There's something very appealing about a streak. The analytical part of my brain knows that performance over long periods of time is what matters, not the way those performances are clustered. The lizardy parts of my brain see consecutive performances as a pattern, suggesting more strongly the possibility of future greatness. If he hit a home run the day before yesterday, and he hit a home run yesterday, and he hit a home run today, I will inevitably be disappointed and in some way surprised when he doesn't hit a home run tomorrow.
Streaks are fun, and fan reaction to streaks are fun. This has been a good year for stretches of dominance; Zack Greinke broke the 40-inning threshold of scoreless innings, written about on this very website by my colleague Chris Teeter, and this past weekend saw the 20th anniversary of Cal Ripken's record-setting 2,131st consecutive game. Because streaks are fun, I wanted to look at the best streaks of this year, for pitchers and hitters, and try to wring some conclusions out of them.
What I did was find the best streaks of every length from 1 appearance (a streak in only the most pedantic sense) to 10 appearances by four different measures: pitcher Win Probability Added, pitcher Game Score, batter Win Probability Added, and batter Total Bases. As a reminder, Win Probability Added measures the impact of a player on his team's likelihood of winning a game, so it is very much context dependent; a home run in extra innings of a 0-0 game means much more than a home run in the 9th down 10-2 (or up 10-2, for that matter). It also gives no credit for fielding and correspondingly all credit for pitching -- a diving catch to turn a triple into an out doesn't get the fielder anything, but it does get the pitcher the same credit as a routine fly ball.
Game Score, on the other hand, is context neutral. It's a Bill James-developed metric which gives a pitcher credit for getting outs, extra credit for strikeouts, and deducts credit for hits, runs, and walks. It is entirely for fun but does a good job of identifying the pitching performances that seem the most impressive. Batter Total Bases is pretty obvious; it's not a particularly sensitive or informative metric, but it's a decent context neutral metric and most importantly is trackable by the Baseball Reference Play Index Streak Finder. On to the streaks!
|Streak Length||Name||Start Date||End Date||MinWPA||Notes|
|5||Zack Greinke||6/18/15||7/18/15||0.344||6 games|
|7||Yovani Gallardo||5/24/15||7/2/15||0.135||8 games|
|9||Dellin Betances||4/17/15||5/8/15||0.071||11 games|
|10||Dellin Betances||4/17/15||5/8/15||0.071||11 games|
This probably requires some explanation. For each streak, I looked at the minimum WPA for each player over each streak of that length and chose the best streak. For example, over his six-start streak from June 18th to July 18th, Zack Greinke's lowest WPA was .344, and that was the highest minimum WPA for any pitcher over any six-game streak (or five-game streak, for that matter). For some context on WPA, the Chris Archer "streak" that tops the one-game list was a one-hit, one-walk, eleven-K complete game shutout. The Rays scored a single run in the fourth and no more for the rest of the game, meaning Archer was constantly pitching in a tight situation and justly rewarded by WPA.
The shape of the chart gives some information about the distribution of WPA. A single game over .800 is extraordinary; multiple would be practically unprecedented, and the level of the streaks drops off rapidly as they increase in length. Zack Greinke's dominant stretch is in there at five and six, helped by some lackluster offensive support in those games, and it isn't until 9 and 10 that a reliever first appears.
That stretch of Betances appearances in April and May doesn't seem unbelievable on its surface, and in a real way it isn't. He pitched well and didn't give up a run in any of the appearances, but they weren't more or less extraordinary than normal-extraordinary Betances. The consistently high WPA figures have more to do with the situations and games he entered. On April 26th, Chris Martin sustained an injury with two outs in the second, and Betances worked a scoreless inning, the WPA inflating because of the timing of his appearance. On April 21st, he entered the game with no outs in the eighth, with the Yankees up by three versus a runner on second, and left the game after three quick outs and a tidy WPA to go with it. What these illustrate is why WPA can be less than satisfying; context-dependent stats rely on a lot more than the performance of the player in question.
Pitcher Game Score
|Streak Length||Name||Start Date||End Date||MinGSc||Notes|
|3||Clayton Kershaw||7/8/15||8/1/15||84||4 games|
|5||Zack Greinke||6/18/15||7/19/15||69||6 games|
|7||Chris Sale||5/12/15||6/19/15||67||8 games|
|9||Zack Greinke||6/7/15||7/31/15||61||10 games|
Remember first half Max Scherzer? He was really good! He threw a one-hit complete game shutout, with 16 strikeouts and one walk, and then six days later threw 96% of a perfect game and still managed a no-hitter. His game scores of 100 and 97 are currently first- and sixth-highest on the season, and he became the only pitcher ever to record game scores of at least 97 in consecutive starts. Let's just not think about second half Max Scherzer, because one should try to be happy, as a general rule.
The drop-off here is much more gradual than in the WPA chart, in part because of the distribution of Game Scores but also a reflection of the greater control an individual has over Game Score as compared to WPA. Greinke's crazy mid-summer stretch makes several appearances in the leaderboard, and the worst appearance in his 10-game stretch was a middling performance against the Mets with three Ks, three BBs, and 2 runs over 7 innings, good for a "mediocre" 61 Game Score. Being consistent enough over basically 2 months that your worst performance is "mediocre" is seriously impressive, but due to the lack of Dodgers offense that night, Greinke finished with a WPA of -.002. That's also why this a) reads like a leaderboard of possible answers to "Who has been the best pitcher in baseball thus far?" and b) features all repeated names, since Game Score is much closer to a measure of repeatable skill than WPA.
|Streak Length||Name||Start Date||End Date||MinWPA||Notes|
|T-9||Josh Reddick||4/24/15||5/4/15||0.026||10 games|
Batter WPA tends to be lower than pitcher WPA for a few reasons. With only four or five times to contribute, a batter's impact is somewhat limited. To put up a really gaudy figure, it takes either a) a lot of clutch hits over the course of a close game or b) a real, honest-to-goodness lifesaver of a hit. On July 27th, the Cubs played the Rockies, and Kris Bryant's first four plate appearances looked like the following: 1-4, 2 Ks, 1 1B. The single was part of a fourth inning rally and was situationally important but nothing special. Over the course of this game, the Cubs had gone from down 4-0 to up 7-4 before a four-run Rockies eighth that put them back in the lead. Kris Bryant stepped up with two outs in the ninth and a runner on second, down by one run. Past teams in that situation have gone on to win 8.7 percent of the time, which honestly seems high. I guess the path to victory is fairly obvious, or at least it was to Bryant. Gratuitous embedding ahead:
The combination of that super-clutch hit and his previous not-very-important outs meant that Bryant's .975 figure was borderline historic. Only 48 other players have had higher WPAs in 9-inning games; the highest of all time comes from Jim Pagliaroni, your friend and mine. On September 21st, 1965, he also hit a walk-off, two-run dinger with two outs after contributing to three prior rallies, accumulating a whopping 1.287 WPA in five plate appearances.
Anyway, there's not that much else to say here. WPA is fun but weird, and it's especially hard for batters to put up consistent figures. At least now you know something about Tommy Pham!
|Streak Length||Name||Start Date||End Date||MinTB||Notes|
|T-4||Nolan Arenado||9/1/15||9/4/15||5||5 games|
|T-4||Nelson Cruz||7/31/15||8/4/15||5||5 games|
|T-7||Nelson Cruz||7/31/15||8/9/15||3||9 games|
|8||Nelson Cruz||7/31/15||8/9/15||3||9 games|
|T-10||Miguel Cabrera||8/15/15||8/26/15||2||11 games|
|T-10||Jose Altuve||4/21/15||5/2/15||2||11 games|
Finally, total bases. Again, not the best measurement, but it is available. I've tried to set all these charts to have a similar maximum on the y-axis, so the shape of the decline can be compared across. Total bases, like Game Score, is more controllable by a batter, and so the decline is less sharp and the names more recognizable. This leaderboard is also a bunch of good hitters and someone named Curt Casali, who apparently is also a good hitter but both injured and a Ray and therefore not on my radar.
Again, streaks are mostly fun, and I'm mostly highlighting these because they're crazy both while they're happening and in hindsight. Remember when Adrian Gonzalez started the season with three straight games with at least 7 total bases? Remember how Nelson Cruz is having a 5-win, 170 wRC+ season? Remember that Miguel Cabrera and Jose Altuve are not only buddies on the above leaderboard, but in real life?
Baseball is goofy, and streaks are goofy. While they can illustrate some points about context-neutrality or the frustrating whimsy of run support, mostly they're good for fun. That's good! Don't read too much into them, obviously, but I am going on the record in favor of fun, and therefore streaks.
. . .
Henry Druschel is a Contributor at Beyond the Box Score. You can follow him on Twitter at @henrydruschel.