Developments in the Baseball Blogosphere
Toward the end of the 2004 season, I wrote about a key late-season Dodgers/Giants game, and I graphed the Win Expectancy of the game, play by play. As far as I know, this was the first time it had been done, and I was pretty proud of myself. It combined two of my biggest baseball interests, win-based stats and graphs, in a useful way. In fact, I was a little disappointed that Bud Selig didn’t come knocking on my door, thanking me for my vital contribution to the greater baseball good.
I plugged on, however, writing a reference article about Win Probability Added, making a WPA spreadsheet available on my baseball graphs site, writing a regular Game in Review column last year and including WPA graphs in the 2006 Hardball Times Annual.
Well, Bud Selig still hasn’t come knocking, but I’m happy to say that Win Probability (same thing as Win Expectancy, by the way), and particularly WPA graphs, are taking off in 2006. Here are the most recent developments I’m aware of (apologies to those I have missed):
- The outstanding Mariners blog, Lookout Landing, posted WPA graphs all last year and this year, along with Jeff’s “patented” video review of pitcher mechanics.
- Many team-specific blogs are focusing on WPA for their teams this year: Will Young is tracking the Twins in his blog, Robin to Ricky is tracking the Brewers, and the Sporting Brew is doing the same for the Yankees.
- A couple of blogs have been created just for the purpose of tracking WPA (and having middling success; tracking WPA for every game is hard work!): the WhiteSoxWPA is tracking the men of the South Side, TigersWPA is doing the same for the Detroit crew. The Crawfish Boxes sporadically posts WPA graphs of Astros’ game.
- Beyond the Boxscore is running a weekly Game in Review this year.
- Chris Shea has updated his historical win expectancy data with 2000-2004 data. Shea’s application is extremely useful as a guideline but shouldn’t be used in detailed WPA analyses.
- Skyking writes about his fantasy league and WPA analysis on his blog. WPA geeks: Check out his most recent analysis of WPA standard deviations for each situation, which is a way of measuring the importance (or leverage) of a situation.
- Most impressively, Fangraphs has automated the WPA spreadsheet and now posts a WPA graph of every game the following day. Bloggers can cut and paste those graphs into their own blogs, as long as they acknowledge fangraphs.com as the source.
- And the Washington Post even has a LIVE WPA graph for each Nationals’ game on their game-tracking page.
Now, some folks feel that player WPA is the “ultimate baseball stat.” I’m not one of them. WPA is useful for analyzing the ins and outs of games, and it’s also very useful for making in-game decisions. But it’s too quirky to use as the primary stat for MVP or Hall of Fame voting. I prefer to use it as a secondary consideration for players who are ranked closely in other stats, such as Win Shares. I’ll try to demonstrate why I feel this way in a minute.
As this blog nicely argues, WPA is all about narrative. It expresses, particularly in graphical form, what is happening during a ballgame. It’s the perfect companion to watching a game and I’m thrilled to see its use expanding. With all of these graphs abounding, I thought you might appreciate a few basic graphs to understand what you should be looking for. Here,then, are a few WPA graphical examples, showing the Win Probability of the home team on each play.
On the right is a simple graph of a couple of pitchers pitching perfect games against each other, three up and three down each inning. As you can see, WPA starts at 50% (since each team has the same chance of winning at the beginning of a game), moves up as the visitors are retired in order, then declines back to 50% as the home team batters repeat the drill.
As the game continues, the home team’s probability rises in the middle of each inning. With no outs and no one on in the bottom of the ninth, in fact, the home team has a 64% probability of winning. The simple explanation is that the longer a tie game goes, the less time the visiting team has of coming back if the home team gets ahead. In the bottom of the ninth, the visiting team will have no chance at all if the home team scores a run. This is why managers often use different strategies on the road than at home.
On the left is a graph of what happens during a game when a batter hits a home run in the very first at bat of the game, and then both pitchers again throw perfect games. The home run reduces the home team’s WP to 40%, where it hovers for awhile and then fades quickly in the last few innings. As in the first graph, the ups and downs get wider as time runs out.
You can probably tell why relievers are so highly valued by WPA. In fact, if a reliever pitched the last two innings for the visitors in this game, he would receive 35 WPA points. The starter, who would have pitched seven innings of shutout ball, would receive only a somewhat higher total of 51 points. The batter who hit the home run, by the way, would receive only six WPA points because he made outs in his other at bats.
Below, I’m putting two graphs side-by-side to really make a point. In the graph on the left, a home team batter hits a home run in the bottom of the first for the only hit/run of the game. The graph on the right is the exact same scenario, except the batter hits his dinger in the bottom of the ninth instead.
The home run in the ninth is obvious, but you can barely see the home run in the first. So the ninth-inning slugger receives four times as much WPA credit as the first-inning guy. Is this fair? Probably not. But it does accurately reflect the “narrative” of the game.
Below are two more graphs that illustrate the game narrative. In both games, each team hits a home run every other inning, ending in a 5-5 tie after nine innings. But in the left graph, the visiting team hits theirs first, while the home team hits theirs first in the right graph:
Two identical games, really. All I did was switch the order in which the teams hit their home runs, yet the graphs differ dramatically. This is the power of WPA: when teams take a lead, their fans feel good. When they have a lead in the ninth, their fans feel very good. And when the team loses that lead, their fans feel terrible.
WPA captures the dynamic of watching a game. It expresses the highs and lows of each event in a game, and also expresses the contextual importance of each situation and game event. Try downloading my spreadsheet and tracking a game next time you watch one. To quote Tangotiger, you may never watch a game the same way again.
Other Developments in the Baseball Blogosphere
Striketwo.net and Ballbug
Have you discovered Striketwo.net yet? Both Striketwo and Ballbug use technology called memeorandum to pull together the best/most viewed blog entries, resulting in a “top news” or “most discussed” lineup. Striketwo is particularly impressive due to its “most popular players and teams” pages, which show the most popular items in larger font in a “cloud” of players and teams. I visit it several times a day.
FirstInning
My buddy Aaron “Mr. SI” Gleeman set me onto another new site called First Inning, which is devoted to minor league players. It doesn’t carry news of minor leaguers (check out John Sickels’ blog, among others, for that sort of content), but it does have a PECOTA-like projection and commentary system for minor league prospects. The site is way cool, with stats, projections, commentary and news for many, many minor leaguers.
Heater
There’s a new baseball magazine in town, and I write for it. The magazine is called Heater, and a 50-page issue will be published every week during the season. Heater is the creation of John Burnson, who writes regularly for Ron Shandler’s Baseball HQ and created the Graphical Pitcher.
John’s vision is to have a magazine that provides what Baseball Weekly used to provide: useful commentary (John Hunt, Deric McKamey and I are the three regular columnists) and lots of useful, visual statistics for tracking your favorite players and teams. Please check it out. John is a very talented editor and you’ll almost certainly find it’s worth the price.
References & Resources
In every WPA article, we should always acknowledge the work of a few key individuals. The Mills brothers and George Lindsey were two early researchers into the subject. Doug Drinen and Jay Bennett have also done a lot with WPA in the past. Most recently, Tangotiger has done a lot of great research with WPA. And special kudos are due to Keith Woolner of Baseball Prospectus, for publishing the key formulas for computing WPA in different run environments. Special thanks to Jon Daly for starting the WPA spreadsheet.