It’s still extremely early in the season, which makes it very difficult to do much meaningful analysis. We’re heavily in the “small sample size” portion of the season, where fans love to overreact to fast starts only to be disappointed when reality brings these players back down to ground. That doesn’t mean that some of these breakouts aren’t genuine improvements in a player’s talent, but it can be difficult to find that in a world currently filled with SSS noise.
So with that in mind, let’s take a look at Merrill Kelly. Most of the players on the team have extensive MLB/professional experience prior to this season, so we have a general sense of what they are or might be in 2019. Merrill Kelly is an exception, having played in Korea the past several years and not throwing a professional pitch in America since 2014.
So far, in 3 starts, Merrill Kelly has actually looked good. He’s averaged 6.33 IP per start and currently has a 3.79 ERA with 17 strikeouts against only three walks. For a player that was destined to be the #5 starter and that didn’t look great in Spring Training, that’s really good.
But how “real” is this start from Merrill Kelly? Kelly hasn’t looked like an ace or even a #2, but his current stat line looks like a solid #3. 3.79 ERA, 3.89 FIP, 8.05 K/9, 1.42 BB/9, 1.42 HR/9. Can he keep this up?
Let’s start with the easiest: the 3.79 ERA doesn’t really mean much at all in 19 IP. It’s terribly unpredictive on its own. So let’s throw that out.
But what about FIP, K/9, BB/9, and HR/9? Well, as you all should remember, FIP is simply a combination of a pitcher’s K/9, BB/9, HPB/9, and HR/9. So while the FIP currently “validates” the ERA (though the sample size is too small for it to be very meaningful... yet), let’s throw it out. Instead, we’re going to look at K/9, BB/9, and HR/9.
In general, the thought process of analyzing anything (whether it’s baseball or something completely different) is that inputs -> outputs. In any equation that is 100% known, the same inputs will always yield the same outputs. However, real life is much more complicated than a simple equation and thus we will never have a “100%” baseball equation (this is what R^2 is telling you on any stat). However, don’t be discouraged - while FIP as a whole might still be unstable in 19 innings, the individual components that make up FIP (K/9, BB/9, HR/9) will often stabilize at a much faster rate.
In general, individual components of an equation will always have less variance and noise than the equation as a whole.
So in the case of Merrill Kelly, we need to look at his K/9, BB/9, and HR/9 to assess how accurate they are to his true talent level. We’re still too early to have a lot of confidence in our assessment, but we can at least test what we’ve seen so far.
To do this, we are going to look at Kelly’s xK%, xBB%, and xHR/FB%. These stats have been referenced here several times (well, xK% and xBB% have), because they have very strong predictive capabilities and they stabilize quickly.
Since we’re using K%, BB% and HR/FB% instead of K/9, BB/9, and HR/9, we are going to need Kelly’s data for that:
So let’s start from the top. How does Kelly’s xK% look? Per Fangraphs, the equation for xK% is as follows:
xK% = -0.8432 + (Str% * 0.2916) + (L/Str * 1.2689) + (S/Str * 1.5334) + (F/Str * 0.9672)
Str%: Strikes / Total Pitches
L/Str: Strikes Looking / Total Strikes
S/Str: Strikes Swinging Without Contact / Total Strikes
F/Str: Pitches Fouled Off / Total Strikes
The R^2 for this equation is 0.931, which means that given an adequate sample size, this equation can accurately predict 93.1% of a player’s true strikeout ability. Kelly is not currently at an adequate sample size, but the linked article above did study this at less than 50 IP for pitchers and still found an R^2 of 0.632, which is still very strong. At this point, we can probably “predict” about 50-66% of Kelly’s true strikeout ability.
To find the above stats is very simple. Just go to Merrill Kelly’s Baseball-Reference page and go to “Finders & Advanced Stats”, click on “Advanced Stats” under “Pitching”:
Then scroll down to “Pitch Summary -- Pitching *”:
Notice how all four inputs are listed in this table - it makes it REALLY easily to calculate for any player!
Anyhow, the stats for Kelly are right there:
Input these into the equation and you get:
Merill Kelly’s “expected” strikeout rate is 21.1% while his actual performance, so far, has been 22.7%. And that’s actually a good thing! While Kelly seems to have overperformed his expected strikeout rate, it isn’t by a super large margin - it’s literally one extra strikeout. Kelly currently has 17 strikeouts against 75 batters faced; drop that to 16 strikeouts and his K% drops to 21.3%, extremely close to his expected rate.
And this is all excellent news for Kelly because it suggests that he might be closer to having an average strikeout rate than expected. The league average for starting pitchers in 2018 was 21.6%. Expectations weren’t high for Kelly as he was primarily viewed as being a “strike thrower”. Even this author was hoping for only 6-7 K/9 at best from Kelly. Instead, we’re looking at the potential that Merrill Kelly might be closer to average than not at getting strikeouts. This bumps Kelly’s upside considerably.
Furthermore, there is another factor here: notice how three of the inputs above all deal with throwing strikes, more-or-less (Str%, L/Str, F/Str). These all play into Kelly’s strength in that he’s a strike thrower. This suggests that Kelly’s strikeout potential might be more believable than we were once lead to believe.
But what about his walk rate? Kelly’s BB% of 4.0% is making Robbie Ray and Zack Godley salivate right now. But can Kelly keep that up?
xBB% = 0.564 — 0.484*Str% — 0.216*K% — 0.528*I/Str + 0.505*(3-0%)
Str%: Strikes / Total Pitches
K%: Strikeouts / Total Batters Faced
I/Str%: Balls Put Into Play / Total Strikes
3-0%: 3-0 Counts Seen / Plate Apperances
Again, all of these inputs (except K%) can be found in this same table. For Merrill Kelly, they are:
And this gives us:
Again, Kelly has overachieved but not by a very large number. This is much less than 1 batter, as adding an extra walk to the 75 batters faced would raise his BB% to 5.3%. In other words, Kelly is well within the margin of error of the statistic.
So while Kelly is hovering just below league average in terms of strikeout rate, he is considerably better than the league average walk rate of 8.0%. And it looks to be pretty real, at least so far.
And again... notice how all but one of the stats rely on throwing strikes: Str%, I/Str, 3-0%. This plays right into Kelly’s strength. Maybe this could be more sustainable for someone like Kelly than expected?
And then there’s the batted ball data. Specifically, homers. Kelly seems to have avoided hard contact for the most part, though his 15.0% HR/FB% is worse than average. It’s also a statistic that is terribly suspectible to noise.
Fortunately, we have a predictor for HR/FB% as well. And while it was designed for hitters, it can be used to decently analyze pitchers as well:
xHR/FB = -0.0066 + ([Fly Ball Pull% + Fly Ball Oppo%] * 0.0671) + (Brls/BBE * 1.2479)
Fly Ball Pull%: Pulled Fly Balls / Total Fly Balls
Fly Ball Oppo%: Oppo Fly Balls / Total Fly Balls
Barrels/BBE: Barrels / Batted Ball Events
This equation is pretty nifty because it shows how useful the “barrels” statistic is from Statcast. The other two stats are mostly hitter-derived stats, not pitchers, so it does make the equation a bit noisier for pitchers. For pitchers, we are primarily worried about the rate at which they give up barrels. In Merrill Kelly’s case, so far, he’s been good.
To find the Barrels/BBE leaders, just go to the Statcast Leaderboards. The link has been filtered to pitchers with a minimum of 40 batted balls this season, which leaves us with 107 pitchers. Out of these 107 pitchers, Merrill Kelly ranks 45th with 7.3%. Pretty middle of the pack.
If you take thie 7.3% and add it to the other two stats above, you get:
In this case, Kelly has slightly underperformed his expected home run rate; however, he is still within the margin of error (if you reduce his homers allowed by 1, his HR/FB% would drop to 10%), so it’s hard to say which way he’ll go at this time. The MLB average for starting pitchers was 13.1% last year so his expected rate is pretty close to average.
The rest of his batted ball data looks pretty good, too. His average exit velocity allowed on fly balls and line drives is the 16th-lowest out of those 107 pitchers. He’s also better than average in Hard%, 95+%, and overall average exit velocity. Hitters aren’t squaring him up, as one might expect with the “strike thrower” archetypes.
So far, this is what we’re seeing from Kelly: average strikeout rates, average home run rates, well above average walk rates. That actually makes for a good, but perhaps unspectacular, pitcher. Putting this all together, and Merrill Kelly, so far, looks like he could have a FIP in the high 3s or low 4s. That is considerably better than the back-end starter expectations that we’ve been giving to Kelly so far.
Just remember: this is only 19 innings. There is a lot of season left. It will be hard for Merrill Kelly to do much better than he’s doing now while there is a lot of room for him to regress negatively. However, the early state of Merrill Kelly is currently backed up by the predictors.
It’s likely that Kelly will continue to excel at throwing strikes. That was his M.O. when he was scouted in Korea and brought back to the US. The real question is whether he can continue to be roughly league average at generating strikeouts and preventing homers. If he can accomplish these two feats, he’s going to be a very solid pitcher. He’s not going to be an ace or a #2 without developing a good strikeout pitch, but he’ll certainly blow away expectations.