Each team is around 18-22 games into the season, which means we’re at the quarter mark. Considering it’s only Week 13 of the NFL season, it feels like the NBA campaign has flown by so far.

Being a quarter through the season means we can actually dive into some of the numbers and know that they aren’t skewed because of small sample size. Some teams have outperformed expectations, while others have supremely disappointed. What’s the reason? Have some teams been merely unlucky; will they get better? Or did the public misevaluate players or rosters and perhaps some teams just simply aren’t as good as initially thought? Thankfully there are some notable metrics that can help us cut through these questions.

### Point Differential and Pythagorean Wins

The most basic analysis of expectation versus reality is to look at each team’s Pythagorean win expectation, which calculates how many wins a team ‘should’ have based on its point differential — which has been shown to be far more predictive of future performance than a team’s win-loss record. For example, a team could be playing opponents close and getting lucky in the last two minutes over a month or so. While their record may be 14-2, are they really as good as a team that is 14-2 and blowing almost every team out? Margin of victory matters, and point differential and Pythagorean wins highlight which teams have been playing better or worse than their records indicate.

Here’s an interactive graph with each team’s actual win total and their Pythagorean expected win total:

There are seven teams whose real win total exceeds their Pythagorean win total by two games: the Celtics, Cavaliers, Pistons, Bucks, Timberwolves, Suns, and Kings. There are five teams whose win total drags behind their Pythagorean total by two games. The only team with a net difference of four or more games is the Oklahoma City Thunder, who have a Pythagorean win expectation of 12 games but are 8-11 — dead last in the Western’s Northwest Division.

Why the huge difference? Let’s dig into a couple more statistics that have proven to be volatile and could thus indicate whether a team has been lucky or unlucky: 3-point field goal percentage made/allowed and clutch performance.

### 3-Point Performance

There have been many studies over the past couple of years suggesting that, while a team can limit the number of 3-pointers an opponent takes, whether those shots go in or not is somewhat random. There are exceptions to this, and there are many players who are better at closing out and creating contested 3-pointers with their length, but in general 3-point defense has been shown by Ed Feng and Ken Pomeroy to be moderately random. Who are the teams that have been lucky or unlucky in this regard?

The Pacers, Blazers, and Celtics stand out among the teams who have likely been lucky offensively and defensively. Indiana actually leads all teams this season with an absurd 40.9 percent 3-point mark. They don’t have non-shooters on their squad, but you can bank on negative regression for some of their main guys:

• Cory Joseph: 33.7 percent career, 47.2 percent this season
• Victor Oladipo: 35.6 percent career, 46.2 percent this season
• Bojan Bogdanovic: 37.7 percent career, 45.6 percent this season
• Domantas Sabonis: 32.9 percent career, 45.5 percent this season
• Myles Turner: 35.0 percent career, 41.2 percent this season

Some of these guys are young and have improved this season, but it’s highly unlikely that any of them will be over 40.0 percent by the end of the year; regression is going to come hard for the Pacers.

The Blazers and Celtics are shooting 38.1 and 36.6 percent, and it’s possible they won’t regress much offensively given their talent. However, they are allowing 33.1 and 33.0 percent from behind the arc on defense; those are top-five marks. While Boston has great length on the perimeter, it is still unclear just how much that affects 3-point performance. It is encouraging these teams are allowing only 25.0 and 26.0 3-point attempts per game — top-five marks again — but eventually opponents will start hitting more shots.

### Clutch Performance

A team’s clutch performance on offense and defense isn’t completely random, but there’s still quite a bit of noise. For example, a team could have a highly negative clutch Plus-Minus simply because a couple players on opposing teams hit buzzer-beaters or half-court shots. After 82 games, clutch performance will likely more closely mirror a team’s overall performance, but there can be huge discrepancies after just a quarter of the season. You can probably guess by the title of this piece who has been ridiculously awful in the clutch:

The Thunder have scored only 0.965 points per possession in the clutch, and they have allowed opponents to score a whopping 1.637 points per possession. Given the star power and elite defensive capabilities of Paul GeorgeAndre Roberson, and Steven Adams, that is an astoundingly bad number.

Overall, they’ve posted a league-worst Plus-Minus of -0.672 per possession in the clutch. The Hawks rank second in the league at -0.464. I wrote before the season that they were likely to struggle offensively early on because of their weighted usage problems, but this is more extreme than anyone anticipated. But it’s hard to believe this will continue: With guys like PG, Adams, and Russell Westbrook, they’re just too talented to post season-long numbers like these.

### Practical Applications

There are obvious DFS and gambling implications here, and Matt LaMarca wrote about tonight’s slate and whether Westbrook could continue his hot streak versus the Magic. Perhaps this is the start of the positive regression for the Thunder.

Vegas lines are obviously affected by a variety of factors, such as injuries and matchups. For those and many more reasons, we can’t just run a simple linear regression to see the relationship between the point spread and a statistic like SRS, which takes into account point differential (so it deals with Pythagorean wins) and strength of schedule. Instead let’s look at just two games tonight:

• Brooklyn Nets (SRS: -3.20) at Dallas Mavericks (SRS: -4.02) — Mavericks -5
• Oklahoma City Thunder (SRS: 3.01) at Orlando Magic (SRS: -3.31) — Thunder -6

Even when you take into account home court advantage for the Mavs, which is probably worth about 2-3 points to the spread, it’s clear the Thunder’s SRS differential over Orlando suggests they should be far greater than six-point favorites, given that the Mavs are five-point favorites over a bad team in the Nets.

Unfortunately betting NBA spreads isn’t as easy as that analysis suggests, but the data does point toward value on the Thunder. Using the metrics we know have high correlation to future performance (point differential and expected win totals) and ones we know are mostly random (clutch and 3-point performance), we can likely gain an edge on the public.