Sunday, October 26, 2008

The Difficulty in Playing in the WNBA Over Time

How much does skill level change in any sport over time? In baseball, you can enjoy a whole evening's worth of hot stove league arguments on this very subject. Pluck Babe Ruth magically out of the 1921 lineup of the New York Yankees and drop him into the 2008 lineup. Would he still be the Sultan of Swat? Would he be an All-Star player? Or would he be washed up trying to hit against 2008 pitching?

We know that in many sports, nutrition advances, improvement in training methods, population growth, and a increase in a sport's popularity (leading more potential athletes to choose the sport) change the skill level of sports leagues year by year.

The shoe can be placed on the other foot. A league's competition level can be weakened by overexpansion or by a decrease in a sport's popularity, giving players a chance to compete that wouldn't have made it in years past. Leagues can be watered-down as easily as strengthened.

Given that, is there a way to estimate how much tougher a league gets year by year?

There's a very simple way to do it:

a) Take a group of players from 1997 and see how well they do against the league,
b) Follow that exact same group of players from 1998 and see how well they do against the league.
c) If the group did better in 1998 than they did in 1997, the league got weaker - the reason that they put up better numbers was that the competition was weaker.
d) If the group did worse in 1998 than they did in 1997, the league got stronger - they put up worse numbers because the competition was tougher.
e) Do this for 1997 vs. 1998, 1998 vs. 1998, etc. until you get to 2008.

This "cohort" way of doing things makes sense. The only problem is how big should the cohort be? The ideal answer is that the cohort should be the entire group of players in 1997 who were also active in 1998.

Unfortunately, I can't do that. I don't have a database of year by year performance for players that automatically does these calculations. So I'll use Basketball Reference and choose the ten WNBA players with the most minutes in one year, and compare the performance of this group over to the next year. When I use the next year as the starting point, I'll perform the same process with a new group of players - the players who had the most minutes in the new starting year.

Occasionally, you get situations where a player was among the top ten in minutes one year and disappears the next year - the Kym Perrot case. In which case, I substituted players from the most field goals category. The goal was to choose players who were very active.

Another problem (which I was forced to ignore) was differences in coaching style. Players from the 2007 Indiana Fever are going to have different levels of production than the 2007 Phoenix Mercury. I don't have a solution for this problem, but I'm not looking for exact values, just an estimate of how much tougher a league gets over time.

The next problem: what metric should I use to measure a player's overall performance?

I chose the metric of Wins Score. I don't like the Efficiency Metric used by the WNBA because I feel it overvalues shooting percentage, awarding points for poor field goal shooting. Furthermore,

a) Wins Score is an "additive metric", giving points for everything a player does right, and
b) It takes into account all facets of a player's game, both offensive and defensive.

I looked at combined Wins Score, and then divided that into a "Wins Score Per Minute" for my particular group of players to take into account different amounts of total minutes players. This "Wins Score Per Minute" rating was tracked over two successive years for each of the cohorts to determine relative strength and weakness of a league.

The Analysis

So here we go. We set the difficulty value for the 1997 WNBA at 1.00.

19971.0008 teams, 28 games, halves, 30 second clock
19980.910expansion from 8 to 10 teams, 30 games
19991.385expansion from 10 to 12 teams, 32 games, ABL players join
20001.113expansion from 12 to 16 teams
20031.601contraction to 14 teams, 34 games
20041.730contraction to 13 teams
20061.429expansion from 13 to 14 teams, movement from halves to quarters, 24 second clock
20071.404contraction to 13 teams
20081.472expansion to 14 teams

A graph is attached below:


Let's look at the graph and the numbers in the table above.

The dip in league quality from 1998 shouldn't be a surprise. With extra teams, we'd expect the team to get weaker, and this was the year when the Houston Comets won 27 out of 30 games. According to the estimate, 1998 was the weakest year in the WNBA's history.

Going to 1999, a flood of American Basketball League stars entered the league and the role players who filled spots on WNBA rosters were suddenly displaced. The league took a 52 percent jump in level of competition, the biggest jump between any two years out of the 13 years the WNBA has existed. Having the former ABL players did a lot for the WNBA.

The league's quality suffered in 2000 with an expansion to 16 teams, the largest number of teams ever in the WNBA. However, you'll notice that the league's quality slowly edged up year after year after year. Most likely, these changes reflect changes in coaching both at the WNBA and college levels, as well as the growing popularity of women's basketball as a sport. The level of play got tighter and tighter year after year, reaching a high in the 2005 season. Contraction down to 13 teams also helped moved the process along.

However, in 2006, the league's quality mysteriously dipped. There were a set of rule changes named for the 2006 season, and the league's strength overall dipped 23 percent. These changes, according to the Chasing the Title blog, had great impact on the game. They allowed players with superior athleticism to dominate other players. Players had a couple of minutes of rest. The shorter clock gave quicker athletes a premium and hurt teams trying to run the old 30 second full court offensive.

My suspicion is that the changes hurt a lot of players that weren't able to adjust. Like changes in the enviroment in an evolutionary example, some of the players in the W found themselves handicapped. Two minutes of rest where formerly no rest existed might help some players over other players.

The last two numbers are paradoxical. You'd expect the league to become tougher in 2007 (contraction and fewer places for players) and weaker in 2008 (expansion). It could be that the 2008 change is reflected by the strength of the draft class in 2008; I don't know what accounts for the 2007 dip. It might be random fluctuation.

Using a model like this provokes further questions:

1. Some American players thought that the WNBA would be able to offer enough money so that they wouldn't have to play overseas anymore. I'm sure that some players enjoyed a couple of years at home in the offseason. When it became obvious that salaries were going to remain small, players had to begin playing year-round again, playing in Europe in the off-season. How did this change competitiveness?

2. After a certain point, the pipeline of European players dried out for similar reasons. Why go overseas when you could just make the same crappy amount of money at home and work in the offseason, possibly for a company that was sponsoring your European team? How did that change competitiveness?

3. Has college basketball gotten tougher or easier over the past 12 or 13 years? How has that changed things?

4. Most of the early WNBA coaches were women, who moved from the college level to the pros. Now, the best female coaches stay at the college level, because it pays more and offers more security. Furthemore, former NBA players are preparing their resumes by coaching at the WNBA level. Do the NBA players bring a better skill set to WNBA coaching? How does that effect league strength?

No comments: