Table of Contents
A top-25 team can lose twice in a week and climb four spots. Another can win both games and fall three. Neither is a glitch. The NCAA basketball rankings scrolling across the bottom of your screen are a blend of half a dozen systems, each answering a different question, and most arguments about them come from people comparing numbers that were never meant to be compared.
What follows is the routine I use from November through Selection Sunday. It takes about twenty minutes a week, and it stops you making the two mistakes that ruin every March group chat: treating the AP poll as a prediction, and treating the NET as a verdict.
Step 1: Match the ranking to the question you’re asking
There is no single correct ranking. There are four or five useful ones, and they disagree on purpose. Before you open any of them, decide what you actually want to know.
- AP Top 25 — roughly 60 sportswriters and broadcasters voting weekly. It rewards wins, good losses, and reputation. Best used as a rough consensus of informed opinion, not a forecast.
- USA Today Coaches Poll — more than 30 head coaches. Moves slower than the AP poll and is prone to lagging a week or two behind reality.
- The NET — the NCAA’s own evaluation tool since the 2018-19 season, built on adjusted efficiency plus a résumé component. This is what shows up on the team sheets the selection committee reviews.
- KenPom and Bart Torvik’s T-Rank — purely predictive, built on adjusted offensive and defensive efficiency per 100 possessions. These answer the question “who wins on a neutral floor tomorrow?”
Use KenPom to settle an argument about who’s better. Use the NET team sheet to understand which games the committee will debate. Use the AP poll to know what your casual friends have heard. Using the AP poll to predict a game is how people get blindsided in the second round. If the distinctions between these systems blur together for you, this breakdown of what team rankings actually measure is worth ten minutes before you go any further.
Step 2: Learn the Quadrant system, because it decides everything
Quad records are the part of the rankings people skip, and they’re the part that matters. Every game gets sorted by two things: where it was played, and how good the opponent is according to the NET.
- Quad 1: home vs. NET 1–30, neutral vs. 1–50, away vs. 1–75
- Quad 2: home vs. 31–75, neutral vs. 51–100, away vs. 76–135
- Quad 3: home vs. 76–160, neutral vs. 101–200, away vs. 136–240
- Quad 4: everything else
Here’s why the geography matters. A road win over a team sitting at NET 72 is a Quad 1 win. Beat that same team at home and it’s only Quad 2. That single line explains a huge share of the confusion in February, when two teams with identical 18-6 records have wildly different résumés. A 6-4 Quad 1 record built mostly on the road is a completely different animal from a 3-6 Quad 1 record built mostly at home, even if the raw win totals look similar.
Step 3: Track the direction, not the number
A snapshot tells you almost nothing. A trend tells you nearly everything.
Every Monday, write down three numbers for the ten or so teams you actually care about: AP rank, NET rank, KenPom rank. Do that for six weeks and you’ll have a curve. A team that has climbed from NET 44 to NET 21 since mid-December is a genuinely improving profile, regardless of what the human poll says. A team stuck at 15 in the AP poll while sliding from NET 18 to NET 33 is being propped up by name recognition.
This is also where the weekly churn gets genuinely hard to follow, with bracket projections, injury news, and four ranking systems all updating at different times. There’s a decent explainer on keeping up with the rankings as they get harder to read that’s worth pairing with this routine.
A worked example
Say Team A sits 22nd in the AP poll, 14th in the NET, and 9th at KenPom, with a 5-4 Quad 1 record and zero losses outside Quad 2. Every analytical number says the voters are three or four spots too low. The committee will almost certainly seed them better than their AP rank suggests, and they’re a sneaky good pick in a bracket pool.
Now flip it. Team B is 15th in the AP poll, 31st in the NET, 1-5 in Quad 1, and has a Quad 3 home loss in December. The voters are looking at the win-loss record and the logo on the jersey. The team sheet says first-weekend exit risk. Two teams, similar-looking résumés at a glance, opposite conclusions. That’s the whole skill.
Step 4: Build a one-page sheet and stop checking your phone
Ten teams, ten rows, one page. Every Sunday, fill in the same fields. It takes twelve minutes and it makes you the person in the group chat who’s actually right in March.
- Team and current AP rank
- NET rank and KenPom rank, side by side
- Quad 1 record, then Quad 1 plus Quad 2 combined
- Best win, with the venue
- Worst loss, with the venue
- Arrow: up two, down five, flat
- One-line note — injury, brutal stretch, three straight road games
The note column is the one people forget. Metrics don’t know your point guard rolled an ankle on January 9. Efficiency numbers need three or four weeks to fully absorb a lineup change, so the rankings you’re reading today are describing a team that no longer exists.
Step 5: Know the traps that make smart fans wrong
November rankings are mostly brand
Preseason polls are built on last year’s roster, recruiting rankings, and reputation. A team ranked 12th in the preseason that starts 4-3 hasn’t “fallen” — it was never accurately placed. Ignore anything before roughly December 15.
Margin of victory matters, whether you like it or not
Efficiency systems count every point. A 20-point win and a 2-point win are identical in the standings and completely different in the NET and KenPom. That’s why a team can win both games in a week and still drop in the analytics, and why garbage-time baskets against a blown-out opponent are worth tracking more than you’d think.
Conference depth distorts everything
Going 11-7 in a 16-team league where the bottom six teams are all top-60 is not the same as going 11-7 in a league with four Quad 4 anchors. Look at the Quad 1 opportunity count, not just the record. Three road games in eight days against tournament teams will wreck a résumé that would have cruised through a softer schedule.
Home court is worth real points
College home advantage runs around three and a half points, and in some buildings it’s more. Any ranking that ignores venue is telling you half a story. That’s also why a road win at NET 70 means more than the number suggests.
Turning the rankings into a bracket that survives the first weekend
Once you’ve tracked Quad records for six weeks, bracket picking becomes much less mysterious. You stop seeding by AP rank and start asking a better question: which high seeds have never beaten anybody good away from home?
Two rules do most of the work. First, in the 5-12 and 6-11 matchups, take the mid-major with a winning Quad 1 record and a senior backcourt over the power-conference team that went 4-12 in league play. Second, fade teams whose analytics rank trails their AP rank by more than ten spots — the market and the voters will both overrate them.
If you’re organizing something for a group, it’s worth setting the rules before the bracket locks rather than arguing about scoring in the second round, and this step-by-step guide to running a bracket pool covers the mechanics. The chaos is the point, after all — the reason March works is that the rankings are never quite right, and the 12-seeds keep proving it.
What the numbers are actually for is narrowing the field of plausible upsets. A 13-seed that won its conference tournament by beating three top-100 teams on the road is a different bet from one that backed in after losing four of six. Check the Quad 1 record. Check the worst loss. Check the arrow. Then fill out the bracket with your eyes open.
