May 27, 2015

Wait, the Twins are winning? How did that happen?

Glen Perkins and Paul Molitor Twins

Coming off four straight 90-loss seasons the Twins were projected to finish in last place by nearly everyone, myself included, but instead they have the third-best record in the American League at 27-18. Since a brutal opening week of the season in which they went 1-6 and were outscored by 33 runs the Twins are 26-12 with a run differential of plus-52 and they've won 22 of their last 31 games. They're winning and they're winning a lot. But how? Here are three big reasons:

Scoring In Bunches

There's nothing impressive about the Twins' overall hitting numbers. They've batted .257 with 35 homers and 112 walks in 45 games, ranking 20th among MLB teams in both on-base percentage (.311) and slugging percentage (.388). They also haven't done much running under new manager Paul Molitor, stealing just 18 bases while being thrown out 11 times. And yet they've scored the eighth-most runs in baseball, including an AL-best 5.1 runs per game since their 1-6 start.

As a team the Twins have hit .257 with a .699 OPS overall, but with runners in scoring position they've hit .294 with an .806 OPS. When the bases are empty they've hit .240 with a .654 OPS, but with runners on base they've hit .282 with a .760 OPS. Whether you want to chalk up those huge differences to clutch performances, pure luck, or something in between it's easy to see why the Twins' lineup has scored a lot more runs than the overall numbers would suggest.

They've also done an exceptional job of clustering their hits together, exploding for big, multi-run innings to knock out the opposing starting pitcher or put a game out of reach. That's partly tied to the aforementioned significant uptick in production with runners on base and especially runners in scoring position, but it goes beyond that to an offense that has focused an unusually high portion of its damage within one inning per game.

Or, put another way: If a team averages nine hits per game they'll score a whole lot more runs if five or six of them are clustered together in the same inning than they would with a more even distribution of 1-2 per inning. Again, whether you want to chalk up the clustering of hits to clutch performances, pure luck, or something in between anyone who's watched the Twins this season can tell you their ability to explode for a big inning has been remarkable to see.

This year the Twins have scored four or more runs in an inning 15 times in 45 games, which is a pace of 54 times per 162 games. On average from 2011-2014 the Twins scored four or more runs in an inning 29 times per 162 games. So they've upped their OPS by 100 points with runners in scoring position, they've maximized a modest amount of overall damage by clustering it together, and they've exploded for a huge inning to put a game out of reach 2-3 times per week.

And that's how a lineup that ranks 20th in OPS can rank eighth in runs scored.

Late-Inning Relief

This season, like last season, the Twins' bullpen ranks dead last among all MLB teams in both strikeout rate and xFIP. Their rank in ERA is essentially unchanged from 23rd to 21st. However, within that all-too-familiar sub par performance from Twins relievers is some very good work in the late innings of close games. They've been bad overall, but in high-leverage situations where giving up a run could change the outcome of a game they've actually been quite good.

Glen Perkins is responsible for a lot of that. His excellence in the closer role is nothing new--he's already one of the three or four best relievers in Twins history--but he's been nearly flawless this season by converting 17 of 17 save chances with a 1.25 ERA and 21/2 K/BB ratio in 22 innings. Perkins ranks third among all relievers in Win Probability Added, which accounts for the situations in which performances occur within games and how that impacts the team's odds of winning.

His primary setup men have also thrived in high-leverage spots, which is shocking considering his primary setup men are journeyman minor-league signings Blaine Boyer and Aaron Thompson. Boyer is 32 years old and came into this season with a 4.63 ERA in the majors and a 5.31 ERA at Triple-A. He was terrible during the opening week, giving up runs in each of his first four games. And since then he's allowed a grand total of one run in 20 innings.

Thompson didn't even make the Opening Day roster, but quickly leapfrogged Brian Duensing and Caleb Thielbar in the hierarchy of lefty setup men. At age 28 he came into this season with 15 innings in the majors and 980 innings in the minors. And now he leads the American League with 23 appearances, 14 of which have come in "close and late" situations, and he's been nearly unhittable in those spots while holding lefties to a .094 batting average overall.

Thompson and Boyer both crack the top 10 in Win Probability Added among all American League non-closers and no trio of relievers in the league has a higher cumulative WPA total than Perkins, Thompson, and Boyer. They've combined for a WPA of 3.23 and the rest of the Twins' bullpen has a negative WPA, because in "close and late" situations the bullpen has held opponents to a .211 batting average and .536 OPS compared to a .314 batting average and .907 OPS in other spots.

And that's how a bullpen that ranks 21st in ERA can rank third in Win Probability Added.

Non-Disastrous Starting Pitching

Make no mistake, the rotation hasn't been good and Twins starters again rank dead last among all MLB teams in strikeouts. However, even being "not good" is actually a step up from the disastrous 2011-2014 rotation that ranked dead last in strikeout rate, ERA, xFIP, Wins Above Replacement, and opponents' batting average. For four years the Twins' rotation was a dumpster fire on which each new starter would pour more gasoline and now it's merely a standard bag of trash.

Depth has played a big part. Instead of constantly dipping down into the minors for a parade of replacement-level (or worse) starters the Twins actually have more decent rotation options than actual rotation spots, which is why Tommy Milone is currently dominating Triple-A hitters and the team is a month away from facing a tough decision when the biggest free agent signing in franchise history, Ervin Santana, returns from an 80-game suspension.

Twins starters rank 25th in xFIP thanks largely to the lack of strikeouts, but because they've been better at wriggling out of jams and limiting damage within troublesome innings the rotation sits right in the middle of the MLB pack in ERA at 15th. Whether you choose to put your faith in ERA or xFIP, going from dead last by a wide margin to somewhere within the realm of respectability has a huge impact on a number of fronts.

Compared to 2011-2014 the rotation is remaining in the game nearly 10 percent longer per start and surrendering 20 percent fewer runs per inning. Not as many games are already out of reach within the first few innings due to the starter blowing up and exiting early, which in turn leads to a less-taxing workload for the bullpen's middle relief underbelly and more of an opportunity for the Twins' lineup to explode for big, game-breaking innings of its own.

And that's how a rotation that ranks 25th in xFIP can still be a massive improvement.

To hear two podcast hosts try to figure out how to feel optimistic about their favorite team again, check out this week's "Gleeman and The Geek" episode.


  1. You forgot to include Torii Hunter’s clubhouse leadership! How could you forget that?

    Comment by sandbun — May 26, 2015 @ 11:43 pm

  2. I love you. Also, is that the Earth coming at us very quickly?

    Comment by Travis Hayes — May 27, 2015 @ 12:39 am

  3. Don’t forget his smile.

    Comment by Eric — May 27, 2015 @ 8:27 am

  4. I think that is huge.

    Comment by siouxya — May 27, 2015 @ 10:47 pm

  5. What about # of double plays? Have the twins turned an unusually high number, average etc. as compared to other teams & years past? I wonder if missing bats becomes a bit less of a priority for success now that less PED’s are used overall. In theory is it possible that hitter contact has become less dangerous overall in MLB but would still lead to as many double plays with runners on base. ( Possibly even a few more since there isn’t as much artificially enhanced speed)?

    Comment by DeWall — May 27, 2015 @ 4:21 am

  6. I’ve followed this blog for years and never commented, but I had to create an account just to say that “For four years the Twins’ rotation was a dumpster fire on which each new
    starter would pour more gasoline and now it’s merely a standard bag of
    trash” is possibly the best line I’ve read in sports writing in years.

    Comment by Woz — May 27, 2015 @ 7:57 am

  7. It seems to me Molitor is like a cold hearted poker player who only makes bets he can win. Don’t play the wrong side of the platoon. Don’t let a gassed starter or overmatched reliever finish an inning. Don’t start 2 inept statues in the outfield. Take extra bases but not risks. These are all just observations but I would guess the numbers back up that these decisons are being made differemtly. Molitor is making decisions that work well with this roster.

    Another thing I see that I don’t know if it’s been coached is the length of starts. Which makes sense because if half of the bullpen is shaky, then three runs in 7 innings is better than 0 runs in five innings, because it maximizes your wins in the long run.

    Comment by justme — May 27, 2015 @ 9:33 am

  8. Good points — it seems like Molitor has got a great handle on this team.

    Comment by funoka — May 27, 2015 @ 10:01 am

  9. This team is showing that sabermetrics shouldn’t be the only tool when it comes to assessing a team/player. It’s a valuable tool. It’s not THE tool.
    The turn around this season can also be attributed to improvements made in the field. Their outfield is actually respectable with Arcia no longer playing one of the corner spots, and Hicks making plays in center. Danny Santana’s defense continues to improve.
    One outstanding defensive play in a close game can and has changed the outcome.
    How much credit does Neil Allen deserve? Paul Molitor? Twins have been fun to watch.

    Comment by Kale — May 27, 2015 @ 9:35 am

  10. I don’t think anyone advocates that sabermetrics should be the only tool. Regardless, what does this team’s performance have to do with downgrading the usefulness of sabermetrics? I don’t think anyone predicted the twins would be sporting this record this late into the season whether you look at old school stats, new school stats or just use your eyeball and gut.

    I think the amount of credit due to managers and coaches is incredibly difficult to assess, which is why the variance in pay across coaches and managers is relatively small. If any coach were a lot better than average and we were able to measure it, they’d have a much larger paycheck. Empirically, teams are willing to pay, what, $7 million per win above replacement for players these days, so if a team thought a manager was going to get them a couple wins above the replacement manager, then top managers would be getting paid closer to $15 million a year than the $1 million coaches actually get.

    The one place we do agree is that it is certainly a lot of fun to watch!

    Comment by Kavan — May 27, 2015 @ 10:07 am

  11. “what does this team’s performance have to do with downgrading the usefulness of sabermetrics?”
    –Did you not read Aaron’s post?

    Comment by kale — May 27, 2015 @ 10:35 am

  12. I don’t understand, are you saying that what Aaron wrote is sabermetric-y and it doesn’t do a good job of explaining what we’ve seen? I guess I would argue the exact opposite, I think this article does a nice job of explaining how the twins have gotten where they are without the use of super complicated advance metrics.

    The article is filled with things about batting average with runners on base, clumping hits together, length of starts by starting pitchers (and runs allowed in those starts) and ERA of the bullpen. Those aren’t sabermetric concepts, the arguments are just based on solid observation and some critical thinking, I guess I don’t consider that sabermetrics. Sure, he sprinkles in some observations based on xFIP and WAR but I don’t this, or hardly any of Gleeman’s posts, hang their hat on sabermetrics.

    Comment by Kavan — May 27, 2015 @ 10:44 am

  13. So essentially, advanced stats will tell you that advanced stats are deceiving?

    Average the OPS+ by position of players who have played the most games. The Twins as a team are well below average.
    Eye test will tell you something else.
    I’m not saying advanced stats aren’t important. However, people on this site get a little carried away with the importance of advanced stats.

    Numbers can be manipulated.

    Comment by Kale — May 27, 2015 @ 11:22 am

  14. This part of Aaron’s post begins with a puzzle: How have the twins scored above average runs with below average hitting? Aaron rationalized the puzzle when he wrote the Twins are clumping hits together, and thats a good way to score a lot of runs. I’m not sure how that is a mark against advanced statistics.

    I think the takeaway from this part of the analysis is that if the Twins continue to hit with the same OPS, one would expect fewer runs in the future. This statement requires the assumption that clumping hits together is not a sustainable skill or strategy. Now if you think that clumping hits together is something the Twins can continue to do, then the Twins can continue to score above average runs with below average hitting (but one might wonder why the Twins should be better at clumping hits than any other team). It just comes down to the assumption we make about the ability to sustain a pattern of clumpy hits. I’m not sure how this is manipulative.

    Comment by Kavan — May 27, 2015 @ 11:39 am

  15. Ha ha ha! You just proved his point.

    “I think the takeaway from this part of the analysis is that if the Twins continue to hit with the same OPS, one would expect fewer runs in the future.”

    Game set match.

    Here is what I say. There are two types of people. People who played, and people who think advanced statistics are the bible.
    Of course there are exceptions, silly! But this is mostly true.

    Comment by Danyo — May 27, 2015 @ 12:42 pm

  16. I didn’t think this was a contest, and I didn’t mean to frame it that way. I actually did play baseball and I don’t think advanced statistics are the bible, neither of which are relevant to the argument.

    I’m still cloudy on what the difference in our opinion is here. Perhaps we disagree that clumpy hitting is sustainable? That’s a completely reasonable counter argument. I don’t happen to agree with it, but it’s not unreasonable. Certainly no need for a ‘gotcha’ moment.

    Perhaps your point is that we can’t measure things, or we are measuring the wrong things and it takes experience playing the game to understand how teams generate runs. Also a reasonable argument, another I happen to disagree with. If that’s the point of contention, then how do we attempt to explain things that appear counterintuitive? Just leave it as “those who didn’t ever play could never understand”? Isn’t there some possibility that thinking about it from an non-players perspective could still shed light on the mechanism, or is that out of the question?

    I’m not out to prove anyone wrong or ‘win’ an argument. I find it interesting that the Twins are scoring more runs than teams that appear to be hitting better. I think it’s interesting to try and reconcile that fact and in doing so we might even be able to learn about the feasibility of sustaining the trend. Most of all I just don’t understand what in Aaron’s original post was controversial.

    Comment by Kavan — May 27, 2015 @ 1:46 pm

  17. I can’t believe this topic is still going. Moving on…

    Comment by Kale — May 27, 2015 @ 3:32 pm

  18. Please do.

    Comment by ML — May 27, 2015 @ 7:13 pm

  19. Oh Snap!!! You go, girlfriend! SASSY!

    Comment by danyo — May 28, 2015 @ 8:39 am

  20. Zzzzzzzzzzzzz

    Comment by ML — May 29, 2015 @ 3:06 pm

  21. Just solid team play is happening, somehow. It’s great when you aren’t digging out of a hole in the early innings, as in the past. And the Twins have no perfect lineup. Dozier is back at lead-off with anyone/everyone else backending the lineup. Hunter at second? But where else would he be, or who would be second otherwise. Mauer is driving in base runners, good. We have no set clean-up hitter. And it’s a lineup that has found places for Escobar, Robinson and the back-down-to-earth Suzuki. You wonder what will happen when the Twins finally decide to play Hicks, Santana, Pinto, Arcia and Vargas everyday — with Rosario and Escobar as the two main bench guys.The bullpen could be a disaster, except Fien is coming back and will take some burden off of Thompson and Boyer. Yet even this last week the amount of innings given to Tonkin, Pressly, Duensing and Graham have been slim, and Stauffer is just taking up space. It’s great when the starters do go long, but you then have to utilize your bullpen so everyone gets work. But IT IS GOOD the starters go long. When Santana returns, who goes. We need to give May, then soon Meyer, and eventually Berrios opportunities to pitch and stay. Poor Milone!

    Comment by Joel Thingvall — May 27, 2015 @ 10:28 am

  22. Meyer? You’re kidding, right?

    Comment by ML — May 27, 2015 @ 7:14 pm

  23. Excellent analysis, Aaron! It does seem like the starting pitching is getting steadily better. Hughes is finally locked in again, Pelfrey should have had a shutout through 7 innings (why are they playing Escobar in left field?), Nolasco is no longer an embarrassment (but still not worth his salary), Gibson and May are solid. I think if you ran the numbers for the month of May, Twins starting pitching would be close to the top of the league.
    The other thing I’ve observed is, the defense is way better since Arcia went on the DL. Rosario/Robinson, Hicks, Hunter is a pretty good defensive outfield. Plouffe and Dozier are playing at a Gold Glove level. Santana, with all his errors, is one of the best I’ve seen at getting to a ball hit in the hole and making a hard throw to first. Again, I think if you ran the numbers for the month of May, the defense would be in the top third of the league. Maybe better. It’s a good thing too, because the offense has been getting worse. This team is starting to look more like the Royals, winning on pitching, defense, and timely hitting.

    Comment by Dave T — May 27, 2015 @ 10:54 am

  24. “I think if you ran the numbers for the month of May, Twins starting pitching would be close to the top of the league.”

    Actually, no. Middle to bottom still

    Comment by ML — May 27, 2015 @ 7:15 pm

  25. Thanks

    Comment by Dave T — May 27, 2015 @ 11:50 pm

  26. For the month of May, Twins are top 10 in all of baseball in ERA. 4th in the AL.

    Comment by danyo — May 28, 2015 @ 8:46 am

  27. What are starters for the month? Season?

    Comment by ML — May 29, 2015 @ 3:06 pm

  28. It’s probably wishful thinking to say our rotation has flexibility (any of these starters could implode at any time) but when Santana returns there are 7 guys for 5 spots.

    I expect the Twins to stick with the veterans, unfortunately. Replace May with Santana. Keep Milone down in AAA until Nolasco totally implodes. If they could get anything for him, I’d like to see them try and trade Pelfrey while he’s on a roll and then bench Nolasco and roll the dice with Milone/May/Gibson/Santana/Hughes.

    Comment by headhunter2 — May 27, 2015 @ 11:14 am

  29. I think Twins will trade someone when Santana returns. IIRC, Nolasco has already said he wishes he were back in the NL. If he keeps performing well, his wish might be granted.

    Comment by Dave T — May 27, 2015 @ 11:59 am

  30. Dumpster fire also brought me here as a first time poster! But while I’m here I’ll add that the ordinal team rankings can be deceptive (just in case that isn’t constantly pointed out here). That is, there’s a lot of clustering around the mean and thus there’s a world of difference between, say, being ranked 21st and having an ERA almost indistinguishable from the mean and being ranked 29th and having an ERA over a full run above it…

    Comment by delc — May 27, 2015 @ 3:11 pm

  31. Great post. Also, Twins are now tied for 1st! This is wonderful.
    It’s so nice to have real outfielders making plays. It changed the game.

    Comment by Kale — May 27, 2015 @ 3:30 pm

  32. Defense is missing from this discussion, as is the role of an added veteran like Hunter. And managing.

    Comment by TR6guy — May 28, 2015 @ 6:05 am

  33. I think overall OPS is a better predictor of future runs scored than OPS with RISP. So based on that alone, one would expect their scoring to drop.

    Balancing out this expected regression, however, is the fact that they have a lot of young hitters who are continuing to learn and improve. I would expect, based on the stage of their career alone, that Santana, Rosario, Arcia, Vargas, Hicks, Pinto, etc., are more likely to improve in overall OPS than regress. Or at least, the ones that continue to get playing time will!

    Add to that the anticipated addition of stud prospects like Buxton and Sano and you have further reason to project an upswing not a downswing.

    Will their runs scored go back to correlating more with overall OPS? Almost certainly. But no one expected them to lead the league in runs per game all season, as they did this month. We just want to see improvement — and boy, are we seeing it!

    The starting rotation is no longer the bottom of the barrel; they have seven guys who can give you a legitimate shot to win every day.

    The relievers are holding their own, and could very likely improve: the late inning guys are fine, and the middle inning guys are likely to be fortified by a starter or two bumped from the rotation. And possibly even Alex Meyer.

    The fielding may not be great, but is more likely to improve than get worse.

    There may be a gut wrenching month or two every now and then when they just stop scoring, or the pitchers implode in bunches. But I don’t think that’s the status quo any more.

    I think the long awaited rise has finally started!

    Comment by by_jiminy — May 28, 2015 @ 8:31 am

  34. Maybe its the optimist in me, but I’d love to identify an explanation for our run-scoring that’s more sustainable than “clutch” hitting. A few ideas/hypothesis:
    1) At some point I read that the twins were emphasizing “quality at bats”. I never saw a definition of this, but if we were to assume that it had something to do with having longer at bats and/or forcing pitchers to throw better pitches (e.g. not chasing pitches that are far out of the zone) than (a) do we know if Twins are actually having more quality at bats than other teams and (b) is it possible that the team is getting something out of these at bats even if they still lead to outs (and therefore don’t influence OBP or OPS), like a better understanding of the pitcher that can be leveraged in the right situation?
    2) One thing that’s been emphasized a couple of times is that we don’t have a regular hitting above 300 — is it possible that a team with a more consistent line-up (ie less deviation amongst the avg/obp/ops of different hitters) has a higher run-scoring potential than a team with the same overall metrics, but less consistency? Or a slightly different version of this question, is it possible that where the hits are coming from in the lineup has an influence in addition to just the total number of hits?
    3) Is it possible that the increase in productivity with runners on is driven not by luck, or “clutch”, but rather by an advantage the Twins have figured out how to exploit? The first thing that comes to mind is stealing signs from the catcher? Or exploiting the change in infield positioning that results from holding baserunners on?

    Would be very interested in hearing if anyone has any thoughts/analysis on any of this ideas! In the meantime, lets enjoy first place while it lasts. Go Twins!!

    Comment by JD — May 28, 2015 @ 6:21 pm

Leave a comment