Last night was close to perfect, then came crumbling apart.
I watched game 7 of the NBA finals, rooting for the Spurs. The game was tense and close throughout, though Tim Duncan, the Big Fundamental, couldn't get a tying layup or tip-in to go, so the King James version of the ending (a 17-foot jumper) won out.
The basketball game disappointed me, but Kyle Seager kept cutting through my twitter feed. It seemed the Mariners were thrashing the Angels early.
I changed the channel just in time to watch Peter Bourjos take Felix Hernandez deep. It turned out to be the first of seven consecutive hits for the Angels.
I thought I was going to watch the Spurs win a hotly contested game seven, then enjoy a rare Mariners rout - of the Angels no less, the team I hate more than any other! Instead, I got a Heat victory, and the worst inning of King Felix's career.
So last night went from nearly perfect to utter disaster.
I knew I had to write something about last night's Mariners game, besides how soul-crushing it was. Part of me thought that I had a skewed view of the game, given that I only saw the Mariners score 1 run, and the Angels score 8. It was one of those weird games with a big swing in the middle, and I only saw the latter half of it. So, it felt like a rout for the Angels as I watched it, even though most of the innings I caught featured a close score.
That got me thinking about the nature of painful losses, and thrilling victories. They are the ones that live on the edge between glory and disaster, the games that could easily go either way. Sabermetrics has a nice way of gauging these situations, through a stat called Win Percentage Added (WPA). It is quite simply a way of calculating the odds of a team winning a game, given the current game situation. Each event changes the odds, so WPA is a calculation of how much the odds changed for or against a team winning.
I had a simple idea this morning: why not take the absolute value of each WPA in a game's play log, then add them all together? In theory, games with big momentum swings would have higher sums, because one team would go from heavily favored to the other. That swing would take lots of changes in WPA, though if I didn't take the absolute value, I wouldn't see the big fluctuation. Most of the WPAs would cancel each other as the pendulum swung back and forth.
I've decided to call the sum "net WPA." Below is a chart summarizing the net WPAs for every Mariners game in June. The 'W's and 'L's show Mariners wins and losses:
Last night's net WPA was 3.99, which means it is the game fourth from the left. The crazy high net WPA came in the 16 inning game, where Seager hit a grand slam to tie the game in extras, only to have the Mariners ultimately lose. It makes sense that that game was so brutal by net WPA standards.
The M's losses this month, according to WPA, are more painful than their victories are satisfying. An average loss this month has a net WPA of 3.16, and an average victory has a net WPA of 2.35. I decided to see if the median net WPAs told a different story, given that the crazy 16-inning game might have skewed the results. Medians told the same story though. The median net WPAs for wins and losses were 2.58 and 1.94, respectively*. The extra inning game had skewed the results some, but not enough to change the story.
*The fact that both medians were lower than the means suggests that WPAs are skewed in general. A logarithmic scale might be more appropriate, though this is also a very small sample size.
Higher WPAs tend to correlate to more memorable moments in a game, because they are attached to moments that greatly alter a game's outcome. So, it follows that the higher net WPAs in Mariners losses makes their losses more memorable. Tack on the fact that the Mariners lose more often than they win, and it's easy to see how losses are the dominant story in June.
Boring wins coupled with painful losses. Is it much of a wonder that fans aren't flocking to Safeco Field these days?