ERA Lie Detector: Finding 2019 ERA Underachievers With SIERA

May 28, 2020

Using the differences between SIERA and ERA, Nicklaus Gaut evaluates starting pitchers who could see positive regression in 2020 fantasy baseball. While not necessarily an ERA prediction, SIERA gives us our best look at the underlying skills of starters and what we should expect in the future.

Skills-Interactive ERA (or SIERA) has been around since 2011 when it was introduced at Baseball Prospectus by Matt Swartz and Eric Seidman before moving to its current home at Fangraphs. Much like xFIP, SIERA attempts to quantify the skills that underpin a pitcher's ERA, albeit in a different manner. Although technically a backward-looking evaluator,  SIERA is slightly more predictive than xFIP in terms of the following year's ERA and most gets at the how and why of a pitcher's success.

FIP and xFIP generally ignore balls in play, focusing on only the things that pitchers can directly control; strikeouts, walks/HBP, and home runs. SIERA tries to take into account how specific skills (strikeouts, walks, ground balls) interact with each other in order to help pitchers limit runs. Strikeouts are even more valuable in SIERA, as high-K pitchers induce more weak contact, thereby running lower BABIPs and HR/FB%. Walks are bad but not as bad if you don't allow many of them, as they have a lesser chance to hurt you. The more groundballs you allow, the easier they are to field and the more double-play opportunities you'll have.

Essentially, instead of giving "flat-rates" for different skills, SIERA weights them, moving up the skills of pitchers who have high strikeout- or groundball-rates, or low walk-rates. While SIERA is not the "final word" in ERA evaluators (as different evaluators have strengths in different areas), it is quite sticky in terms of the following year's ERA. After previously looking at pitchers who had a higher SIERA than ERA, let's move to the flip side. We'll start today by looking at three pitchers who put up a lower SIERA than ERA in 2019, and judge whether they are worth the draft prices they are currently fetching.

• Fantasy baseball draft kit
• Fantasy baseball rankings
• Team Sync platform and Draft Assistant
• Fantasy baseball mock draft simulator
• Fantasy baseball draft cheat sheets
• Fantasy baseball closer depth charts
• Fantasy baseball prospects

Largest ERA Underachievers in 2019

Looking at starting pitchers in the top-300 of ADP in NFBC leagues, there were just 20 players who had a lower SIERA than ERA in 2019. For sake of consistency, I'll be using dollar values derived via the Fangraphs auction calculator for 5x5 standard 12-team leagues.

Blake Snell, Tampa Bay Rays

2019: 4.29 ERA, 3.56 SIERA (0.73 run differential)

Currently getting roasted for reasons other than performance, Snell didn't reward owners who believed in a repeat of 2018 when he posted a 1.89 ERA and finished as the SP 4 and ninth-highest earner overall. After being drafted around a 30 ADP in NFBC leagues, Snell came nowhere close to earning his draft price in 2019. Between a brutal first half and missing most of the second half following elbow surgery, Snell finished as the SP 66 and 366th overall, putting up a 4.29 ERA and 1.27 WHIP over 107 innings.

Looking just at just the numbers, it seems it was an unmitigated disaster of a year. But was it really? Snell made 20 starts before being shut down in late-July to have surgery to remove loose bodies in his elbow. What jumps out when looking at his game logs, is just how much boom-or-bust he had in him:

When you allow less than two runs in three-quarters of your starts, further examination is warranted. The thing is, Snell didn't have a bad first half; what he had was a bad June. Not just bad but rather a no-good, absolutely rotten and horrible June. Over six starts, he gave up 25 ER in just 23.1 innings, running a 9.64 ERA.

He was better in July, allowing just two runs in three starts, including back-to-back one-run performances against the Yankees. Those starts were deceiving, however, because while Snell may not have allowed many runs, his underlying metrics didn't agree with the results. His FIP and xFIP were both well over 5.00 for both games and Snell stranded 100% of the baserunners he put on.

Sometimes the simplest solution is the most logical. Maybe, just maybe, Snell had been having elbow issues long before he was finally shut down on July 25th. One of the surest red flags for a pitcher having arm problems is diminishing velocity, so how was Snell's speed leading up his eventual shutdown?

The highlighted bar is Snell's first start in June and was also his highest average velocity of the year. In fact, Snell's velocity had actually been slightly rising all season until that zenith. Is it a stretch too far to think that elbow issues had a lot to do with his struggles prior to corrective surgery?

His elbow may have not just been affecting his velocity. Looking at his usage rate, there was a dramatic shift in Snell's pitch mix between April/May and June/July.

The differences are even more clear when looking at his LHB/RHB splits. Along with his fastball, Snell attacks right-handed batters mainly with his curveball and changeup, along with the four-seamer and occasional sliders. Lefthanders, on the other hand, don't get any changeups, with Snell instead switching to a steady diet of sliders and fastballs, while mixing in curveballs.

Right-handed batters had a .161 AVG and 49.3% K-rate against the curveball in April/May and left-handed batters had a 36.4% K-rate and .000 AVG against the slider. And yet, pitching for one of the most data-driven teams in baseball, Snell suddenly flipped a switch after two months of success?

Looking at his heatmaps, is it any wonder that Snell turned down the dial on his breaking pitches, given the difficulty he had locating it? Here's the curveball, his main weapon against right-handers, over the two different periods:

And here the slider, his main weapon against left-handers, over the same periods:

Illusion of Failure

The thing is, is that Snell may have actually been better in 2019 than he was the year before. Even including the terrible June/July, Snell was again a Statcast darling, posting a 4.7 Brl% that was in the top-10% of the league and down from the 7.2% Brl% in 2018. He also had a .205 xBA (top-9%), .327 xSLG (top-8%), and a .264 xwOBA (top-8%) - all were better marks than in 2018.

Let's compare the bottom lines in 2018, April/May of 2019, and June/July of 2019:

Even pitching in a juiced-up offensive environment, it seems that Snell was virtually the same pitcher according to his ERA-evaluators in the first two months of 2019, as he was in 2018. Plus, he increased his K-rate three points and shaved two points from his walk-rate. Then "mysteriously" his velocity ticked down, his pitch mix changed, and Snell proceeded to get knocked around for two months before hitting the IL.

2020 Outlook (48 ADP in NFBC leagues)

As far as I'm concerned, you can take Snell's June/July and throw it out the window. Do we really need more evidence of how long he was pitching hurt and how it changed his pitching and performance? My inner Ockham is screaming the obvious; Snell had to adjust his pitching the more bothersome his elbow became. His changeup usage stayed the same but the curveballs and sliders that put more stress on his elbow were swapped out for more fastballs, even though his velocity was slowly ticking down.

No one should expect the 1.98 ERA that Snell put up in 2018 but his 3.30 SIERA that season says he probably didn't deserve every bit of that shiny ERA. He "only" had a 3.08 ERA in April/May of 2019 but it was right in line with his 3.16 SIERA. The SIERA-based skills say he was the same pitcher before the assumed elbow issues but he had increased his K-rate and decreased his walk-rate. All of that smells elite, to me. The various projections seem to agree, although they're not quite as high as I am.

In our latest staff rankings update, I have Snell as my SP 7 and the 32nd player overall. The only question for me is the health of his elbow. Last summer's surgery may have only been to remove "loose bodies" but Snell had to receive a cortisone shot on March 1st after making only one start in spring training. The Rays said at the time that they did not feel it was a serious issue and Snell has now had an extra three months (at least) to let it rest. He's also reportedly throwing three times a week at home during the pandemic layoff without any issues so I'm optimistic that he'll be fully healthy by the time the season actually starts.

Yu Darvish, Chicago Cubs

2019: 3.98 ERA, 3.55 SIERA (0.43 run-differential)

In many ways, Darvish's final line for 2019 was the best he's put up in years, even though it didn't translate to big fantasy earnings. After only pitching 40 innings for his new team in 2018 due to various injuries, Darvish returned to form in 2019, putting up a 3.98 ERA and 1.10 WHIP over 178.2 innings, going 6-8 with 229 K.

However, with few wins and an ERA near four, Darvish only finished as the SP 34 and 199th-overall. How much drag was the ERA on his fantasy value?

Woof. So, is Darvish just what he is at this point in his career, giving you a lot of strikeouts but paired with an ERA approaching four?

As usual, the devil may lie in the details. While Darvish put up numbers closer to the best versions of himself that we've seen before, his overall numbers simply don't do justice to the second-half dominance the likes of which haven't seen before.

For a moment put aside his ERA and WHIP that looked like the prime Darvish we haven't seen since 2013. Even pass over the 37.8% K-rate; Darvish has always had elite strikeout abilities and sky-high numbers like that are certainly within in grasp for stretches at a time. What really stands out is the basement-level walk-rate of barely 2%. Darvish has never been confused for a control expert, as 7 - 8% has been the norm since the 9 -11% days of his first two years in the Majors.

Combining the minuscule walk-rate with a strikeout-rate approaching 40%, Darvish posted a gaudy 35.6% K-BB% that was the third-highest among starters who threw at least 50 innings in the second half.

That is called keeping good company. SIERA is the king of the ERA evaluators when it comes to predicting next year's ERA but K-BB% is arguably better as a predictive stat. Given Darvish's array of pitches and proven track-record of whiffery, it's not a big stretch that he'll again put up a K-rate over 30%. The real question about which Darvish we'll see in 2020 will come down to whether or not we believe his walk-rate. It's probably foolish to expect a 37.8% K% and a 2.2% BB% but a 27 - 28% K-BB% would still put him in an elite tier of starters.

Location, Location, Location

Looking at Alex Chamberlain's Pitch Leaderboard (which tabulates data from Baseball Savant), Darvish's 2.2% BB% doesn't appear to be much of a mirage, posting a 4.5% xBB% that was the ninth-highest among pitchers who made at least 10 starts in the second half. He also had a 35.4 xK%, giving him a 30.9% xK-xBB% that was the second-highest such mark and less than 1% behind Gerrit Cole.

Darvish boasts a wide variety of pitches but his primary mix consisted of a cutter, four-seam, slider, and two-seam. Comparing the two halves, he spread his mix out by increasing his cutter usage and decreasing the use of his four-seam, slider, and two-seam. He also increased his curveball and split-finger while adding a knuckle-curve that was thrown about 4 mph higher than his regular curve:

It's obvious from above that all pitches allowed dramatically fewer walks, even as the overall performance didn't improve across the board. Let's look again at his first and second half splits, this time focusing on the movement of the pitches that he threw in both halves:

Differences in movement can be caused by a variety of factors, with varying spin and speed affecting how pitches move. Looking at Darvish, however, the biggest difference between the halves seems to be in the consistency of his release points. Not only do consistent release points lead to consistent movement and location but they also make it more difficult for batters to recognize what pitch is coming based on where it's released.

Here are the release points of Darvish's three main pitches in the first and second halves:

It's easy to note the overall tightened cluster of the second half, compared to the first half. Let's drill down more and look at the individual pitches and their subsequent heat maps. First, the cut fastball:

In the second half, the cutter sits on the outside edge for right-handers and the inside edge for left-handers. In the first-half, however, the cutter sat center-cut.

Moving to his four-seamer - of the first-half 18.6% BB% and 1.242 OPS - the story is much the same:

Looking at the above heat-map, are the results any wonder? He was either wild all around the zone or he was railing it in dead-center in the first half. And in the second half, the four-seam stayed up-and-away to right-handers and up-and-in to left-handers.

And finally, here's Darvish's slider:

Understanding why the second-half heat map is preferred means understanding what Darvish was trying to do with the slider. Starting to pitch "backward" more, Darvish used the tightened up slidepiece to steal more strikes early in the count, while still maintaining it's swing-and-miss ability. In the first half, his slider had a 14.5% SwStr% and 21.8% called-strike. In the second half, it dipped slightly to a 13.8% SwStr% but raised way up to a 34.6% called-strike.

Put them all together and you have three pitches with different movement profiles that are being released out of similar slots. Let's see the method in action in a late September at-bat versus Paul Goldschmidt.

Pitch one is an 80 mph slider that starts in on Goldschmidt before sliding over for strike one:

Pitch two is an 85 mph cutter starts middle and cuts over to the outside edge for strike two:

After getting a foul-tip on pitch three, Darvish finishes him off with a 96 mph four-seamer up in the zone. Goldschmidt has no chance:

2020 Fantasy Outlook (56 ADP in NFBC)

Darvish's ADP has only barely risen since the 59 ADP he had during the pre-pandemic month of February. In our latest RotoBaller staff rankings, I have him as my SP 12 and 46th-overall. Like all NL pitchers, Darvish will take a hit if baseball returns with a universal-DH in the shortened season. Even with that in mind, I like him at his current price. Worst-case scenario, I know I'll get a bunch of strikeouts and hopefully a lot more wins given that the Cubs look to be much improved. Best-case, I also get the Darvish from the second-half with the lower walk-rate and ratios that would make him knocking on the top-10 door.

More Sabermetrics & Statcast for Fantasy Baseball