Saturday, July 5, 2008

The Mann/Bales Trade and More Crazy Stats

I was very much influenced by a post at Rethinking Basketball that was used to evaluate rookie talent in the WNBA. The point of the post was to determine which rookies had the potential for success given their limited playing time and which were likely to fail. This inspired me to do the same regarding Mann and Bales. Neither gets starters minutes so the tools used by Q to test rookies could be useful.

The first tool I want to develop is something called "Wins Score". "Wins Score" is what we call a linear metric. All that means is that Wins Scores is one of those tools that gives a player x points for something something good, y points for doing something bad, and adds all the points together for a final total. (Remember 3x - 2y in algebra? Just like that.)

If you just can't stand not knowing the metric, here it is:

(Points + Rebounds + Steals + ½Assists + ½Blocked Shots – Field Goal Attempts – Turnovers - ½Free Throw Attempts - ½Personal Fouls)

The WNBA uses a similar metric called "Efficiency"

((Points + Rebounds + Steals + Assists + Blocked Shots) - ((Field Goal Attempts - Field Goals Made) + (Free Throw Attempts - Free Throws Made) + Turnovers))/Games

There are two differences in the two formulas. The second divides by "games" at the end. All this does is reduce the second formula to a "per game" basis.

The other difference is that Wins Score only gives half credit for assists and blocked shots. It ignores field goals and free throws made, only gives half credit for free throw attempts, and unlike Efficiency, Wins Score takes points away for personal fouls whereas Efficiency ignores personal fouls.

Most Basketball APBRmetricians like "Wins Score" a lot better than they do "Efficiency". When I switched to "Wins Score" the following results popped up.

Betty "Noox" Lennox: 26th in Efficiency, 55th in Wins Score
Ivory Latta: 40th in Efficiency, 58th in Wins Score
Tamera Young: 51st in Efficiency, 53rd in Wins Score

Furthermore, Stacey Lovelace pops up to 56th in Wins Score, above Ivory Latta. Why does everyone take such a hit in Wins Score?

First, Wins Score doesn't care how well you shoot, as opposed to Efficiency. Wins Score just cares how many points you put on the board. Secondly, turnovers hurt you in Wins Score, and they don't hurt you in Efficiency.

Betty makes a lot of turnovers for the points she makes. As the end of 2007, she was 12th in turnovers all time in the WNBA. Wins Score appropriately devalues her overall worth to the team. With Tamera Young the highest overall Wins Score leader for the Dream -- at 53rd in the wNBA -- it's no wonder that the Dream is 0-17.

All right. Let's assume that we agree that Wins Score does a better job than Efficiency in predicting a player's worth. Where do we go from there? What does this mean both for Kristen Mann and for Alison Bales?

The first thing we need to do is create some way to answer the question, "if these two players played starters minutes instead of sub minutes, what is their potential for upside?" As it turned out, this question had already been answered by a man named Kevin Broom.

He created something called a "Diamond Rating". The point was to find "diamonds in the rough" which had been overlooked, players who might be able to make something happen if they had playing time.

The formula is a bit more complex, so I'll just summarize how it works. I changed Broom's formula and assumed that a WNBA starter would play 33 1/3 minutes per game. (His formula is an NBA formula, and he uses 40 as the starter's benchmark.)

a) Figure out what the Wins Score would be if the player played 33 1/3minutes and divide that by games played,
b) Subtract their current Wins Score per games played,
c) Add the difference between their win score per 33 1/3 minutes divided by games played and the league's average win score per 33 1/3 minutes divided by games played.

A and B determines how close the player comes to a starter. C gives them credit if they're producing more than an average player.

We then exclude certain groups of players. Anyone who's playing less than 2.5 minutes per game is chopped off -- we don't have enough data on them. Anyone playing more than 21 minutes per game is chopped off -- they're earning close to starter's minutes already.

WNBA leaders in PADR (Petrel Adjusted Diamond Rating)

CON Danielle Page 39.83
HOU Sancho Lyttle 19.28
MIN Vanessa Hayden 13.25
NYL Tiffany Jackson 10.95
WAS Crystal Langhorne 10.52
PHX LaToya Pringle 9.89
SAC Crystal Kelly 8.96
LAS Sidney Spencer 8.54
CON Tamika Raymond 6.98
MIN Kristen Rasmussen 6.66

As it turned out, Sancho Lyttle is lighting the league up now that she's had a chance to play. She lit up the Dream for 18 rebounds on Thursday. (Danielle Page is an outlier, as she barely makes the 2 1/2 minutes per game cutoff. I'd ignore her.)

Let's look at the Diamond Ratings of Alison Bales and Kristen Mann.

Bales 6.31 12th
Mann -0.61 67th

According to the Diamond Rating, Bales might have something extra to offer if you give her time. (We don't know yet where the cutoff in Diamond Rating for the WNBA rests. In the NBA, "diamonds in the rough" usually score eight or better in Diamond Rating.) Whereas Mann, according to Diamond Rating, doesn't have much to offer if you give her extra time.

Is Bales a real "Diamond in the Rough?" Is this going to be one of those trades that's going to work out very well for the Dream in the end and help us turn things around? The only way we'll know is that if Alison Bales is given significant minutes on the floor. We probably won't see that happen in todays Sky/Dream game, but we might see it later on in the year.

And just remember...sometimes the statistics lie...!

1 comment:

Q McCall said...

Nice post. I especially like the idea of adjusting the Diamond Rating for the WNBA. Didn't think about that.

Crystal Kelly is also an interesting one to me. I've only watched the Monarchs a few times and she didn't really stand out to me.

The only problem with using win score in this capacity is that it overvalues rebounding and doesn't separate value for Off and Def rebounds.

Here is an interesting post from the APBRmetrics board about "alternative win score" that I've been mulling over...