and a matchup from five years back counts a quarter. The ranking stays current as applicant preferences shift, without throwing away the older data that anchors schools lower on the list.

To earn a published rank, a school needs enough head-to-head data to place it without guessing. The thresholds: at least 150 matchups on record, at least 10 different opponent schools, and enough weight on each matchup that the result isn't riding on a single applicant. Schools that fall short are shown as unranked — the data isn't dense enough to place them with confidence, which says nothing about the school itself.

Part-time programs and Canadian law schools are excluded entirely. Part-time programs draw from a different applicant pool (working adults, evening students) and compete in a different market. Canadian schools admit applicants who are typically not weighing US offers.

When two schools fall within about 10% of each other in strength, we treat them as tied. Tied schools share the same rank: if three schools tie at 7, the next school is ranked 10. Forcing one school to 8 and another to 9 on a gap smaller than the noise in the data would make the ranking look more precise than it actually is.

Exact ranks are published for positions 1 through 50. Schools in positions 51 through 75 are labeled "Top 75," and anything below that is "75+." Past the top 50, the cross-admit signal gets too thin to support a specific number.

The rankings page and individual school profiles also list US News, Above The Law, and Peer Assessment ranks. Those are shown for reference. They don't feed into the LSD ranking at all — the "LSD" column is the only one computed from enrollment choices. You can sort by any of the other sources separately.

Rankings for completed cycles are recomputed as applicants keep reporting outcomes, and are cached daily. During the active cycle — while applicants are still deciding where to go — the ranking is marked as "in-flux" and calculated on the fly with looser sample thresholds. Expect some movement near the boundaries until the cycle closes.