skyking162

I’ve been waiting for league leaderboards over at Fangraphs.com, and now they’re up! You can find the leaders for any stat on the site — including Win Probability Added, Leverage Index, GB%, and the usual suspects — for any league in any year 2002-2006. I always like clicking my way through sortable stats, but this page is especially fun and informative. David’s pretty much said it’s in a beta stage, so look for more improvements down the road.

Popularity: 2% [?]

Below is the chart from last time, showing the change in probability of the Yankees and Red Sox winning the 2004 ALCS (from FanGraphs). The top represents a 100% chance the Yankees win (they got dangerously close) while the bottom represents a 0% chance. Each game has its own win probability graph ranging from 0% to 100%, but taken in the conext of the seven game series, each game has a different range in Series Win Probability. For example, after Game 1, the Yankees could have a 34% or 66% chance of winning the entire series, a range of 32%. But Game 4 only has a range of 12%, from 88% up to 100%. While Game 4 was exciting, it was only one-eighth as important as Game 7, which had a Series Win Probability range of 100% (win or go home).

Here are the overall WPA totals for the entire series. Each team is sorted by Series Win Probability added, where the differing leverage of each game is staken into account. The second column is the sum of all seven game’s worth of normal WPA. Notice that both the total WPA is either +/- 50 (the Red Sox won one more game than the Yankees) and the total Series WPA is either +/- 50 (the Red Sox won one more ALCS than the Yankees).

David Ortiz was definitely the big hero and Kevin Brown the goat — Tony Clark tried his best to challenge Brown, however.

If you enjoy the useless elegance of SWPA, the concept could easily be extended to encapsulate all three levels of the playoffs. The World Series should be weighted twice as much as the League Championships Series and four times as much as the Wild Card round. Couple this Uber Playoff Series Win Probability Added metric (UPSWPA) with the ridiculousness of Playoff Probability Added and you’ve got yourself quite a statistic. On one hand it would reveal the players who did the most to help their team wins the World Series and on the other hand it would be completely meaningless. What a combination!

Hat Tip…

… to David Appelman for computing the WPA data and graphs at his site, FanGraphs.com.

Popularity: 2% [?]

David’s got a cool article up over at FanGraphs that dissects all sevent games of the 2004 ALCS (Boston’s miracle comeback over the Yankees in case you’ve got memory problems.) Even sweeter is his composite graphs of all seven game WPA graphs. Check it out.

I thought I’d take things one step further and compute Series Win Probability Added. The WPA numbers in the article treat all games equally, whereas Game 7 is much higher leverage game than Game 1. The Series Win Probability (SWP) for the Yankees going into both Game 1 and Game 7 was 50%. However, by winning Game 1, they “only” increased their chances of winning the series by about 16%, wherease winning game seven would have bumped them up 50%. To convert game’s WPA to SWPA, you need to multiply the WPA numbers from the game by the difference in SWP between the Yankees winning and the Red Sox winning (listed on the chart below).

Dave Roberts’ thievery of second base in Game 4 is remembered as the series’ defining moment. But the numbers have a different memory: Roberts’ steal was worth less than both Millar’s walk and Mueller’s subsequent RBI in that same inning. It was also worth less than Manny’s leadoff single and Ortiz’s walk-off blast in the 12th. On top of all that (and not to be a downer), Game 4 was the least important game of the series. Really. Going into the game, the Yankees had a 94% chance of winning the series. Even with the Red Sox pulling out the miraculous victory, the Yankees were still in the driver’s seat up three games to one. Winning the last three games was only a 1 in 8 shot for the Sox.

It’s ironic that the most important game — in terms of series leverage — was also the least exciting. Game 7 was basically over after Damon’s second inning grand slam put Boston up 6-0. You won’t see many WPA graphs more boring than that one. But considering an entire seven game series boiled down to just that one game, David Ortiz’s first inning two-run homer was the one play that most changed the probability of the Red Sox winning the ALCS.

(I plan on posting the total SWPA for all players, but it’s going to take some time since I can’t just copy and paste the data from the FanGraphs post. I have a feeling David Ortiz comes out looking pretty good, though.)

Popularity: 3% [?]

FanGraphs got a mention in the Pittsburgh Post-Gazette as their Stats Geek column dissected the Pirates from a Win Probability Added point of view. The author, Brian O’Neill, goes a good job of working with WPA and explaining it in layman’s terms at the same time. He tackles the Big Hit Theory, the idea that coming up with big hits is a unique, repeatable skill, rather than a reflection of other actual skills (like being a good hitter).

Evidently the readers in Pittsburgh are smarter than the average bear in order for their newspaper to print an article like that. Our local paper seems to be stuck in the middle ages of baseball fandom.

Popularity: 2% [?]

Win Probability Added has been gaining steem for a few years now, but it’s popularity really seems to be boiling over this season. I’m a big fan of the statistic for a variety of reasons, none of which are because it carries any significant contribution to the discussion of who’s a better player:

• WPA’s methodology is still being worked out. It’s a fun puzzle to figure out how to calculate WPA and leverage. (Hey, I’m a math geek.)
• WPA is a foundation for showing that modern bullpen usage is suboptimal. Teams could probably squeak out another few wins each year just by deploying their relievers better.
• WPA lends itself quite well to visual presentation. Graphs like these are a nice change of pace from the traditional game writeup.
• WPA reflects the fan’s sense of drama while watching a baseball game. A homerun hit in the first inning doesn’t evoke the same emotional response as a walk-off homer.
• WPA accounts for how much each player actually changed his team’s chances of winning a game. There’s no guessing or estimating as to how much a single generally helps a team. A specific single in a specific situation has an exact value to a team.
• WPA is the ultimate MVP metric.

I know Dave Studeman disagrees with that last statement, so let me explain. In my mind we should not be awarding an MVP — it’s just too vague an award, one that I’ve stopped really caring about. What we should be awarding is a Best Player award to honor the player that achieved the highest level of performance — the player that would be picked first before the season if every general manager was omniscient. Best Player would not include team-dependent influences such as RBI, game-winning HRs, or the “skill” of carrying a team to the playoffs.

But since the baseball writers don’t award the BP and instead hand out MVPs, we can at least limit the damage using WPA. Most voters see the MVP as the player who did the most to help his team win. That includes being clutch and performing when the possibility of making the playoffs hangs in the balance. Isn’t that exactly WPA?

Actually, not quite — WPA is the next-to-ultimate MVP metric. The ultimate would be Playoff Probability Added — a combination of WPA and playoff odds. Just like WP is the probability of winning any particular game given the current game situation, PP is the probability of reaching the playoffs given the current game situation and league standings. Helping your team win a game early in the season isn’t as dramatic as hitting a walk-off homer on the last day to clinch the Wild Card. And hitting a walk-off homerun on the last day for a team in last place wouldn’t exactly be MVP-caliber material.

Conveniently, Baseball Prospectus posts a Playoff Odds report that presents the likelihood of each team making the playoffs. For each game, you’d simply need to multiply each player’s WPA by the difference in playoff probabilities between his team winning and losing that game. Sum over the whole season and you’d find out which players did the most towards increasing (or decreasing) their team’s chances of making the playoffs. Hello, MVP.

I find it ironic that the ultimate MVP stat is perhaps the most convulated, math-geekified calculation I can think of. I mean, if MVP voters would pay any attention to it, they’d probably first accept that Best Player is a much better award than MVP. It’s just a fun little toy. Like WPA.

Popularity: 2% [?]