Analyzing the NCAA Tournament’s ACC Powers Using STATS SportVU


Questions arise annually for the NCAA Tournament selection committee when the bracket is revealed on Selection Sunday. No one ever is fully happy with the results when bubbles burst, and those who received a bid often wonder about their seed or where they’re headed on opening weekend.

The ACC doesn’t have much to complain about this year, though, tying the conference record set last season with nine teams in the tournament, the second-most ever for any conference. Virginia is the tournament’s No. 1 overall seed, North Carolina and Duke both landed on the No. 2 line, and a couple teams toward the bottom of the league standings snuck in, too.

During the ACC Tournament, STATS had its SportVU player-tracking cameras placed above the Barclays Center court and accumulated data during each game to deliver advanced analytics to the ESPN broadcast and fans watching or following on social media. Now, we’ll go even deeper into that data for the first time.

Below is a breakdown of SportVU data gathered during Virginia, Duke and North Carolina games and how those analytics could help or hurt each team heading into the NCAA Tournament. We’ll also delve into some metrics from the mind of Ken Pomeroy, who receives his statistics from STATS’ industry-leading data feeds to power his proprietary model and deliver the most respected college basketball advanced metrics in the game.

According to KenPom, Virginia has the second-best chance of winning the national championship at 17.6 percent, with Duke third (12.1) and North Carolina sixth (5.7).

Virginia (31-2)

KenPom rating: 1
Seed: No. 1, South Region
First-round opponent: UMBC

If you haven’t heard overwhelming praise for Virginia’s defense by now, you wouldn’t be reading this. The Cavaliers rank No. 1 in KenPom’s adjusted defense rating with an average of 84.4 points allowed per 100 possessions. Virginia is dead last in the nation with an average of 59.1 possessions per 40 minutes, slowing the game down to a snail’s pace and forcing long, closely guarded possessions on the defensive end. Does that mean the Hoos can’t score? Absolutely not.

The Cavaliers are 21st nationally in adjusted offense with an average of 116.5 points per 100 possessions. They balance their attack with only three guys averaging double figures in Kyle Guy (14.1), Devon Hall (12.0) and Ty Jerome (10.5) with a very deliberate offensive scheme.

Just how deliberate? During the Cavaliers’ three ACC Tournament games tracked by SportVU, they threw 0-2 passes on 15 percent of possessions, 3-5 on 41 percent, and 6+ on 44 percent. Virginia averaged 1.00 points per possession with 0-2 passes, 1.04 when throwing 3-5, and 1.31 points with 6+ passes.

For comparison’s sake, during Duke’s two ACC Tournament games, the Blue Devils threw 0-2 passes 45 percent of possessions and 6+ only 16 percent. Virginia is patient when it comes to shot selection, constantly setting screens to free up an open man.

During its three victories to win the ACC Tournament – including a win over the Tar Heels in the title game – Virginia led all teams with an average of 1.28 points per possession when setting a screen. It set 46 screens for Jerome according to STATS SportVU tracking, resulting in an average of 1.43 team points per possession. SportVU also compiles who sets specific screens, and the list below shows Virginia’s top three combinations involved Jerome during the ACC Tournament (click to enlarge).

A No. 1 seed has never lost to a No. 16 seed in the NCAA Tournament, and Virginia’s body of work suggests UMBC doesn’t stand a chance. Some might say UMBC needs to get extremely lucky. It so happens KenPom defines luck in college basketball as “a measure of deviation between a team’s actual winning percentage and what one would expect from its game-by-game efficiencies.”

Which team ranks No. 1 in KenPom’s Luck Rating? UMBC. And that luck seems to be kicking in already. Virginia announced Tuesday that ACC Sixth Man of the Year De’Andre Hunter suffered a broken wrist in the conference tournament and will miss March Madness.

If Virginia can overcome the UMBC luck factor and Hunter’s injury, it’ll still have to survive a bracket featuring No. 2 seed Cincinnati, third-seeded Tennessee and fourth-seeded Arizona, which seems awfully low for a team with a talent like Deandre Ayton.

Duke (26-7)

KenPom rating: 3
Seed: No. 2, Midwest Region
First-round opponent: Iona

The preseason No. 1 team is back to being a national title favorite despite finishing just 2-2 down the stretch. The Blue Devils certainly appear to have the most talent with freshman Marvin Bagley III – who is No. 3 in KenPom’s national player of the year standings – and senior instigator Grayson Allen, along with a supporting cast that features lottery pick Wendell Carter Jr.

Duke is one of only two teams to finish in the top 10 in both KenPom’s adjusted offensive efficiency (third) and defensive efficiency (seventh), with Michigan State – the No. 3 seed in the Midwest – being the other. The Blue Devils can beat any team inside or out given their shooters (Allen, Gary Trent Jr.) and their front line (Bagley, Carter).

STATS SportVU tracked four of Duke’s games this season, with two at the ACC Tournament and one each at Indiana and at Wake Forest. In those games, Bagley and Carter combined for 81 close touches (originated from a pass within 12 feet of the basket), leading to 1.26 team points per possession. Carter’s free-throw rate (free throws made/field goals attempted) was 64.3 percent on close touches and he shot 78.6 percent after a close touch.

The inside domination doesn’t stop there. According to SportVU, Bagley shot 12 of 14 when contested at the rim by two or more defenders, and Carter went 4 of 5 in those same situations. With only a single defender, Bagley was 8 of 14 and Carter 7 of 10. Essentially, if the ball gets inside, forget about it.

And if Allen is on his game from deep, the opposition stands little chance – unless defenders can keep him moving. In Duke’s tracked SportVU games, Allen shot 5 of 15 (33.3 percent) on non catch-and-shoot 3s. When he caught and fired? A red-hot 8 for 13 (61.5 percent).

Allen is at his best when he doesn’t have to create. When taking zero dribbles, Allen shot 11 of 18 overall and 9 of 14 from the 3-point line.

As for Trent? He’d rather be on the move. Trent went 11 of 22 (50.0 percent) from the field when taking at least one dribble in Duke’s four games SportVU tracked. When he didn’t take a single dribble, he went 7 for 23 (30.4 percent) overall and 4 of 17 (23.5 percent) from deep. Have a look at Allen’s and Trent’s numbers from the ACC Tournament (click to enlarge):

Bagley, Carter, Allen and Trent all played well in the ACC Tournament semifinals, but the Blue Devils ran into the Tar Heels, who came away with a 74-69 victory. Now the defending national champions are seeking a repeat.

North Carolina (25-10)

KenPom rating: 7
Seed: No. 2, West Region
First-round opponent: Lipscomb

A 10-loss No. 2 seed that finished sixth in its own conference with a home loss to Wofford on its resume? Crazier things have happened, even though North Carolina is 3-3 in its last six games. But a run to the ACC Tournament title game convinced the committee big things are in store for the Heels.

Theo Pinson, Luke Maye – who is No. 9 in KenPom’s national player of the year standings – and last year’s NCAA Tournament’s Most Outstanding Player Joel Berry II are the nucleus from a title team looking to make another run. And since the Tar Heels ranked 104th in the nation in 3-point percentage, they’ll go as far as their attack-the-hoop game takes them.

In North Carolina’s four ACC Tournament games tracked by SportVU, Pinson scored 16 points on a team-high 17 drives, which are defined as touches that start at least 20 feet from the basket and end within 10 – excluding fast breaks. Those attacks led to 26 team points, more than any other UNC player who drove to the hoop in Brooklyn.

But Berry and Pittsburgh graduate transfer Cameron Johnson had success, too. Have a look below at the team points per drive for each player, according to SportVU (click to enlarge):

All three teams have their strengths, but anything can happen during March Madness. STATS SportVU gave us a deeper look into ACC powers, and the advanced metrics listed above are just a sliver of the data collected during the conference tournament.

Imagine if we could evaluate every college basketball team the same way with SportVU player tracking.

Analyzing Purdue vs. Michigan State with Advanced Metrics


A breakdown of the first-ever top-five meeting between the Boilermakers and Spartans using STATS research and KenPom’s advanced analytics powered by STATS’ college basketball data

Identifying an advantage for either team in a marquee matchup between national title contenders can be an exhausting task that often leads one to settle for the basics.

Fourth-ranked Michigan State hosts No. 3 Purdue on Saturday in the only clash of the season between the Big Ten juggernauts. We can give the Spartans the edge and point to the fact the Boilermakers haven’t beaten a top-five team in a true road game since winning at No. 3 Michigan on March 6, 1994, especially since Glenn Robinson isn’t suiting up this weekend to provide an impact.

It’s easy to counter that Purdue is 7-0 in true road games this season – and that the previous 24-plus years of Purdue basketball has no impact on this matchup. But then again, the quality of competition hasn’t been near what it’ll face against the Spartans. The basics only take you as far as a winding circle most of the time.

Enter STATS’ college basketball data, which exclusively powers the model behind, the most in-depth and respected advanced analytics website in the sport. STATS’ data API pumps information into Ken Pomeroy’s complex formulas that calculate the advanced metrics referenced by some of the best college basketball analysts in the country.

KenPom’s model gives Michigan State a 56 percent chance to beat Purdue and has a projected final score of 75-74 – essentially even with the Spartans’ home-court advantage throwing the overall edge their way. Let’s get into the data and break down the matchup further, shall we?

When Michigan State has the ball

Tom Izzo teams make a point of getting the ball across halfcourt quickly with strong outlet passes and attacking if the opportunity is there. Just as often, the Spartans gather themselves and run through an organized offense. Michigan State’s ranks 74th in the nation with an average time per possession of 16.4 seconds, and it is 194th with a 68.0 adjusted tempo, which measures the average amount of possessions per 40 minutes.

The Spartans wouldn’t earn an assist on 68.3 percent of made baskets – a percentage that is tops in the country – if they ran frivolously. And sophomore Cassius Winston is the floor general helping Michigan State rank 10th in offensive efficiency with 119.9 points per 100 possessions.

Winston’s 44.5 percent assist rate – assists divided by the number of field goals made by teammates while he’s on the floor – leads the Big Ten and ranks third in the country. He also averages a conference-best 7 assists per game while playing just 27 minutes. Winston’s 10.5 assists per 40 minutes lead the nation and are three more than second place – teammate Tum Tum Nairn, who averages 7.3 per 40.

Miles Bridges gets most of the scoring credit given his team-high 17.7 points per game. However, Winston’s 12.2 scoring average, a 52.5 shooting percentage from 3-point range that ranks second in the country, and a 65.4 effective field-goal percentage that is second best in the conference – combined with the aforementioned assist rates – show he can do it all. Winston’s 129.0 offensive rating, which factors in total possessions and points produced among many other variables within a complicated formula, leads the Big Ten among all players used on at least 20 percent of team possessions.

Cassius Winston is Michigan State’s floor general. (Graphics by Stephan van Niekerk)

And if Winston and the Spartans are having an off shooting night? Nick Ward leads the country with an 18.9 offensive rebounding percentage, which takes into account possible team offensive rebounds, minutes played and defensive rebounds allowed.

How do the Boilermakers counter the Spartans’ offensive efficiency? Purdue’s 94.7 adjust defensive efficiency – points allowed per 100 opponent possessions – ranks 13th nationally, and its 99.5 mark in Big Ten play is second in the conference. It also ranks seventh nationally holding opponents to 43.5 percent on two-point field goals and 10th in the country in effective field-goal percentage defense at 45.5.

Despite the Boilermakers’ 19-game winning streak ending with Wednesday’s 64-63 home loss to Ohio State, they finished with a 96.5 adjusted defensive efficiency. The problems arose mainly on offense, which was a surprise considering how well Purdue has performed on that side of the ball this season.

When Purdue has the ball

Purdue’s 12 turnovers in the loss to Ohio State are its most in nine games after combining for 11 over the two games prior. The Boilermakers had a 98.2 percent win probability with just over 10 minutes remaining, according to KenPom.

That advantage quickly dissipated, and Purdue’s 95.0 final offensive efficiency rating was only the 14th-best rating against Ohio State this season. Vincent Edwards went 1 for 9 on two-pointers, leading to a season-worst 72 offensive rating. Among Big Ten players with at least 24 percent of possessions used, Edwards’ 119.8 offensive rating is fifth, sandwiched between Ward and Bridges.

It was an unexpectedly poor offensive performance for the Boilermakers despite another solid effort from Carsen Edwards, who finished with a 148 offensive rating after scoring a career-high 28 points while going 8 of 13 from the field, 4 of 7 from deep and 8 of 9 from the line. Purdue ranks third in the nation in adjusted offense at 123.5, which is nearly 30 points better than Wednesday.

The scoreless performance from P.J. Thompson didn’t help. Thompson has the Big Ten’s highest offensive rating at 133.6, ranks third with a 63.7 effective field-goal percentage and sits fifth at 46.5 percent from 3-point range, but he went 0 for 5 against the Buckeyes with every shot coming from behind the arc.

Purdue should just chalk it up as one tough performance and move on quickly. Michigan State’s two-point percentage defense ranks first in the nation at 37.9, but it is 69th in 3-point percentage defense at 32.9. Purdue is second in the country shooting 42.7 percent from 3-point range, and its 38.4 percent of total points coming from the 3 in conference play is the most in the Big Ten.

Also, the Spartans beat Iowa 96-93 on Tuesday, but it wasn’t anywhere near pretty. It’s the most points they’ve allowed in a regulation game since losing 98-63 to North Carolina on Dec. 3, 2008, at Ford Field. Iowa’s 124.2 offensive rating makes it the most efficient game a team has played on offense against MSU this season. The Hawkeyes are 12-14 overall, 3-10 in the conference and lost by 23 to Purdue on Jan. 20.


Saturday will boil down to which team can expose the other’s weaknesses more during an important game for the Big Ten title race – and it likely will be decided outside the paint.

Michigan State freshman phenom Jaren Jackson Jr. ranks second in the Big Ten with a 15.44 block percentage, with Purdue’s Matt Haarms not far behind at 14.77. The best bet for either team is to get them back to the bench quickly. Jackson’s 5.72 fouls committed for 40 minutes is the worst rate in the conference, and Haarms also is towards the bottom at 5.36.

The advanced metrics show two evenly matched teams going through struggles in different aspects of their games. Michigan State has the slight advantage after running the numbers.

STATS NBA 2017-18: An Advanced Look Into The Offseason Shakeup


The ground has most certainly shaken in the East with the most unlikely of trades between conference frontrunners.

Kyrie Irving and Gordon Hayward have landed in Boston to join forces with the latter’s former coach. LeBron James has been dealt an unlikely top-three scorer.

Transcontinentally, 2016-17’s DIYers James Harden and Russell Westbrook have been given some serious help. Or is it the other way around: Chris Paul, Paul George and Carmelo Anthony still don’t have a true last name among them, but are they the main beneficiaries?

Yeah, there’s a Jimmy Butler-Tom Thibodeau reunion on in Minnesota, but the Timberwolves also brought in a point guard who may better suit their bigs.

Lob City is no more, but what it ultimately comes down to is everyone’s still chasing Dub City. Come along with STATS and the most reliable and fastest NBA data available as we take a deep diagnostic dive to weave select stories of how the pursuit may go down in 2017-18.

Player Tracking and Deep Learning Innovations Lead to Greater Interactions with Data


Fans regularly consume STATS’ data without knowing the original source of the information, and that isn’t necessarily changing as the company explores future applications of richer data. What is different in 2017 from 1987, 1997 or even 2007 is STATS’ specialization from often drawing the line at providing sports information to more properly contextualizing it for clients.

One of the leaders making that happen at STATS emerged from behind the scenes to speak with Sirius XM Wharton Moneyball hosts Adi Wyner and Shane Jensen earlier this month. Patrick Lucey, STATS Director of Data Science, reflected on five years of STATS SportVU tracking data collection and provided insight on the answers now coming from growing sample sizes and industry advances.

STATS began tracking NBA games with its optical system SportVU half a decade ago, which happened to coincide with what Lucey marked as a particularly interesting time in sports science.

“Something happened in 2012. Deep learning came along,” said Lucey, co-author of the 2016 MIT Sloan Sports Analytics Conference first-place research paper on patterns of player movement and ball striking in tennis.

“It’s like a perfect storm. I like to think of it as a three-legged stool. First of all, for deep learning to work, you need three things. The first thing is you need a lot of data. Luckily at STATS, we have that. The second thing is you need the computational power. … The third thing is having the people to actually know how to work or set up the network.”

For nearly two years, Lucey has sat atop that stool at STATS and helped expand player tracking into additional sports with an emphasis on the global game of football. Plenty of his current work, including his 2017 MIT Sloan Conference runner-up paper on shooting styles in basketball, has involved finding the proper ways to ask and answer the what-if questions that weren’t confrontable in STATS’ earlier years.

“The big thing – and this is what we can do really well – we can contextualize the data,” Lucey said. “…You can ask us specific questions. You can say, ‘Well, what’s the likelihood of this player making that shot? Or, what happens if I switch that player with another player?’

“So, we can really do fine-grain simulations and ask those what-if questions. … We have the box score or we have these players’ stats, but (I think the next stage in analytics) is to simulate these kinds of specific plays and see what a different player will do in these situations. In a given situation, we can model that context and we can give more precise answers because we understand the data in those situations.”

The particulars of body pose have become one of his key interests. Specifically, advances in player tracking – and the potential for more accurate predictive analysis – lies in an increased emphasis on measurements that can go beyond using a basic center of mass to establish the XY coordinates in optical tracking.

“Once upon a time, we could capture body pose, but we had to do that in a lab setting,” Lucey said. “… Now we can actually do it in the wild.”

Lucey’s team is constantly honing in on the next big thing in its field, and his view for how performance will be evaluated more closely in the future might surprise some. It’s not necessarily about creating fresh analytical metrics. It’s about the relationship between people and technology.

“I don’t actually think new metrics is the way to go,” Lucey said. “I don’t think that’s the future. I think it’s a symbiosis between a human and a computer. So can we develop new technology to help a domain expert do their job better. I think that’s really the next step in sports analytics. Just enabling, creating this kind of technology just to help coaches or analysts or people at home to be able to ask these what-if questions. I don’t think we’re that far off.”

Tough Shots No Problem for Cavs’ Smith in NBA Finals


Irving, Durant Also Stand Out in STATS SportVU’s Tough Shot Score Metric

There are many ways to measure the greatness of players like LeBron James, Kevin Durant and Kyrie Irving, but perhaps the simplest involves no advanced stat at all.

They make it look easy. Durant gliding down the left wing and stopping on a dime for a pull-up 3. James darting into the lane and spinning past his defender for a lay-up in one silky smooth motion. Irving leaving a would-be road block in his dust after a series of crossovers and using the perfect amount of English to will the ball in off the backboard.

It’s not.

STATS has created a metric called Tough Shot Score with the use of its revolutionary STATS SportVU cameras that measure how difficult a given shot is based on a variety of factors: the position of the defender, where the shooter is on the floor and whether he’s moving or stationary. The score is normalized from 0-100, with 100 being the most difficult shot possible. Here’s a little of what we found out about just how tough some of the NBA Finals’ biggest stars had it on basketball’s grandest stage.


Watch J.R. Smith play for any five-minute stretch of his career and it’s obvious that he doesn’t mind taking tough shots. It’s almost like he prefers them, often dribbling into what seems to be a more difficult, suddenly contested shot than what he had when he caught the pass.

Of the eight top scorers in the Finals, Smith’s regular-season Tough Shot Score (0.61) was easily the highest. So it shouldn’t be surprising that among the 11 players in the Finals with 20 field-goal attempts, Smith’s TSS (0.62) was the only one above 0.58.

Player Total FGAs Tough Shot Score
Smith (CLE) 36 0.62
K. Thompson (GS) 70 0.58
Curry (GS) 87 0.57
Irving (CLE) 123 0.54
Love (CLE) 67 0.54
Durant (GS) 106 0.54
James (CLE) 117 0.50
Green (GS) 55 0.50
Iguodala (GS) 34 0.49
Livingston (GS) 28 0.47
T. Thompson (CLE) 22 0.44


Not that the level of difficulty seemed to bother him. After going scoreless from 10:26 of the first quarter of Game 1 until Game 3, Smith was unconscious in the series’ final three games, going 17 of 27 from 3-point range. That includes going 7 for 13 when his defender was contesting within 0-2 feet (very tight) or 2-4 feet (tight).

There’s one player in the series who took more tightly contested 3s than Smith, and he happened to be the one who wound up raising the Finals MVP trophy.

Durant took 312 3s during the regular season, less than a third of which were tightly contested. He hit just 31.4 percent of those, a big drop from the 40.3 percent he hit when open (4-6 feet) or wide open (6-plus feet).

In the Finals? Durant was masterful in transition and was a huge factor defensively, but there was one major reason he wound up sharing a handshake with MVP namesake Bill Russell at the end of Game 5. Of the 38 3s Durant took against the Cavs, 24 of those (63.2 percent) were tightly contested. He made 13 of them, including the Game 3 dagger over James that gave Golden State a lead it would never relinquish. It stands out as the series’ signature moment. All that added up to was a 0.65 TSS from 3-point range, which was actually a shade under his regular-season TSS from beyond the arc (0.66).

Irving had more than a few eye-popping performances in the series, going 7 of 12 from 3 in the Cavs’ lone win. But it was what he did around the basket – particularly in Cleveland’s Game 3 loss – that really stood out. Irving went 14 of 28 on very tightly contested 2s in the series, giving him 11 more attempts with a defender draped all over him than James and two more than the two most heavily guarded Warriors (Durant, Draymond Green) combined.

Only Eric Bledsoe and the Splash Brothers themselves shot better than 50 percent on 2s when tightly contested during the regular season, so for Irving to hit that mark in the Finals while wearing Klay Thompson like a compression shirt is quite a feat. Yet his TSS from inside the arc was higher during the season (0.51) than in the Finals (0.49), so Irving’s Finals’ circus act wasn’t really much different than what he’s been doing since October.

Brett Huston is a Senior Editor at STATS LLC. Contact him at or on Twitter at @BHuston_STATS.

2017 NBA Finals Preview


The top of any list of great sequels in cinema could easily double as a list of some of the finest – and most financially successful – movies ever made. The Empire Strikes Back, The Godfather: Part II, The Dark Knight, Terminator II, Aliens – we could go on.

That’s typically where the creative juices stop flowing.

Sure, there are some noteworthy third acts. The Return of the King is the best Lord of the Rings movie, though that’s more properly viewed as one colossal installment instead of three smaller ones. Indiana Jones and the Last Crusade both made up for the weirdly terrifying Temple of Doom and was popular enough for Harrison Ford to keep playing the character as a senior citizen. Goldfinger was arguably the best James Bond movie.

There haven’t been many third acts in the sports world, which is just one small reason why Cavs-Warriors: Part III should be so compelling. The Lakers and Celtics have met in 12 Finals, but never three in a row. It’s a first for the NBA and just the fourth such threequel in American professional sports history, the likes of which haven’t been seen since the Red Wings and Canadiens battled for three straight Stanley Cups in the 50s.

We’ve already detailed the lack of big-screen triumphs when it comes to third acts, but the success of screens could have everything to do with a Finals that’s rightly drawing as much hype as anything since the days of Jordan, Magic and Bird. Let’s take a look at what the Cavs need to do to repeat and what the Warriors can do to make these Finals more Rise of the Machines after last year’s epic Judgment Day.

When you have 30 percent of the All-Stars from a few months ago in one Finals series, it’s easy to get excited about matchups. Will Steph Curry or Klay Thompson guard Kyrie Irving? Does LeBron have no choice but to spend most of his time checking Kevin Durant? We know Draymond can cause issues for Kevin Love, but can Love be matched up with him on the other end?

Those seven will all find one another at some point, but the winner of this rubber match figures to be the team that consistently creates, and then takes advantage of, the most opportune mismatches.

That the Warriors move the ball and move away from the ball better than any team in the league is no great secret. It’s what almost every team in the league dreams of emulating and one day building themselves. Setting screens is still a major part of an offense that hums like a Ferrari when it’s at its peak, but ball screens are a different story. Golden State set 3,324 of those in 2016-17, per STATS SportVU, its third straight season bringing up the NBA rear in the category. Orlando was 29th, yet the Magic set more than 4,000.

There’s less movement in Cleveland’s offense because it’s less necessary. Possessions can come to a screeching halt in the final 10 seconds of the shot clock and Irving and James can save the day as few individual players can. Irving was the best isolation player in the league this season at 1.12 points per ISO, and he’s been even better in the playoffs. James can’t blow by defenders 1-on-1 like he used to, but his 42 percent success rate from 3 in the postseason adds a more complicated wrinkle for opponents than Batman suddenly wielding an assault rifle would for Gotham miscreants.

The Cavs relied on ball screens to generate offense less this season than they had in the past two, funneling through around 57 per game instead of the 65 or so they’d used in James’ first two seasons back home. Whatever way you slice it – and given the overall levels of talent and execution, this shouldn’t be a surprise – both Cleveland and Golden State get a lot out of their screens. Individually, the Warriors were third in the regular season in points per play (a screen that results in a field-goal attempt, foul or turnover by either the ballhandler or screener) at 0.934; the Cavs were fifth (0.925). As far as team points per possession – this adds in the other three offensive players on the floor as potential factors post-screen – Golden State was fourth (1.12), a tick ahead of Cleveland (1.11).

But those numbers take into account Derrick Williams setting a pick for Kay Felder on a cold February night in Minnesota or James Michael McAdoo trying to free up Patrick McCaw on a November back-to-back in Milwaukee. Let’s eliminate some of the noise and concentrate on what both teams should be focusing on – and what they must work to avoid at all costs.

Golden State DO: Get Curry/Draymond rolling

There was no Love in the 2015 Finals and there was no Irving for the final five games, and while those two are a generally dubious defending combination on ball screens, the Curry/Green combo likely couldn’t have done much better if they were both on the floor. Those six games featured 85 Green screens for Curry, which resulted in the Warriors eviscerating the Cavs defense for an average of 1.26 points.

Fast forward to 2016 and it turns ugly for Golden State. Seven games, a total of 39 Curry/Green ball screens and just 0.78 team points per action.

Green is fronting for Curry 6.7 times per game in these playoffs with excellent results: 1.30 team PPP. If that number stays in that vicinity – like it did two years ago – start sizing up the Warriors for their rings, and perhaps 16-0.

Cleveland DO: Target Curry when he’s guarding the screener

Irving has been known to struggle when he’s checking the ballhandler in the pick and roll, often never finding his original man or the roller and easily providing the opposition with a 2-on-1 toward the hoop. But that Irving/Love combo we discussed a few paragraphs ago? They actually defended quite well when put on an island in the 2016 Finals. There were 19 ballhandler/screener combos that defended at least 10 screens last June, and Irving/Love was by far the MOST effective despite getting torched overall in the postseason (1.31 PPP). Irving fared pretty well in the Finals when paired with Tristan Thompson as well.

Ballhandler Screener Screens Defended Team PPP
Irving Love 39 0.59
K. Thompson Green 12 0.73
Iguodala Livingston 14 0.77
Livingston Green 11 0.78
Irving T. Thompson 41 0.85
Curry Bogut 18 1.00

As for Curry, Cleveland preferred to have whomever he was guarding set the pick for the ballhandler. With Curry already banged-up to some degree in the Finals, the Cavs were physical while guarding him and made him work overtime at the other end. Curry was involved in 88 screens as the screener, nearly 50 more than the Warriors made Irving take on. A look at the difference in how both point guards were attacked in the pick and roll in last year’s sequel:

Player Screens starting on ballhandler Team PPP Screens starting on screener Team PPP
Irving 118 0.805 33 1.06
Curry 68 1.096 82 1.07

On Christmas Day in Cleveland, the Cavs ran Curry through nine more with him initially on the screener, scoring 12 points. Klay Thompson was the targeted on-ball defender – often on Irving – with Cleveland putting him through 28 screens and scoring 40 points. Overall, the Cavs celebrated their comeback win at The Q with 75 total points (1.19 team PPP) as the result of screens – 51 more than Golden State (0.71).

Cleveland DON’T: Let Iman Shumpert get screened into submission

The Warriors’ holiday in Northeast Ohio may have been dampened, but they took out seven months’ worth of frustration on the Cavs three weeks later in Oakland. Golden State used 46 ball screens in this one and particularly attacked Iman Shumpert on the ballhandler, often when he was checking Curry. Ten screens of Shumpert led to 22 Warriors points, further lending credence to this stat: In the 128 minutes Shumpert was on the floor in the 2016 Finals, the Cavs were outscored by 13.4 points per 100 possessions. In the 208 he sat, Cleveland enjoyed a plus-9.1 edge.

Golden State could drive Shumpert off the floor entirely in these Finals. In theory, he’s an ideal guy to stick on Curry or Thompson to hide Irving for a bit, but in reality he tends to get lost when he’s asked to do more than guard someone 1-on-1. Richard Jefferson played a key role against the Warriors last season and seems more suited to have a chance of defending Durant than Shumpert. With Kyle Korver a potentially vital offensive piece to stretch the floor, Shumpert may wind up a DNP-CD (can’t defend).

Golden State DON’T: Ignore Kevin Durant as a ballhandler

Let’s get to the elephant in the room of why many expect this series to be short. The Warriors added one of the three best players in basketball at the expense of Harrison Barnes, who went 5 for 32 from the field once Golden State went up 3-1 last year.

As we’ve covered, the Warriors aren’t going to rely nearly as much on the ball screen as the Cavs. But when things start to break down – particularly in the fourth quarter – there will be instances when it could be a necessity.

Logic tends to dictate that should a critical Golden State possession become bogged down, Durant will ISO, Curry will launch a 3 or, perhaps, Durant will come to the ball and screen for Curry. But there’s another option.

Durant has an awfully good handle himself. Curry screening for him should allow KD a moment to turn the corner and pop away from the secondary defender for an open 3. And if Curry can’t get free, Durant proved during the regular season that he was fantastic finishing in these situations. Of the 144 players who participated in 300 screens as the ballhandler, only Wilson Chandler and Paul George scored more points per individual screen than Durant (0.48).

It’s been even more absurd during the playoffs. Durant’s 0.66 average is a full tenth of a point better than any of the other 46 players who have participated in at least 50 screens. From a team perspective, the Warriors’ 1.36 PPP off screens with Durant as the ballhandler is second – and the chart below shows how infrequently that’s used compared to some of the other big names at the top.

Ballhandler Screens Team PPP
Stephenson (IND) 56 1.38
Durant (GS) 77 1.36
James (CLE) 271 1.35
Leonard (SA) 222 1.28
Curry (GS) 256 1.25


They’ve only broken the Curry-screening-for-Durant combo out 13 times during the playoffs but it’s led to 21 Warriors points, and frankly, there was no need to even do it that much. It’s a wrinkle that Steve Kerr and Mike Brown have largely been saving to unleash only when they need it, and that alone should terrify the Cavs.

Cleveland and Golden State DO: Get the big men involved

There have been 72 two-man combos that have run at least 30 screens in the playoffs, and the top two involve, as you might expect, James and Curry. But the other half of those equations probably isn’t who you’d expect. JaVale McGee has teamed up with Curry for 52 screens that have resulted in 1.47 Warriors PPP, tied with James and Tristan Thompson for the most effective in the league this postseason.

The James and Thompson combo has been a special kind of deadly on their 112 screens. James has hit 7 of 14 3s directly after Thompson frees him up, and the duo is 31 of 53 (58.5 percent) overall immediately after Thompson screens for James. Thompson is one of the league’s best at rolling off a screen and flushing an alley-oop from James or Irving, and he and James went for an impressive 1.14 PPP in last season’s Finals as well.

There you have it. There’s no shortage of storylines in the most star-studded Finals since Lakers-Celtics was in its mid-80s heyday. Durant’s chasing his first title. LeBron is chasing MJ’s legacy. Curry and Green are seeking Finals redemption. Klay Thompson wants to prove his subpar playoffs so far have been a fluke. Love wants to show that he can play – and play effectively – against the Warriors.

Golden State knows what’s coming. It’s up to the Warriors to keep Cleveland from catching them in bad ball screen combos while picking and choosing their own spots to use them in an offense that rarely does.

In a third act worthy of the big screen, we’re about to find out how big the screen can be.

Applying 3D Modeling to Player Performance


STATS, the world leader in sports intelligence, has created a new tracking and analysis method that has powerful applications in terms of determining the future success of sports teams. Using 3D mapping to plot players’ every movement, each shot, step, dunk, or pass can be captured in detail not seen before. This data tells the full story of the game – which position had the highest likelihood of scoring points, why a certain style was successful or how a blocked pass could be avoided in the future.

The current STATS SportVU Basketball Data system reveals data on the types of plays teams run, and estimates on their future success. Six cameras are installed throughout an arena, capturing the position of players and the ball a remarkable 25 times per second. However, the new 3D model can provide an even deeper look. It can provide key insights into each player’s signature movements across a variety of plays by accounting for differing physical attributes as well as tracking what moves are unique to each player.

This 3D modeling approach can also equip STATS to flag potential athlete injuries. By using models created specifically from previous player behavior, researchers can identify when a player’s technique falls outside of the expected model and could possibly lead to harm. After an injury has occurred, the same method can be used by tracking if a recovering player is mimicking his previous actions.

The data that can be gathered using this method compared to earlier methods could be thought of as the difference between a standard pedometer and a Fitbit. Our previous ability to track and analyze information provided a solid, streamlined picture on what mattered. Now, we can expand into an entire new level of understanding and assessment on the performance of players and what this means for their teams’ success.

Analyzing a Potential NBA Finals Rematch


Before the season began, the Warriors (86 percent) and the Cavs (48 percent) had the best chances of earning the top seeds in their respective conferences, per FiveThirtyEight. As of February 17, right before All-Star weekend, both teams’ probabilities had increased to 95 percent and 77 percent. The highly desired NBA Finals rematch was looking promising.

The Cavs, of course, ended up finishing second after dropping their final four regular-season games, twice resting LeBron James and seemingly ceding that top spot to the Celtics. Unexpected, sure, but hardly earth-shattering considering they went .500 in their final 46 games.

Perhaps the most shocking FiveThirtyEight statistic suggests that, going into the playoffs, Cleveland’s chances of making it to the Finals stood at a mere 11 percent (to Golden State’s 66). Betting markets were far more bullish on the Cavs, giving them approximately a 75 percent chance to make it out of the East despite their underwhelming regular season.

The Cavs have doubled their FiveThirtyEight chances after sweeping the Pacers in Round 1, but they’re still the East’s third-likeliest team to make the Finals (through Monday’s games), behind a pair of teams that are far from a sure thing to make it out of their opening series:

  • Celtics (26%)
  • Raptors (23%)
  • Cavaliers (22%)
  • Wizards (16%)

Bet you didn’t see that one coming.

Next up is either Toronto or Milwaukee, though the Raptors swung things considerably in their favor with Monday’s 118-93 rout. Should the Cavs be rooting for a playoff rematch against Toronto or the young Bucks to break through?

Young and inexperienced, Milwaukee has given Toronto fits at times with its athleticism on defense, posting a 100.1 defensive rating that’s the playoffs’ second best behind only the Warriors’. The return of Khris Middleton – who didn’t face the Cavs in the regular season – mixed with the ever-growing confidence of Giannis Antetokounmpo could hand Cleveland’s already-shaky defense some problems if the Bucks advance.

Antetokounmpo played all four regular-season games against the Cavs and averaged 24.0 points while getting to the free-throw line 43 times. He had 34 points, 12 rebounds, five assists and five steals – Anthony Davis was the only other player to post such a line – in Milwaukee’s only win in the series, which also happened to be the lone time LeBron James, Kyrie Irving and Kevin Love all played.

The defender who gave the Greek Freak the most issues might surprise you.

Offense Defense Games Matchup Time Points FGM-FGA FG% FTA Drives Drive Points
G. Antetokounmpo Richard Jefferson 4 15:29 6 2-12 16.7 4 4 1
G. Antetokounmpo LeBron James 4 9:20 11 3-6 50.0 4 3 2
G. Antetokounmpo Iman Shumpert 4 7:12 9 2-3 66.7 5 5 5
G. Antetokounmpo Derrick Williams 1 3:52 4 2-4 50.0 0 2 2
G. Antetokounmpo J.R. Smith 2 3:32 9 2-3 66.7 6 2 3
G. Antetokounmpo Tristan Thompson 4 3:31 9 4-8 50.0 1 3 4
G. Antetokounmpo Kyrie Irving 4 3:07 6 2-3 66.7 6 5 5

Richard Jefferson fell out of the Cavs’ rotation by the end of their series against the Pacers, but it’ll be interesting to see if Tyronn Lue brings him back in a potential Bucks matchup.

Jefferson also did a nice job against DeMar DeRozan as the Cavs won three of four against Toronto. So did the man who saw the majority of Jefferson’s minutes in the first round, Iman Shumpert.

Offense Defense Games Matchup Time  Points FGM FGA FG% FTA Drives Drive Points
DeMar DeRozan J.R. Smith 2 9:08 10 4 9 44.4 2 6 4
DeMar DeRozan Richard Jefferson 3 7:43 18 6 15 40.0 8 9 5
DeMar DeRozan LeBron James 3 6:47 2 1 6 16.7 0 3 2
DeMar DeRozan Iman Shumpert 3 4:51 2 1 10 10.0 0 4 2
DeMar DeRozan Tristan Thompson 3 3:07 8 4 10 40.0 0 6 4
DeMar DeRozan Kevin Love 3 3:02 14 7 12 58.3 0 9 4
DeMar DeRozan Kyrie Irving 3 2:34 9 2 6 33.3 6 4 2

Considering Cleveland has yet to see a Milwaukee team with Middleton, and hasn’t faced the full-strength current iteration of the Raptors – Serge Ibaka and P.J. Tucker weren’t around for the Cavs’ three victories – the second round may be the toughest obstacle for James and Co. on the path to a third straight Finals.

Golden State’s biggest test isn’t likely to come until the conference finals. Both the Jazz and Clippers are dealing with key injuries, but looking down the road to a matchup with Houston’s high-octane 3-point variance or the Spurs’ balanced machine reveals a more complicated mission.

After combining for 140 wins last season, the Spurs and Warriors never did get to meet in the playoffs. That might change in Round 3 this time and, if it does, the league’s two best defenses will be on full display.

In two games against Golden State, Kawhi Leonard guarded Draymond Green for a span of 6 1/2 minutes, holding him to a mere two points.

Offense Defense Games Matchup Time Points FGM FGA FG% FTA Drives Drive Points
Draymond Green Kawhi Leonard 2 6:26 2 1 4 25.0 0 4 0
Stephen Curry Tony Parker 2 6:02 8 3 5 60.0 0 1 0
Klay Thompson Danny Green 1 4:54 5 2 2 100.0 0 0 0
Kevin Durant Kyle Anderson 1 3:03 11 4 4 100.0 2 1 2

Draymond dominated on the other end of the floor, however, allowing LaMarcus Aldridge to scored just 11 points in 17 minutes of matchup time. Likewise, Andre Iguodala held Kawhi to eight points in 9 1/2 minutes.

Offense Defense Games Matchup Time Points FGM FGA FG% FTA Drives Drive Points
LaMarcus Aldridge Draymond Green 2 17:29 11 4 10 40.0 2 0 0
Kawhi Leonard Andre Iguodala 2 9:24 8 3 6 50.0 2 3 0
Tony Parker Stephen Curry 2 8:18 9 4 5 80.0 0 5 2
Pau Gasol Zaza Pachulia 3 7:34 7 2 4 50.0 2 1 0

It’s hard to read too much into those numbers considering of the three meetings, the only one in which each team had a fully functional roster took place on opening night in October – a 29-point Spurs win at Oracle Arena. It took five months, but the Warriors got revenge in San Antonio, coming from 22 down to beat the Spurs 110-98 on March 29 even without Kevin Durant.

Vegas considers a Warriors-Cavs rubber match a near certainty by exact matchup standards, giving it a 59 percent chance of happening after both teams’ opening sweeps. There are still some stars that need to align, but it looks even more probable now than it did prior to the playoffs.

It’s certainly more likely than a team blowing a 3-1 lead in the Finals again.

Photos By: AP Photo/Ben Margot/Darron Cummings

Why the Final Four Shouldn’t Be Such a Shock


The NCAA Tournament is always full of surprises, and this year has been no exception. However, even with two No. 1 seeds and one No. 3 seed in the Final Four, this round of March Madness has somehow proven to be one of the most unpredictable.

Per ESPN’s Tournament Challenge, only 0.003 percent of submissions correctly guessed the  Final Four. Last year, three times that fraction of brackets were picked correctly even though there was a No. 10 seed (Syracuse) crashing the party.

Year Seeds % Correct
2017 (1) North Carolina, (1) Gonzaga, (3) Oregon, (7) South Carolina 0.003%
2016 (1) North Carolina, (2) Villanova, (2) Oklahoma, (10) Syracuse 0.009%
2015 (1) Kentucky, (1) Wisconsin, (1) Duke, (7) Michigan State 1.360%
2014 (1) Florida, (2) Wisconsin, (7) Connecticut, (8) Kentucky 0.006%
2013 (1) Louisville, (4) Michigan, (4) Syracuse, (9) Wichita State 0%
2012 (1) Kentucky, (2) Ohio State, (2) Kansas, (4) Louisville 0.220%
2011 (3) Connecticut, (4) Kentucky, (8) Butler, (11) VCU 0%

In 2014, there was a No. 7 seed and a No. 8 seed in the Final Four, and 0.006 percent (double this year’s share) of people perfected their selections. Why is this combination such a surprise?

The answer is all in the brand. As of late, the two years that produced zero flawless Final Four predictions were the two years that included mid-majors (Wichita State in 2013, Butler in 2011).  Few had even heard of Wichita State (no, Wichita is still not a state) much less picked them to make that sort of run – remember, the Shockers’ 34-0 regular season wasn’t until the year after they made the Final Four. And that 2013 roster wound up having three future NBA players.

Butler was a slightly different story. The Bulldogs had already made it to the national championship game in 2010 and headed into the 2011 tourney having won nine straight. Yet Brad Stevens’ team wasn’t even favored to make it past Old Dominion in the first round, let alone be a member of the Final Four. Hardly a household name to fans then, even if it’s become one since.

Still, these projections somewhat made sense on paper, given that Wichita State and Butler both would have to beat a No. 1 seed to even make the Sweet 16. Going into this year’s tournament, these were the chances of each remaining team making it to the Final Four, according to FiveThirtyEight:

Team % Chance
Gonzaga (1) 41.5%
North Carolina (1) 29.9%
Oregon (3) 6.6%
South Carolina (7) 1.1%

Gonzaga had the highest probability in the entire tournament pool of making it to the Final Four, yet just 37 percent of brackets put them there.

Back to brand. Especially when money is on the line, people are most comfortable choosing teams that have an established name. In other words, people are most comfortable choosing teams that have high seedings, flashy players, and measurable amounts of experience, regardless of anything else that may be relevant.

North Carolina checks all the criteria on that list. It leads the NCAA in Final Four appearances (20) and ranks second in total tournament appearances (48). Think about it. UNC has made it to the Final Four in over 40 percent of its total tournament trips. So, not only did the Tar Heels enter as a No. 1 seed, but one could say that, well … they’ve been here before. Throw in likely lottery pick Justin Jackson, who has averaged 19.8 points, 6.3 rebounds, and 4.3 assists through the Elite Eight, and it’s no surprise that 45 percent of brackets picked UNC to make it this far.

Gonzaga, on the other hand, has been in the Big Dance 20 times and has never played in a single Final Four game. But they received a No. 1 seed for a reason. The Zags suffered just one loss in the regular season, posting a points per game differential of plus-23.4 – the best in Division I since Duke’s 1998-99 juggernaut that featured five lottery picks. The next best this season was Wichita State at 19.6.

Point differential matters here. To win 30-plus games in a season is no easy task, but to win them by that much is truly historic. Gonzaga’s schedule was no cakewalk, either, containing fellow NCAA Tournament teams in Iowa State, Florida, Arizona and Saint Mary’s (three times). Still, people chose brand, so more trust was put into teams like Duke (40%), Arizona (45%), Villanova (48%), and Kansas (58%, the most commonly selected Final Four team via CBS).

Just 9 percent of brackets placed Oregon in their Final Four, compared to 27 percent in favor of fellow-No. 3-seed UCLA despite the Ducks finishing ahead of the Bruins in the Pac-12. Part of that could have had to do with the season-ending injury to big man Chris Boucher, but there are perhaps two bigger reasons. One, UCLA has 48 tournament appearances to Oregon’s 15, including 18 Final Fours – all of which came well after the Ducks’ lone previous trip, in 1939. And second, Lonzo Ball.

We can’t forget South Carolina, which is easily the biggest reason for such bracket mayhem. Sure, they’re a No. 7 seed, they have zero Final Four experience and don’t have a single player who’s a surefire first-round pick. Sure, they had a one percent chance of making it to the Final Four. Sure, just 0.2 percent of nearly 19 million people picked them to make it this far. But should we be this surprised that they did?

The Gamecocks had regular-season wins over Michigan, Syracuse and Florida and has a top-10 KenPom defense that’s one of the most aggressive in the country. They have the SEC player of the year in Sindarius Thornwell. And they have a coach with this perspective (if you have the time, listen to the full 8-minute interview – it’s worth your while).

Maybe it’s not such a surprise. Maybe it’s time that “brand” gets re-branded.

Photos By: AP Photo/Charlie Riedel/Young Kwak/Julio Cortez/Gerry Broome

Impact Percentage: A New NBA Fingerprint


No two teams in the NBA are alike when it comes to the mixture of player tendencies, lineup combinations and styles of play. One metric that tells a story of player involvement (and furthermore, can serve as a team’s fingerprint) is usage percentage. This estimates the portion of team plays used by a player while he was on the floor. In other words, it shows us how often a player ends his team’s possessions. Players like Russell Westbrook, DeMarcus Cousins, and DeMar DeRozan currently lead the league in usage percentage.

Using our revolutionary STATS SportVU data, we have recently developed an improvement upon traditional usage percentage. Now, not only can we tell how frequently a player terminates a possession (with a FGA, FTA, or Turnover), but we can also quantify how often they impact a possession (with a drive, a ball screen, an isolation or a post up). We’ve appropriately named this value impact percentage.

Why is this valuable? Consider players like Goran Dragic and Mike Conley, who rank 38th and 49th, respectively, in usage. Although a glance at these numbers might persuade one to think that these two aren’t chief possession-influencers, they are both among the top 10 in the league in impact percentage.

This new statistic was mentioned in ESPN the Magazine’s newest analytics issue, which hit the stands on March 17. Today, we are going to use it to analyze Monday’s matchup between Oklahoma City and Golden State.

As mentioned earlier, Russell Westbrook leads the NBA in usage percentage – but he also sits at the top of the leaderboard in impact percentage. When you look at OKC exclusively, no one comes close to the amount of impact that Westbrook has on their possessions.


Usage plays a part here, but when you break down OKC’s SportVU plays – drives, isolations, post ups, and ball screens – it’s evident that these are where the bulk of Westbrook’s impact lies. Westbrook accounts for over 51 impactful plays per game on average. The next closest Thunder player is Victor Oladipo at 16.


Let’s talk about variability. In statistics, standard deviation measures how spread out a distribution is. A low standard deviation tells us that most of the numbers in a sample are close to the sample’s average – in other words, there isn’t much spread. A high standard deviation tells us that the numbers are more scattered. Unsurprisingly, due to Westbrook’s outlandish numbers, OKC has a very high standard deviation when considering impact percentage.

The Thunder’s standard deviation comes to a whopping 15 percent. In comparison, Golden State, before Kevin Durant’s MCL sprain on February 28th, had a standard deviation of 8 percent.


The biggest and most obvious takeaway here is that these are two completely diverse teams when it comes to how the ball is facilitated, and further, who the offense revolves around. Yet, a deeper dive into the Warriors’ Impact numbers reveals that they’ve had to adjust their strategy in KD’s recent absence.


Steph Curry’s impact percentage has increased from an already team-leading 49 percent to an even higher 55 percent while Klay Thompson’s has shot from 31 to 37. This is all expected. What’s more compelling, though, is how a guy like Ian Clark is doing exponentially more with his limited minutes.

In his first eight games* since Durant’s injury, Clark has nearly doubled his ball screen usage (2.6 to 4.8 per game) and increased his drives (1.2 to 1.9 per game). On top of that, his minutes slightly decreased – even with him playing 34 minutes March 11 against the Spurs as Steve Kerr rested all his stars.

It’s important to mention that impact percentage is not necessarily a reflection of efficiency. Just because a player is influencing his on-court possessions doesn’t mean that he’s influencing them positively. It’s no secret that the Warriors are struggling to fill Durant’s shoes, especially when it comes to productivity – but it does appear as if guys are at least stepping up to the challenge and getting involved.

With all of this in mind, it’ll be interesting to see how OKC handles a Golden State team that finally seems to be adjusting to Durant’s absence.

*all numbers are as of 3/17/17

Photo By: AP Photo/Alonzo Adams
Illustration By: STATS/Andrew Skweres