Posts

Kingdom of Kevin: Why Durant is the NBA’s first-half MVP

If it seems like a basketball lifetime ago that Kevin Durant, Russell Westbrook and James Harden all shared the same hardwood, that’s because it was.

It’ll be five years this June since one of the most talented Big 3s we’ve ever seen disbanded before the NBA would truly grasp the potential that Oklahoma City trio had, none having reached his 24th birthday prior to getting the little brother treatment from LeBron James and friends in the 2012 Finals. Harden was shipped to Houston before the following season began, and after four years of not quite getting over the hump together, the Durant/Westbrook duo broke up with the former’s decision to bolt for the Bay Area.

Each finally has a franchise to pilot on his own, and in the case of Westbrook and Harden, that’s being taken quite literally. Jilted at the altar by Durant, Westbrook has channeled his manic on-court energy into a one-man show the league has rarely seen. His 41.1 usage rate is the highest since the league started keeping track in 1997-98 – Michael Jordan’s swan song from relevant hoopdom (sorry, Wizards). The season Kobe averaged 35 for a mediocre Lakers team? The era of Iverson’s “practice” rant? Both positively passive compared to what Russ is doing.

Harden is just three spots behind, using 34.3 percent of the Rockets’ possessions while leading the league in minutes. As he continues to rack up assists in his new role as Houston’s point guard, Harden has touched the ball 5,730 times this season. Only Westbrook (5,504) is within 1,000.

Either has a perfectly good case as the league’s MVP. Westbrook has kept the Thunder firmly in the Western Conference playoff picture by himself. He’s averaging a triple-double. He’s pulling down basically the same number of rebounds per night as DeMarcus Cousins while carrying a team that scores 106.6 points per 100 possession with him on the court and just 97.2 – think last season’s 10-72 Sixers level – when he’s on the bench.

Prior to New Year’s Eve, no player in NBA history had recorded a 50-point triple-double. Before the end of January, Harden had done it twice. Until 2016-17, there had been five seasons in NBA history where a player had averaged double figures in both assists and free-throw attempts – all by Oscar Robertson in the 60s. Westbrook and Harden are on pace to give the Big O company for the first time in a half century.

And neither should be the NBA’s midseason MVP.

Durant left a good situation in Oklahoma City for a historically great one in Golden State, and while there’s not going to be a follow-up 73-win season for the Warriors, make no mistake – this is a better team than the regular-season juggernaut of 2015-16. They’ve been 12.6 points better than their opponents per 100 possessions, a full point above last year’s 9-loss squad and 3.6 ahead of the league’s next-best team (San Antonio) this season. For as incredible as Golden State was last season, the Spurs had a better point differential.

It wouldn’t have taken a fortune teller to figure out that Durant would score less for the Dubs than he did with the Thunder, but he’s still leading Golden State by averaging 25.8 points – just 2.4 fewer than he did last season on 2.2 fewer shots. Durant is producing 1.52 points per field-goal attempt, second best in the league other than the DeAndre Jordan/Rudy Gobert/Dwight Howard troika which rarely takes a shot outside the restricted area.

Who’s the only one ahead of Durant? That would be Harden (1.55), but the Rockets’ star is getting there with a 52.5 effective field-goal percentage. Durant’s at 59.5. His true shooting percentage of 65.2 is the best of his career, better than any non-big other than the Wizards’ Otto Porter.

Durant’s shooting 37.4 percent from 3-point range, which while far from a bad number is his lowest since 2010-11. But consider what he’s doing from inside the arc. As NBA.com’s John Schumann points out, he’s finishing at an elite level both inside the paint and from mid range.

Schumann_graphic

What Durant has done when he drives to the basket separates him from any MVP candidate, LeBron James included. Seventy-five players in the league have driven toward the hoop at least 200 times. The only player scoring more than one point per drive is Durant, and he’s nearly a full quarter of a point (1.14) ahead of No. 2 Tobias Harris. Last season, no one was higher than 0.88 per drive – Durant himself.

Perhaps an even better measure is team points per drive, which takes into account more than just the individual’s finishes. Durant’s head and shoulders above the rest of the league here, too, with a top eight that’s basically a who’s who of NBA megastars.

Drives Team Points Per Drive
1. Kevin Durant (GSW)
235
1.55
2. Chris Paul (LAC) 212 1.36
3. LeBron James (CLE) 490 1.35
4. James Harden (HOU) 628 1.32
5. Stephen Curry (GSW) 336 1.32
6. DeMar DeRozan (TOR) 520 1.31
7. Kyle Lowry (TOR) 565 1.30
8. Jimmy Butler (CHI) 473 1.29

In the two years STATS SportVU data on drives has been fully available, the top finishers were Harden (1.37 in 2014-15) and Curry (1.38 in 2015-16).

Durant is shooting 72.4 percent when meeting resistance at the rim, tops in the league among 108 players with at least 100 contested field goals. Finishing in the restricted area overall? 78.1 percent, 2.7 above LeBron James at No. 2 and miles better than any big (Howard, Jordan, Whiteside, et al) who makes his living inside the few feet around the basket.

It only seemed logical that moving to Golden State’s ball-movement favoring, constant motion offense and leaving behind Oklahoma City’s ISO-heavy sets would decrease Durant’s need to create by himself once a play breaks down. And it has. Durant had the seventh-most ISOs in the league last season, going at his defender 1-on-1 on 9.2 percent of his possessions. That percentage is down to 6.8 with the Warriors – 19th in the NBA – but he’s been even a tick more effective, scoring 1.00 points per ISO after putting up 0.99 with the Thunder. Westbrook (0.90) and Harden (0.88), who ranked first and fourth in total ISOs, are considerably behind.

They’re not as far back of Durant there as they are in transition, however. Let’s start by pointing out that the three teams we’re looking at are the three who most frequent the fast break. Let’s continue by mentioning that there are 36 players in the league, as of the All-Star break, to attempt at least 100 field goals in transition. Durant happens to cash in more often than any of them, averaging 1.34 points per transition bucket while Westbrook (0.99) and Harden (0.97) sit at 31st and 32nd. Part of the reason? Durant rarely coughs up the basketball.

Transition turnover percentage Rank (out of 36 qualifiers)
Durant
9.4
25th
Westbrook 21.1 2nd
Harden 29.2 1st

Neither Durant, Westbrook nor Harden has the reputation as a lockdown defender, and it’s still difficult to find a reliable all-encompassing defensive statistic to go by. ESPN’s defensive real plus-minus, measured in net point differential over 100 offensive and defensive possessions while adjusting for teammates and opponents, has Durant ninth among small forwards (1.89), Westbrook 19th among point guards (-0.09) and Harden 72nd among shooting guards (-1.72), a position he doesn’t even really play. A lot of noise there, too hard to draw a huge conclusion.

But remember how effective Durant is when he drives to the basket? He’s been nearly as good when he’s the one defending the drive. Last season, of the 126 players to stand in front of at least 200 drives, Durant ranked 106th while allowing 1.22 team points per drive. As we inch toward the three-quarters mark of the 2016-17 season, let’s use 150 as a minimum threshold. With the Warriors, Durant is seventh of 117 qualifiers at 1.04.

That’s just one example, but Durant has taken on the challenge of protecting the rim after going from a team that had multiple great defensive options inside to one without any particularly good ones. His 1.7 blocks per game and easily a career best and he’s accounted for 36.4 percent of the Warriors’ blocks overall. He’s defending slightly more post plays per game than he did in OKC and he’s doing it well. Durant’s limiting the player posting him up to 0.40 points per post, ninth best in the NBA of the 60 players to defend at least 75. And consider the company. Marc Gasol is giving up the exact same number. Teammate Draymond Green is at 0.41. Likely defensive player of the year Rudy Gobert is at 0.59. Does that mean Durant is a defender on par with those three overall? No. He’s had roughly two-thirds of the amount of post-up defensive opportunities as Gasol, Green and Gobert. But does it mean Durant can hold his own on key possessions down low against the likes of Kawhi Leonard, Blake Griffin, Kevin Love and James come late May? Quite possibly.

Consider one other part of his game that doesn’t get a ton of credit. The Warriors are the league’s third-best team from behind the 3-point arc, shooting 38.8 percent. Golden State is shooting 41.2 percent on 3s off passes from Draymond Green, 39.9 percent from Curry and 37.7 percent from Andre Iguodala, their first-, second- and fourth-ranked assisters overall. On passes from Durant, they’re shooting 47.9 percent. Curry’s percentage on passes from other Warriors is 40.3, and on Durant dishes it kicks up to 49.5. Klay Thompson goes from a 41.7 percent shooter from deep on passes from non-KD teammates to a 53.7 deadeye when firing off a feed from No. 35.

Westbrook and Harden have been fantastic this season, the NBA’s two most overwhelming forces lifting what are likely lottery teams without them to playoff squads (and, in Houston’s case, home-court advantage) with them. But Westbrook has the ball in his hands more than a quarter of the time he’s on the floor. Harden’s a smidge under the 25 percent mark.

Sure, that’s their job. Ball dominance shouldn’t preclude a player from being the league’s MVP. But Durant is finding a way to take over games while having the ball in his hands just 7.5 percent of the 34 minutes a night he plays. He’s been the best player on a team with the two-time reigning MVP, a team that’s statistically even better than last season’s regular-season behemoth.

Durant won’t lead the league in scoring, rebounding or assists and he won’t turn in lines every night that would make Oscar Robertson blush. The Warriors don’t need him to. What they do need from Durant has been delivered on a higher plane than any other player in the league. And that’s why he’s the NBA’s midseason MVP.

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

Photos By: AP Photo/George Bridges/Sue Ogrocki/Marcio Jose Sanchez
Illustration By: STATS/Andrew Skweres

Mitigating Risk: Analytics and Player Recruitment

Player recruitment is an important strategic pillar for any professional sports team and is an area in which the use of data can have a transformational impact.

Over the last decade, English Premier League teams have spent £900 million on players that have gone on to play less than 50% of possible minutes at their new club. In fact, only 34% of signings have played more than 50% of minutes, a figure that illustrates the costly inefficiencies in the transfer market.

As hundreds of thousands of players ply their trade in leagues around the world, it is virtually impossible for clubs to efficiently monitor prospects across the globe while operating within their budgets using only a traditional scouting network. However, the application of objective analysis methods can have a profoundly positive impact on the efficiency and cost of player recruitment strategies.

Taking The First Step

When clubs come to Prozone for an objective angle on particular recruitment issues, they general want to do at least one of three things:

  1. Benchmark transfer targets against existing squad members.
  1. Find players similar to X (e.g. if they are looking for a direct replacement for a player they are likely to lose).
  1. Look for players with specific characteristics.

As a first step in assisting with requests of this nature, the Prozone Performance.LAB is able to conduct a principle component analysis that enables players to be searched across multiple variables.

The results are then processed to create similarity measures that highlight the levels of correlation between a player or group of players and the desired characteristics. The resulting data can then be plotted to identify the players that are the best fit with the requirements of the buying club and is a useful starting point in the technical scouting process.

Portability of Talent

Of course, scouting is about far more than raw numbers. Once a group of players has been identified according to statistical criteria, it is essential that clubs look at broader and sometimes less quantifiable questions.

One of the major issues is the portability of talent between specific regions or leagues. A leading example of this has been the transfers of forwards from the Dutch Eredivisie to the English Premier League, with players such as Luis Suarez and Robin Van Persie flourishing while others have been extremely disappointing despite good goalscoring records in the Netherlands.

Before signing a player from a different league, clubs seek to understand several things. First, they need to analyse how performance in the league the player currently plays in translates to the competition they will be joining. Then they need to understand how repeatable their performance is likely to be and whether long-term performance can be reliably predicted.

A common method for approximating the portability of talent is through ELO rankings. A system originally developed for chess to determine player ability, the ELO methodology can be used to determine the strength of a specific team or competition. Highlighting the teams that play at a similar level to the recruiting club, ELO rankings provide an extra layer of information and are a useful first step in terms of quantifying the likely portability of talent.

Taking the scouting process to the next level, it is possible to use Prozone’s playing styles analysis to identify teams and players that play in a similar way to the buying club. Quantifying and defining tactical characteristics, playing styles gives clubs the opportunity to find players from leagues and teams that fit with their own profile, thereby enhancing the chances of the transfer target fitting with the system.

Age Matters

When considering signing a player, it is vital that clubs take into account the age of the target and factor in how they are likely to perform in the future. By analysing the usage rates of players in different age categories, Prozone can build a model to demonstrate the development of a Premier League signing in terms of minutes on the field.

Player EVO - Player Recruitment

Interestingly, within four seasons of being signed, players in the under 21, 21-14 and 25-29 age brackets all record very similar rates of usage. For quick results, however, it is clear that buying players in the 25-29 category is best in the short-term. If clubs are building for the future with a long-term strategy, under 21 players from abroad (represented by the yellow dotted line) generally outperform domestic talent in the same age bracket over a seven-year period.

That being said, the best performing category of all is 22-24 year old players signed from British clubs (solid green line). Acclimatising to the Premier League quickly, those players record almost 50% usage by their fourth season and go on to record the highest average usage by year seven.

This model is not a guarantee that those players will always perform to an elite level, but it provides unique information that can usefully inform recruitment processes and further mitigate the risks involved with transfers.

There will always be inherent risks in the signing of new players and there is no magic bullet that can ensure the success of new acquisitions, but the advancement of analytics is helping clubs to be better informed than ever before as they seek to gain a competitive advantage in the transfer market.

Missed our player recruitment webinar? Click here to receive a full recording of the session straight to your inbox.

Recordings of all previous webinars are available on our website where you can also sign up to our newsletter for the latest analytics news.

Interested in speaking to our sales team about recruitment analytics for your club? Click here.

Game 7: Theory vs. Reality

Miami Heat forward LeBron James faces off against Oklahoma City Thunder forward Kevin Durant in the first half of their NBA basketball game in Oklahoma City.

Photo by Bill Waugh/REUTERS

In introductory statistics classes, it is not uncommon for a teacher to ask students to find the probability of an evenly match, best-of-seven series, to go the full 7 games. There’s a method of calculating the odds using a series of binomials, which yields the theoretical probability of going the distance in the series. This method ultimately yields a 31.25% chance that a series ends in a deciding game seven.

The theoretical approach has its limitations, specifically in its evenly matched condition. While it is true that in any given series both teams have a chance to win any given game, a first round series featuring the regular season conference champion, and a lowly eight seed, is far from even. In fact since the NBA switch to all 7-games series in 2003, over 36% of 1-8 opening round matchups end in 4 games, which is triple the expected proportion. So when making predictions about how long a series will last, how does the experimental value compare with the theory?

I have compiled all of the playoff matchups since 2003, and examined how reality catches up to the theory as teams move deeper into the playoff.

The Theory

While traditional mathematical approach that requires a series of binomials may seem intimidating, there is actually a very simple way of approaching this problem that only requires some basic math.

Let’s suppose we have two teams, team A and team B, and that both teams have a 50% chance of winning any given game in the series. There are two ways that this series can end in 4 games: team A can sweep or team B can. The chance that team A wins the series in four games can be calculated by multiplying the chance they win one game (.5) four times, which is .54, which equals .0625. Since team B has exactly the same chance of sweeping the series, we simply multiply the previous value by two to compute the theoretical chance of the series ending in four games.  This yields a .125, or a 12.5% chance of the series ending in four games.

While there is only one way for each team to win a series in four games, there are several for a team to win in five games. For team A to win in five games, team B needs to win one of the games, and this can happen four ways. There four combinations are as follows: BAAAA, ABAAA, AABAA, and AAABA. Each capital letter represents who won each game, so in the first example team B won the first game and team A won the next four. Please note that AAAAB is not a possibility because that series would just end in four games. Theoretically speaking, all four of these possibilities have an equal chance of happening. The chance of any one of these results can be calculated by multiplying the chance that each team wins one game (.5) five times, which is .55, which equals .03125. Multiplying this value by the four ways to win gives us the chance that team A win in 5 games, which is .125. Doubling this gives us the chance that team A or team B wins in 5 games, and that value is .25 or 25%.

Now we know that there is a 37.5% chance that a series finishes in four or five games. The only way for a series to go a game 6 or 7, is for one team to be leading 3-2 after five games. Let’s now say that team A was in the situation, leading 3-2 after game five.  They have 50-50 chance of winning game six and ending the series, and losing and forcing a decisive game seven. This means that the chances that the series ends in six or seven games is equal, and if they process is repeated with team B leading 3-2 after five, the results will be duplicated. Since the chances of the series ending in six or seven games are equal, the chance the series wins in six or seven games can be calculated by finding probability that the series does not end in five games and dividing by two. The change that the series goes past five games is one minus the chance it ends in four or five games, which is 1 – .375 or 62.5%. The chance that the series ends in six games is half of that or 31.25%, and the chance that it ends in seven is the same. The results are summarized below.

[table “2262” not found /]

The Reality

As I have already explained, the assumption that both teams have the same chance of winning a series in an NBA, best-of-seven series is far from valid, especially in the early rounds. Not so surprisingly, the opening round has the highest variation in relative strength. In the 1-8 matchups, the winning team is hardly in question, while in the 4-5 matchups, anything can happen. In total the 1-8 matchup, ended in four or five games nearly 60 % of the time, which is over 50% more than expected theoretical value. The 4-5 matchup was the opposite: over 77.28% of the matchups went longer than five games, which is significantly closer to the theoretical value. The first round in total is as follows.

[table “2265” not found /]
[table “2266” not found /]
[table “2267” not found /]
[table “2268” not found /]
[table “2270” not found /]

The second round series in total was far more competitive than the first round, with only 37.6% of matchups ending four or five games. The frequencies for the number of  6 game and 7 game series were nearly identical in the Eastern and Western Conference, which shows that over the 10 year stretch, the talent distribution amongst the top teams is about the same in each league.

It’s worth noting that in all four rounds, more series end in game six than in any other individual game. The second round is no exception with 36.4% of matchups going to a game six, which is nearly identical to the first round proportion. Whereas the first round was skewed towards the shorter series, the second round is far more evenly distributed about game 6.

The data for the second round is as follows:

[table “2273” not found /]
[table “2269” not found /]
[table “2274” not found /]

For every round we move deeper into the playoffs, there is half the number of data points, due to the nature of the single elimination playoff system. The law of large numbers states that as there are more and more data points, the experimental results will approach the theoretical values. Even though we have narrowed the number of game to which we can collect data from, we can still make some observations.

The proportion for series ending in four or five games had less than a 10% error, but for games 6 and 7 the results were extremely skewed. Of the twenty-two conference finals series from 2003, twelve ended in six games, and only 3 ended in seven. It is worth noting, however, that the proportion of games that end in short series (4 or 5 games) and the proportion of games that end in long series (6 or 7 games), it is very close to the expected probability.

The data for the conference finals is as follows:

[table “2271” not found /]

The Finals, despite having the fewest number of data points, was a close it could possibly be to the theoretical values. Once again there were more series than ended in game six than any other game, but it was only one more than number of games that ended in game seven. Only 1 series ended in 4 games, which is as close as the experimental results could be to the expected 1.375 series.

The data for the finals is as follows:

[table “2272” not found /]

Conclusion

As expected, as we moved deeper into the playoffs the actual values I collected approached the theoretical values we calculated. The first round demonstrated how not all teams have the same chance of winning any given game, and finals showed how evenly matched championship series can be.

The other interesting conclusion is number of series that ended in game six. Whether it was the mentality of teams to close out the series to avoid game seven, or the nature of the home field advantage in the playoffs, game six was the most common close out game.

This year as you’re watching the playoffs, be sure to see how the duration of the series changes as teams move towards the finals, and see how many series really do end in game six.

How Important Are Rebounds?

Favors-and-Kanter_Jim-Young-REUTERS-e1383337557649[1]

Photo by Jim Young/REUTERS

APBRmetrician, Dean Oliver postulated that a team’s ability to win games could be boiled down to a simple set of just four statistics, which he calls the four factors. The four factors, shooting, turnovers, rebounding, and free throws, were somewhat crudely assigned weights: 40%, 25%, 20%, and 15% respectively in order to analyze a player’s value. It is well known that shooting is extremely important when determining the value of a player, but what other statistics are really that important?

On the surface, it would seem that rebounds would be important, after all rebounds create possessions, and more possessions mean more points. Oliver specifies that just looking at rebounds alone doesn’t tell the whole story. Obviously a 10-rebound game in a 70-possession game is more important than a 10-rebound game in a 100-rebound game. For this reason, Oliver uses rebound percentage rather than total rebounds to measure a players effectiveness, and this point I do not disagree with. I believe, however, that rebounds could be a compounded variable.

Offensive rebounds allow a team to gain an extra possession before the opposing team has a chance to score. Although this may seem attractive for a coach, some teams like the San Antonio Spurs forgo the opportunity to get the board to allow the big men to get back on defense. The Spurs had an offensive rebound percentage of 8.1 percent, which was tied for worst in the NBA last season. Other teams, like Indiana, that have a dominant front court, have much higher rebounding percentages.

Offensive rebounds can also be influenced by shot selection. If a team in general shoots a higher percentage from the floor, the big men are more likely to box out since there is a higher probability that they will be given the opportunity to get a rebound. Teams that shoot a higher percentage from the floor often have their big men run back since there is a smaller chance at an opportunity for an offensive board. We see in the graph below that there are two peaks in the data, one with teams that have few wins like New Orleans, and one with winning teams like Denver.

offence

We see a similar trend with defensive rebounds because, like offensive rebounds, defensive rebounds can be compounded with shooting. We know teams that shoot a lower field goal percentage are less likely to win, and since defensive rebounds happen at a much higher rate than offensive rebounds, winning teams often have a high number of defensive rebounds, and this is shown in the graph. However, if you look at the graph you see a few peaks in the teams with lower win totals. This is could be because of team strategy. As I explained earlier, teams like the Spurs rarely go for the offensive rebounds so they can get back on defense. This allows bad teams to collect the defensive rebounds. This explains the two peaks we see in the graph below.

defense

When we look at the graph of wins vs. total rebounds, we see that the double peak feature of the graph becomes more profound.

total

We can conclude that rebounds cannot be as significant as Dean Oliver suggested because the teams that excel in rebounding are at the extremes. Rebounding is probably a compounded variable with shooting and team strategy, and it is simply not as important as we once thought.

The Significance of Basketball Statistics

 

Photo by Steve Mitchell/USA TODAY Sports

Photo by Steve Mitchell/USA TODAY Sports

Since the beginning of the statistical revolution, the number of statistics and metrics has skyrocketed, to the point where the amount of data at our fingertips is simply overwhelming. We have evolved from a simple box score with point, assist, and rebound totals, to an overwhelming pool of statistics at our disposal measuring everything from block percentage to win shares. But what exactly do these numbers tell us? What use is it to have all these numbers and not know which ones actually matter?

When we watched Gregg Popovich’s Spurs and Erik Spoelstra’s Heat, we saw perhaps the league’s two most dominant teams face off in a series for the ages. The teams have contrasting styles: where as the Spurs relied on pick and roll offence from their big men and Tony Parker, the Heat utilized the big three as primary ball handlers, with a wealth of shooters on the perimeter. But other than their ability to reach the finals, these teams much have shared some characteristics, right?

I decided to investigate which box score statistics had a significant correlation to a team’s ability to make the playoffs. Pulling data from the previous five seasons, ranging from the 2008-2009 season to the 2012-2013 season, I was able to identify some common characteristics of the playoff teams.

To analyze the data I used a Z-test to see if each statistic for playoffs teams was significantly higher than the average. Anything with a p-value below .05 is significant. The results for shooting based statistics is as follows:

Statistic

P-value

Significant?

Statistic

P-value

Significant?

Points

0.013440444

Yes

3-Point %

0.004730563

Yes

Field Goals Made

0.099288557

No

Free-Throws Made

0.075157925

No

Field Goals Attempted

0.997532259

No

Free-Throws Attempted

0.056351965

No

Field Goal %

2.44206E-05

Yes

Free-Throw Percentage

0.550878918

No

3-Pointers Made

0.075122268

No

Points Per Shot

4.26039E-05

Yes

3-pointers Attempted

0.170776882

No

Adjusted Field Goal Percentage

1.51501E-05

Yes

 

What does this tell us? Notice that the number of shots taken and the number of shots made, both for field goals and for three point shots, had no significant difference between the playoff teams and non-playoff teams. Any difference between the two groups can be attributed to natural variation, and not a difference in style or skill. Additionally, free throws had no statistical significance whatsoever. What were significant, however, were the efficiency ratings. This makes sense intuitively: one would think a team that makes a higher percentage of shots would have a better chance of winning. This point may seem trivial, but when we’re comparing of the best players in the game, how often is adjusted field goal percentage the first stat that’s brought up?

Possession based statistics, like assists and rebounds, are trickier to analyze from box score statistics because, for example, a five steal game in a 70-possession game is much more significant than a 5 steal game in a 100-possession game. Statistics like assist or rebound percentage may give us more insight in some instances. With this being said, I decided to look at 5 statistics that I thought could still provide statistical insight. The results are as follows:

Statistic

P-value

Significant?

Assists

0.078203253

No

Steals

0.045952507

Yes

Blocks

0.034938444

Yes

Turnovers

0.001775388

Yes

Assist/Turnovers

0.002723667

Yes

 

The fact that steals were significantly higher and that turnovers were significantly lower for playoff teams is perfectly reasonable, because when a team turns the ball over, they not only forfeit their own possession, they give an extra possession to the opposing team. Coaches have preached the importance of ball security, and these numbers back it up.

Assists/turnovers was shown to be higher for playoff teams, but the number of assists was not. Playoff teams are often lead by ball handlers, like Carmelo Anthony or Kobe Bryant, that create their own shots without the help of a teammate, so it is no surprise that assists were not significant. The fact that assists to turnover ratio was higher for the playoff teams is simply due to just how significant turnovers were.

Blocks were significant, but this is most likely and example of correlation and not causation. Since the playoff teams block more shots, the other team misses more shots, which ultimately contributes to their victories. Block percentage would have probably given more insight as to how significant blocks are.

What this data shows us is that efficiency is key. Whether it’s knocking down corner threes, tipping in the put-back lay in, or capitalizing on possessions with ball security, efficiency ratings are the most effective way of measuring a team’s ability to achieve success. So the next time you’re debating LeBron vs. Durant, don’t forget which statistics really matter.

Statistics powered by STATS Basketball Analytics Solutions.