One of the oldest and simplest tenets in sports analytics can be summarized like this: trust points over wins. Call it Pythagorean expectation, point differential, Hollinger Power Ranking, whatever. What it boils down to is this: the difference between points scored and points allowed is a better predictor of future record than a team’s actual win-loss record.
Part of it is simply expanding the sample. In most sports, there are many more points scored (or allowed) than there are games won and lost. Since the score eventually determines the binary win/loss in a contest, it makes sense to incorporate all points into the calculation.
But it’s also about the randomness and chance baked into any sporting event. In a close contest, one bounce — fortuitous or not — can decide the entire game. Miss or make one shot, throw one pass just out of reach or right into the shooting pocket, and the entire result changes.
Another (related) old principle of sports quants? It’s very rare for any team to show a consistent ability to win those close games from year to year. Study after study has shown a lack of predictive power for a team’s record in close games from one year to the next. There’s some evidence in NFL football that great quarterbacks can help you beat the trend, and the Memphis Grizzlies do it in the NBA every damn year, but generally — your record in close games, between seasons, boils down to a lot of luck. Similar evidence has been uncovered in the college game.
So which teams have been good (and, perhaps, overperforming) in crunch time so far this year? KenPom’s ‘luck’ rankings are out there for all subscribers to see, so I figured I’d dive a bit deeper. I took the current top 100 teams — as ranked by KP — and analyzed their games to get their point differential, stats, and records in games with ‘clutch’ time. ‘Clutch’ time here follows the classic NBA stats definition — last 5 minutes or overtime of a game, with a close score — except I used 6 points (or two baskets) as the cut-off point, rather than 5 points. What follows below will be some analysis of the best teams in crunch time, why they’ve been good, and whether it looks sustainable.
Well, that works
Coastal elites Gonzaga and Saint Mary’s have not trailed in clutch time yet this year — apparently, they just don’t understand fan angst like us normal folks. And they’ve largely avoided being in close games altogether. Gonzaga has spent just 21 offensive possessions and 16 defensive possessions ALL YEAR with the other team within 6 points of contact. They’re topped by Saint Mary’s, at 7 offensive possessions and 6 defensive possessions.
Not fair, you guys.
What Gonzaga and St. Mary’s are doing is not fair, especially to fans of teams on the list. In the table below are the ten teams who have spent the MOST possessions in ‘clutch’ time so far this year.
Butler and Oklahoma have spent, at their average pace, about two entire games in ‘clutch’ time so far this year. Stressful. Only 11 teams of those analyzed have spent more than 100 possessions in close games, and about a third of the teams haven’t spent the equivalent of a game (per KP, 69.2 possessions on average) in close contact at the end of the games over the first two thirds of the season. While most of the teams in the table have a positive point differential, this is somewhat skewed by the fact this analysis includes the top 100 KenPom teams — essentially, most of the teams with overall positive point differentials in all of D-I. For reference, the average margin per 100 possessions in clutch time (including when teams are ahead and behind) for all of these teams is +6.
Kings of clutch
The table below shows the number of possessions, efficiency margin, and close game won-loss record of the teams that have performed the best so far this year in ‘clutch’ time.
We’ll go through a few of the teams on this list, especially those who have larger samples. I’d love to write blurbs on all of them, but my poor editors would (rightfully) kill me.
This helps unpack a bit of how the Bears have run out to a 20-2 record this year — they have been simply outstanding in the clutch. This can be seen, by proxy, in KenPom’s luck rating — Baylor is currently rated as the 37th ‘luckiest’ team in the country. But that understates their performance in close games.
When the game is within six points in the last five minutes of the second half or OT, Baylor is outscoring its opponents by FIFTY points per 100 possessions. This includes 78 offensive possessions, and 66 defensive possessions. Their offense scores nearly 1.4 points per possession, and the defense is very stingy, allowing just 0.89 points per possession, with defensive effective field goal percentage of 31.
Much of this is traceable to how well the offense has performed in these situations. The Bears have been hitting 50 percent of their 3s and shooting over 80 percent at the rim in clutch time this year. This helps them set their 2-3 zone on the other end, anchored by a starting frontcourt with nearly 22 feet in aggregate wingspan.
While Baylor’s play in clutch time is almost unsustainably good (see the Kansas loss last night), their close-out ability this season has already banked them a bunch of wins. They’re a rock solid team beyond their record, and their clutch play so far has helped set up a resume that will place them among the top protected seeds.
The wheels may be coming off somewhat here. Butler has dropped its last two contests at home, one of them a closely contested loss to Georgetown, and the other a near double-digit loss to a Mo Watson-less Creighton. But Butler started the season 18-3, with an outstanding record in close games. In all contests that have included ‘crunch’ time, the Bulldogs are currently 10-3, and have a net rating of +35 points per 100 possessions.
As noted above, Butler are in closely contested games a lot — their 130 offensive possessions spent in clutch time currently rank as the second-most in the country. And while they’ve been great overall in the clutch, it’s when they’re behind that really stands out.
The Bulldogs are currently playing at a net positive of 41 points per 100 possessions while they’re behind/tied in clutch time. The team is significantly better than its typical margin in these situations — per KenPom, their adjusted efficiency margin is closer to 21 points per 100. They’re mainly carried in these situations by their ability to get to the free throw line — when tied or behind in clutch time, they have a free throw rate well north of 50 percent. While this is typical for a lot of teams in this analysis — whether ahead or behind, teams tend to foul a TON in clutch time — it’s the biggest reason for their comeback ability. They’re also quite good at the rim, having hit 75 percent of their shots in close when the game’s on the line.
It’s pretty safe to say Butler has been living dangerously, though. While a 10-3 record in close games is great, that also comprises around 56 percent of their overall contests so far. The total number of games (13) that hit clutch time is nearly tops in the teams analyzed. While they’ve shown excellent ability to finish in these games, at a certain point you’re tempting fate a bit too much.
Maryland has ranked among the ‘luckiest’ teams in the country through the duration of Melo Trimble’s career. In his freshman campaign, the Terps were ranked No. 2 in KenPom’s metric. After a drop back to the pack last year, they’re back up to 17th through 22 games this year. They’re currently sporting an 11-1 record in games that get into clutch time, among the best in the country.
It may point to a some inherent advantage for teams helmed by a calm point guard presence that can create his own shot and get to the line in late-game situations. To look at a proxy — Trimble has taken 51 of his team’s 210 shots over 25 seconds into a possession (or after an offensive rebound), and hit 24 of them (41.9 percent on 3’s!) for an effective field goal percentage of 56.9. That’s very high for these type of shots; the Maryland team as a whole (including Trimble) has just a 48.3 effective field goal percentage on these attempts.
The Terps have been comeback kings at an even higher level than Butler this year. They currently have the second best net differential when behind or tied in clutch time at +54 points per 100, with a total of 39 offensive possessions and 30 defensive possessions. Again, much like Butler, this is driven by an incredible ability to get to the free throw line — Trimble has led them to a ~75 percent free throw rate in these situations. They’ve also hit six of their 13 attempted 3’s – something that probably should come down a bit as the season goes on.
It’d probably be fair to call their defense more lucky than good in crunch time. Despite a solid rating of 76.7 points per 100 possessions, too much of that is built on teams going 0-for-8 from jump shooting range. But it’s been a key part of their incredible net rating so far — it’d just be tough to count on the same going forward.
The Choke Squad (Probably just unlucky, though)
The table below summarizes the stats of teams in KenPom’s top 100 that have performed worst in ‘clutch’ time so far this year.
Poor San Diego State.
And, a few words on some of the notable teams on this list.
A quick inspection of the KenPom rankings probably could have told you the Virginia Cavaliers belong on this list. The ‘Hoos, despite a record that includes four losses, are currently ranked as the second best team per adjusted efficiency. They also happen to land near the bottom of the country in luck — 238th as of this writing.
All of their struggles in ‘clutch’ situations stem from holding leads — they’re currently sitting at a -45.8 points/100 possessions net rating in clutch time while they’re ahead. They actually have a (minorly) positive net rating while behind in the same situations. So what’s the problem?
Their defense has been having a really tough time in the admittedly tiny sample. In 37 possessions faced while ahead in crunch time, the defense has allowed 49 points, for a defensive rating of 1.32 points per possession. Teams are hitting nearly 75 percent of their 2s in these situations, and 35 percent of their 3s. The offense also hasn’t helped — they’ve managed just 26 points on 30 offensive possessions, for a rate of 0.87 points per possession. The offense can’t buy a bucket from 2, hitting just five of 16 attempts, and they’ve clunked a bunch of their free throws. They’ve hit just seven of their 14 freebies when in this situation. It all adds up to a few close, tough losses — especially those against West Virginia and Villanova, where the ‘Hoos led in crunch time for periods before eventually losing.
You’d have to think the defense has the capability to pick it up in these situations. After all, Virginia currently has the third-best defense in the country by adjusted efficiency. There’s no particularly good reason they’d be significantly worse when the game slows down — that’s how they like to play, anyway.
Wake Forest Demon Deacons
And for our last team writeup, Virginia’s ACC compatriot, the Wake Forest Demon Deacons. The Deacons are an interesting case study for clutch vs. unlucky so far this year. Despite a middling 13-9 record and a quality win resume of “Miami,” they’re currently ranked as the 34th best team in KenPom’s rankings. Much of this is based on their ability to lose ‘closely.’ In their nine losses this year (all of them to KP top 51 teams), the deficit has only been in double-digits three times. Five of the losses have been by five points or less.
Essentially, they’re good enough to hang with the better-to-best teams in the country, but they haven’t been able to finish any of them off. Much of this is tied to their terrible point differential in the crunch. As noted above, their -24 points per 100 net rating in ‘clutch’ time is seventh-worst among KP’s top 100. It’s been a big factor in most of their close losses.
As you’d expect, they’ve struggled on both ends of the floor to achieve this differential. The offense has scored just 94.9 points per 100 in all ‘clutch’ situations, 21 points per 100 less than their 114.0 ‘raw’ mark overall this year. And the defense hasn’t been much better, coughing up about 1.2 points per possession and letting teams shoot well above average from just about everywhere (71 percent at rim / 43 percent on 2-point jumpers / 65 percent from 3). This is at least more in line with what you’d expect from their D. Wake Forest is in the Marquette/UCLA mold — great offense (13th in adjusted efficiency) with a bad D (146th in adjusted efficiency).
If Wake Forest is going to turn it around in crunch time, the offense will probably need to start picking up its own slack. The Demon Deacons probably won’t be able to defend at a consistent enough level to prevent teams from getting buckets, but they have shown the ability to outscore them. Just need to bring it to the parts of the game that matter most.