The All-Star Game is only a few weeks away, and baseball's top starters have thrown nearly 100 innings apiece. It's far enough into the season that we can take a look at which pitchers have taken a huge step forward from last year, and more important, try to figure out why. (We did this last year around this time, identifying Justin Verlander and Tyler Glasnow at the top of the list.)

Since we're trying to get at underlying skills here, it's not good enough to look at ERA, which can be negatively affected by one terrible start or poor defense or relievers. (Look no further than Max Scherzer for that; he's held his strikeouts steady, lowered his walks and homers ... and has seen his ERA increase.)

Instead, we'll look at the Statcast Expected wOBA metric. The short version is that it looks at quality of contact (exit velocity and launch angle) and amount of contact (strikeouts and walks) in an attempt to get to a defense-free measure of how a pitcher is performing, outputting a number on a scale similar to on-base percentage. For context, the 2019 Major League average is .322.

We'll set a minimum of 200 batters faced last year and 150 faced so far this year. That does cut off some interesting improvement by relievers, including Ian Kennedy, Francisco Liriano and John Gant -- and it's pretty difficult to ask someone who was already elite like Scherzer to actually improve -- but we have to draw the line somewhere. Here are 2019's most improved pitchers.

Lucas Giolito, White Sox

-.086 (down from .350 to .264)

Likely reason: So many reasons

This first one is easy, because we went in-depth on Giolito's transformation from "first-round bust" to "AL Cy Young contender" just a week ago -- the day before he struck out 11 Royals in 7 2/3 scoreless innings.

As we noted, there wasn't any single reason to point to that explained Giolito dropping his current ERA to 2.28 from 6.13 last season. There were so many. Giolito overhauled his training, both physically and mentally; he shortened his arm action, which appears to have helped with his consistency and deception; he's throwing his four-seamer harder; he's mostly ditched his ineffective sinker for more of that strong changeup. There are so many things that had to happen here for success to follow, and Giolito has more than earned a nearly certain spot among the American League All-Stars.

Martin Perez, Twins

-.076 (down from .364 to .288)

Likely reason: That new cutter

Perez seemed like he might be on his way out of baseball after an injury-plagued 6.22 ERA season for the Rangers last year, but he landed in Minnesota with new mechanics and a new pitch. The former has helped him up his velocity to 94.5 mph from 92.8 mph; the latter, a cutter, has suddenly become his primary pitch, as he's using it 34 percent of the time to great effect (a .162 average against). Overall, no AL pitcher has allowed a lower hard-hit rate.

Perez has had a recent rough patch, allowing 15 runs (12 earned) in his last three starts. That he still ranks this high even after all of that shows just how far he'd ascended over last season.

Tyler Glasnow, Rays

-.073 (down from .305 to .232)

Likely reason: A change of scenery

Yes, Glasnow also appeared on this list last year, though at that time he was moving into the bullpen for Pittsburgh. He moved back into the rotation with the Rays after they acquired him in what now looks like a brilliant -- or disastrous, depending on your perspective -- trade for Chris Archer, and his true strides came in his first eight starts of 2019, when he posted a 1.86 ERA while striking out 55 and walking only nine. (He landed on the injured list last month, but is expected back shortly after the All-Star break.)

Like Charlie Morton and Gerrit Cole before him, Glasnow appears to have benefited by a change of scenery out of Pittsburgh. In this case, it's a little about pitch usage -- he was throwing 25 percent curveballs with the Pirates before the trade last year, and that was up to 31 percent with the Rays this year -- and a lot about control, as his percentage of first-pitch strikes has jumped to 63 percent in 2019 from 52 percent pre-trade. With his elite stuff, that might be all he needs.

Frankie Montas, A's

-.068 (down from .351 to .283)

Likely reason: That new splitter

When we wrote about the best new pitches of 2019 earlier this year, Perez and his cutter were at the top, and Montas and his splitter were close behind. Montas made it very clear in Spring Training that adding a usable third pitch was important to him, and it's been something of a game changer, as he's using it a quarter of the time and allowing just a .197 average on it.

Montas has a 2.84 ERA this year, and his strikeout rate is up to 25 percent from 15 percent. It's a massive jump, and a long time coming for a thrice-traded former top prospect.

Stephen Strasburg, Nationals

-.061 (down from .306 to .245)

Likely reason: More grounders

After all these years, all the ups and downs, it feels like we still don't fully appreciate how great Strasburg is. Since his 2010 debut, he's 11th in Wins Above Replacement and 15th in ERA. This year, he's 6th in FIP (Fielding Independent Pitching) and first -- first! -- among pitchers with 200 batters faced in Expected wOBA. If he's not quite Scherzer, he remains one of baseball's best starters.

It's not even like he was coming off a down 2018, either, though his 3.74 ERA was a career high. This year, his strikeouts are up slightly and his walks are down slightly, but the difference here is clear: Only five pitchers, among those with 80 innings both this year and last year, have added more grounders. Last year, Strasburg had a 43.6 percent grounder rate, and now it's north of 51 percent.

It's not hard to see why, either. Strasburg's two best grounder pitches are his curve and his sinker, and he's throwing both more, at the expense of a four-seamer that allowed a .526 slugging percentage last year. More strikeouts, more grounders, and fewer walks? It's a nice combination.

Matt Andriese, D-backs

-.056 (down from .336 to .280)

Likely reason: Huge drop in hard-hit contact

Andriese had been a valuable swingman for the Rays, but after being traded to Arizona last summer, he struggled, posting a 9.00 ERA in 14 games for the D-backs. Even now, his ERA is still only 4.63, which isn't terribly impressive. So what gives? How is he on this "most improved" list?

It's a little bit about strikeouts, because he's upped that to 25 percent from 23 percent, but it's mostly about avoiding hard-hit contact. Last year, Andriese allowed a 46 percent hard-hit rate -- meaning nearly half of the batted balls he allowed were hit at 95 mph or harder -- which was the sixth-highest among pitchers allowing 100 batted balls. This year, that's all the way down to 31 percent, and when he allows two more batted balls to qualify, it will be a Top 25 mark. It's the largest yearly drop of our qualified group.

The results haven't followed yet, and we can't guarantee they will. But there's at least something to like here, something to find interesting. Andriese has found a way to avoid all that hard contact.

Miguel Castro, Orioles

-.054 (down from .353 to .299)

Likely reason: More velocity

Castro has a 5.86 ERA. He had a 3.96 ERA last year. We're going to have to explain this one, clearly. There are a few things to consider here, the first being that the 3.96 ERA wasn't really a great indicator of his 2018 skill to begin with. Advanced ERA estimators, like FIP (5.11) and DRA (6.49!) suggested that his 3.96 shouldn't have been taken that seriously.

Second, despite the ugly 5.86 -- mostly due to two April games where he allowed four runs apiece -- there has been improvement here. Castro has increased his strikeout rate to 22 percent from 15 percent. He's cut his walk rate to 11 percent from 13 percent. He's talked about mechanical changes, but it's a lot easier to see the huge jump in his sinker velocity, up to 96.9 mph from 95.4 mph, the seventh-hardest in baseball.

Since that second four-run blowup on April 22, Castro has allowed a line of .183/.258/.378 with a 3.42 ERA. He's not elite, or really close to it. But he's better, and still improving. It feels like forever ago that he was part of the Troy Tulowitzki/Jose Reyes trade, yet he's still just 24 years old.

Jake Odorizzi, Twins

-.054 (down from .338 to .284)

Likely reason: More velocity, more cutters

Even if you don't believe in Odorizzi's 1.92 ERA -- and you probably shouldn't -- there's been an undeniable step forward here from a veteran starter who never seemed to be able to make that next leap above a solid mid-rotation guy. In Odorizzi's case, it's a little that his strikeouts are up (to 29 percent from 23 percent), a little that his walks are down (from 10 percent to 8 percent) and a lot that he seems to be the only pitcher in the game to avoid a home run problem. (Last year, Odorizzi allowed 1.1 HR/9, and this year, that's been cut in half, to 0.54/9.)

Over the winter, Odorizzi became one of many pitchers to experiment with new technologies like Rapsodo, saying he "feels tremendously better than last year." He's upped his fastball velocity to 92.8 mph from 91.1. He's also throwing a cutter 17 percent of the time, up from 2 percent. (Also, for what it's worth, Perez credits teammate Odorizzi with helping him fine-tune his own effective new cutter.) This is oversimplifying a little, but both Perez and Odorizzi have Top 10 cutters this year -- and Odorizzi seems to be responsible for both.

Luis Castillo, Reds

-.054 (down from .320 to .266)

Likely reason: More ground balls

Castillo -- one of many Reds pitchers taking a nice step forward this year -- has upped his strikeout rate, to 29 percent from 23 percent, but he's also seen his walk rate nearly double to 12 percent from seven percent. That's still a net win for him, but the real difference here is where his batted balls are going.

Remember above when we said that only a few pitchers had added more grounders than Strasburg? No one has added more than Castillo. He's now got baseball's third-highest grounder rate, except neither Dakota Hudson nor Mike Soroka miss bats like Castillo does.

That's a jump to 57 percent from 46 percent, and that's enormous. Pair it with a drop in hard-hit rate to 33 percent from 40 percent, you can see what the effect is here: More balls on the ground, and more balls hit softly. (If the ball is hit at all, obviously, which isn't a given considering the jump in whiff rate.)

There's no secret here as to why. Castillo's changeup gets the fourth-most grounders of any changeup, and he's throwing it more this year, up to 32 percent, nearly as much as his four-seamer.

Ariel Jurado, Rangers

-.047 (down from .366 to .319)

Likely reason: Changes in pitch usage

It's easy to look at the ERA drop to 3.02 from 5.93, but we're trying to get deeper than that. A good place to start would be with strikeout rate: Jurado's 9.1 percent was the lowest in baseball in 2018. This year, that's up to a more manageable 19.2 percent, even as he's recently moved into the starting rotation. So there's that, but also he's cut his hard-hit rate down to a roughly average 36 percent from last year's too-high 48 percent.

More strikeouts and fewer hard-hit balls is a pretty nice combination. It only makes Jurado a league-average pitcher, but there's value in that, especially since he was something much less than that last year.

How, though? Since so little worked last year, changes were necessary. He's stopped throwing his sinker so much, down to 41 percent from 56 percent, and his four-seamer is coming in slightly higher, and, maybe most important, he's figured out a way to solve lefties.

Last year, Jurado threw the sinker to lefties 46 percent of the time. It got pounded (.478 average, .804 slugging.) This year, that's down to a quarter of the time, replaced by the four-seamer and the changeup -- which has been, so far, one of the three most effective changeups against lefties in the game.