clock menu more-arrow no yes

Filed under:

A Statistical Framework For Team Building

Updated on 5/22/13.

Christian Petersen

After tying the mark for all-time women's professional basketball coaching wins after a win against the Chicago Sky on September 11, 2011, Seattle Storm coach and GM Brian Agler offered brief insight related to what it takes to build a winning team.

"When we first got this opportunity I thought Sue and Lauren could be the best at their positions in the world, so what kind of people and players are going to be a good match with those guys?" Agler said. "Obviously, they're both quality people but they also are finesse players so you need to put toughness and tenacity and physical play and people to do the dirty around them. That's what we have and everybody plays and accepts their role. The longer we've played the tighter we've gotten."

Although building around Bird and (a healthy) Jackson might sound relatively simple formula for success to many, Agler was confronted with a more difficult team building challenge this WNBA off-season.

Faced with salary cap constraints and aging stars, it was obvious that the Storm had to make changes. Yet compounding the challenge was the desire to remain competitive in the present - with star center Lauren Jackson missing half the season due to the Olympics - while simultaneously building for a sustainable future. To do that, Agler pursued a combination of experience and youth this off-season.

It's a delicate balance and, as Agler suggested in an article by Jayda Evans of the Seattle Times, what he's attempting to do is something that's very difficult in any professional sport.

"What we're trying to do is very difficult - maintain a high level of competitive drive and success by striving to win championships but yet still get younger with the roster," said Agler, whose Storm has reached the WNBA playoffs a league record-tying eight consecutive times. "Those two things really don't work together sometimes. We're trying to do it."

Bird and Jackson are clearly still the focal points of the Storm's plans this season, but they're clearly preparing to move in a new direction. Only time will truly tell how well Agler did balancing competitiveness with building for the future, but it does raise the question

So that raises the question of what goes into building a competitive team and how we might evaluate the off-season changes of Agler's or any team. I have my thoughts on the Storm's path that I'll share (sooner than) later, but the following is the framework I'm going to use for season previews that I've described before and is essentially all described in our statistics glossary. This is just an update to some of that thinking with the use of additional statistics that I didn't previously have available to me for the WNBA.

Four Factors strengths and weaknesses

Before scrutinizing the performances of a team's individuals, it actually helps to take stock of how the team performed as a whole.

Although win percentage, point differential, or defensive/offensive ratings are all strong indicators of success, for an analysis of how well a team is put together it really helps to have a sense of specific strengths and weaknesses relative to opponents. The best way to do that while still keeping an eye on wins and losses is to use Dean Oliver's Four Factors to create a statistical profile for a team.

Ken Pomeroy summarizes Oliver's Four Factors well and Ray Floriani has provided plenty of game examples throughout the WNBA and NCAA women's basketball season at Swish Appeal with his "stat analyses", but for now I want to revisit the matter of weighting each one of those to determine strengths and weaknesses.

The Four Factors are as follows: effective field goal percentage, turnover percentage, offensive rebounding, free throw rate. But as one might suspect, not all of those things are created equal. Take, for example, the Four Factors for this 59-49 win by the New York Liberty over the Chicago Sky on August 4, 2011 summarized statistically by Ray:

Team eFG% Tov% OReb% FT Rate
Sky 35% 23% 34% 31%
Liberty 41% 11% 29% 20%
NYL Differential +6% +12% -5% -11%

Four Factors for Liberty vs Sky on 8/4/11.

So, what do we make of this? On the surface of it, the game looks like a wash - each team "won" a category with each essentially dominating one and playing closer in another. But common sense might tell us that a few of those battles were more important than others, as described here previously.

Oliver has established weights to determine how important each of those Four Factors are that help shed some light on how the Liberty won that particular game, which you can find here. But to summarize, those weights are as follows:

  • Shooting efficiency: 10
  • Turnover percentage: -8.4
  • Offensive rebounding percentage: 4.1
  • Free throw rate: 2.0

So applying those weights to that game, we'd get the following weighted differentials to help us determine the most significant factors in winning the game:

eFG% Tov% Oreb% FTR
+60 +10.8 -20.1 -22

Weighted Four Factors differentials for the NYL in their win against Chi on 8/4/11.

In other words, although each team won two of four categories, the Liberty "won" more significant Four Factors and thus established more significant advantages that helped them win the game.


The same analysis above can be applied to looking at a team's performance relative to their opponents over the course of a season: championship teams tend to establish significant advantages in at least three of four categories relative to opponents, average teams in two of four significant categories, and poor teams have negative differentials in all four.

So the best teams are built to win as many of those categories as possible. And after looking at the team level, we can look at the individual players that make up a team.


In addition to balance on the team level, a team's versatility in terms of their ability to put combinations on the floor that can produce those Four Factors stats is rather significant to a team's success.

Another way to look at quality is relative to players' style of play - as important as it is to win those Four Factors with quality players, so is having the different types of players that are particularly adept at doing those things. For example, a team full of wings who can contribute to the Four Factors is nice, but they'll inevitably struggle when they facing a team with interior players that they don't match up with well. Championship rosters tend to have a mix of quality players across the spectrum of playing styles who allow them to adjust to different situations fluidly.

The SPI Playing Styles framework has already been described elsewhere and the importance of that to this discussion should be somewhat obvious: 11-player rosters have made versatility even more important in the WNBA in terms of both individual player skill sets and types of players.

So another important way to assess the quality of a team's roster is not only by standard metrics, but with a breakdown of whether each player is a) above average in each area of the Four Factors b) relative to other similar players by SPI Playing Style.


What might matter even more than balance and versatility is having players who can make high quality contributions to the Four Factors. Having four players who are capable of rebounding is not bad on an 11-player roster; having four rebounders as the 8th, 9th, 10th, and 11th players on the bench might not be a formula for success.

There are a number of ways to look at player productivity, but one that I'll rely on early on is 4-year Regularized Adjusted Plus Minus (RAPM). To oversimplify, 4-year RAPM numbers basically tell us how many points a player's team allowed and scored when that player was on the court independent of who they were playing with over a 4-year span.

The advantages of using RAPM over a four-year span have been described here at length (with soundtrack), but one major asset is that RAPM helps us figure out which players are productive defensively, something that is extremely difficult to measure otherwise. With all the player movement and personnel changes that happen over four years, it's a rather valuable statistic. Obviously, younger players haven't played for four years so I'll also use 1-year RAPM numbers when applicable.


Another key is that you want players that not only have a clear division of labor but a distributed one - a team that relies on one player heavily to produce one of the Four Factors is going to be easier to neutralize than one that has multiple contributors to that area (as you might guess, this is especially true of offensive rebounding, which is usually dominated by post players who get in foul trouble more often). Conversely, if you have multiple players that can contribute to the Four Factors, your roster is more adaptive than a team that relies heavily on one player to produce Four Factors stats.

However, the other challenge is getting a mix of players that come together to be "greater than the sum of their parts" as the old cliche goes. In other words, you want players that make each other better when playing as a unit instead of stepping on each other's toes, especially offensively. Synergy is definitely part of that: having players who are willing passers (high assist ratio) and able to create scoring opportunities for themselves (high usage rate). And you can click here for descriptions of how I've looked at that in the past using "synergy rating" and "adjusted synergy rating".

But what also matters is that each player does those things efficiently.

Players with high usage rates - meaning, they shoot a lot - will hopefully have good true shooting percentages for their style of play (meaning they create points efficiently). Players with below average usage rates should hopefully do other things on the court that as role players that give them a high floor percentage, if not scoring efficiently then creating assists and getting offensive rebounds to help generate scoring possessions.

A team that has both efficient scorers and efficient role "players" can really put a lot of pressure on a defense because more people on the court are "threats" to help the team create scoring possessions. A team without both isn't exactly lost, but might face bigger challenges. A team with a bunch of efficient high usage players could obviously be dangerous but there's only one ball on the court meaning they could stagnate if somebody isn't on the court to pass the ball. A team with a bunch of players who struggle to create their own shot but can score efficiently would have to have strong ball movement to be successful.

But an ideal team will have some sort of complementary mix of scorers and role players that maximize the potential of the whole.

The Structure of Championship Teams

The basic idea is that flexibility and productivity matter more than simply being "deep" in the WNBA. Realistically, most teams are only going to play 7-9 players consistently to begin with but have to find a way to manage injuries, fatigue, leaving little room for development unless those top rotation players are outstanding enough to fill the Four Factors on their own (e.g. Seattle, 2010).

In ideal terms, a team that can put five players on the floor that can contribute to the Four Factors at an above average level on the interior and perimeter while scoring points is better off than a team for which that contribution is concentrated in one place because it's easier to find a way to establish advantages against a range of opponents.

Over the last few years, Minnesota (2011), Seattle (2010) Indiana & Phoenix (2009) had rosters that exemplified all of these qualities entering the season in terms of the players who can contribute to those Four Factors. They all ended up in the WNBA Finals. Those squads stand in direct opposition to Atlanta (2010 & 11), which has been an interesting example of a team that had multiple players that contributed to the team's dominance in two of those Four Factors (offensive rebounding and turnover percentage), which ended up fueling impressive finishes to the season and consecutive runs to the WNBA Finals.

2012 will be the first year since the 11-player rosters have been implemented that we'll have an Olympic break prior to which some key foreign players will be absent, which could make this matter of roster flexibility - specifically the ability to make up for the absent player by productively changing style of play if not replacing them - even more important.

Percent valuable contributions

The best way to examine how much a team might lose due to a player absence is to look at percentage of valuable contributions for each player on the team or the weighted value of everything they contributed to the team from assists to missed shots. That's all described here, but this is essentially what that would look like (using the 2011 Tulsa Shock as an example):


In the case of teams that will be missing players due to injury or Olympic absence, this helps to understand how significant that loss is statistically.

Projecting player contributions

But along with knowing how valuable a player was to their team's success last year, we also might want to know how much value a player contributes relative to the proportion of team minutes they received. Valuable contributions ratio (VCR) helps to figure out a few things in that regard:

  • Whether a player is contributing quality minutes when they're on the court in terms of their statistical contribution relative to the percentage of team's minutes they receive.
  • Whether a player can be expected to play starter, rotation, or reserve-level minutes, even if they changed teams.
  • Players who are poised to have a breakout season in terms of how well they produced in the minutes they received in the previous season and whether they'll be able to maintain their productivity if given more minutes.
  • How much room rookies have for development.

Essentially, VCR answers the question: "How much more (or less) would this player contribute to the team if given more minutes, assuming they don't improve their game?" That's why VCR is also useful for determining the Most Improved Player award: if a player does improve their game in some way (e.g. basketball IQ, strength, skill, etc) the rate of statistical contribution relative to the proportion of team minutes they receive would increase. That's different than players averaging more because they played more - it's a measure of whether they did more damage in whatever time they did get.

Just win baby

That's a lot of information to digest at once, which probably makes it seem a lot more overwhelming that it really is (which is why it deserved its own post rather than trying to cram the explanations into the team analyses). As I said at the outset, it's really a list of stats that I use to look at the structure of teams that will be used as a reference point in the future.

So to summarize in the broad statistical terms, the teams with the broadest range of playing styles that can establish advantages by contributing to the Four Factors at above average rates tend to win championships. Conceptually, the teams with the greatest diversity of quality players have won the WNBA title for the past three years.

In real plain terms, teams with the most talent do well even in a league where team play might be of higher value than one-on-one play.

All we're doing with all this statistical stuff is defining what "talent" is and how to turn a collection of talent into a great team. While it might seem overwrought, all of this might assume even more importance than usual as we enter a season that might be heavily shaped by who is present come playoff time.

Click here for the 2011 version of this framework.