JABO: Bryce Harper’s All-Time Breakout
Bryce Harper was voted the National League’s Most Valuable Player. This is because he was almost certainly the National League’s most valuable player. I don’t want to get into the argument about the definition of “value,” and there are some halfway decent arguments that might conclude Harper got topped, but Harper is a wonderfully deserving winner, and his win comes absent any real controversy. The voting was unanimous, I should probably say. Harper got every single first-place vote. Which meant zero first-place votes for Paul Goldschmidt and Joey Votto, who each had maybe the best seasons for a first baseman since Albert Pujols in his prime. The National League had some awesome players! Harper ran away with things.
This was the year Harper reached a new level. It was a level people long suspected would be achievable for a player with Harper’s skills, but you have to realize how uncommon it actually is for a player to perform around his ceiling. Mike Trout has spoiled us, but now Harper’s up there, too, having broken out. Here’s an easy way to visualize this: Before the year, at FanGraphs, we published player projections. Player projections are everywhere, and we had our own numbers. Then players subsequently posted real numbers. It can be fun to compare the actual numbers and the projected numbers, and here’s a table of the 10 hitters who batted at least 500 times and who beat their projected OPS figures by the most:
Jeff made Lookout Landing a thing, but he does not still write there about the Mariners. He does write here, sometimes about the Mariners, but usually not.
Yes, Beltre moved into a hitter’s graveyard and that’s why he had a crazy career year right before he was a free agent that he never came close to duplicating. *wink wink*
It’s safe to say that Bryce Harper has surpassed me in life, but at least I know how to say the word MEME.
i said the word wrong dozens of times before i finally learned how to pronounce it. dude is human.
With my comment I was being SO MEAN to MVP Bryce Harper.
Give me a break person guy.
Can’t we just enjoy the hilarious comedy that happened? Harper made a tactical smart move by refusing to answer the fool’s question but then went right into saying maymay which is hilarious. Which internet do you use buddy?
chill brah
Harper had an absolutely amazing year based on the normal ways we measure success, but I think it’s worth pointing out that his exit velocity numbers simply don’t corroborate this amazing power breakout he’s experienced in 2015 (obviously doesn’t take away from his amazing OBP, but still).
He ranked 3rd in baseball in HR/FB rate, which would imply he had elite exit velocity in some manner.
He didn’t. He was 41st in exit velocity, behind Juan Lagares and Logan Morrison.
If you want to get more precise and just look at balls in the air, it doesn’t get any better. He was 43rd in line drive/fly ball exit velocity, behind Jayson Werth and Brett Lawrie.
If you prefer fly ball distance, it’s even worse. He was 67th in average fly ball distance. He was 189th (!) in average home run distance, in between Ender Inciarte and Juan Lagares. There were only 221 hitters on the list, via baseball savant.
What does this all mean? While we should definitely appreciate Harper’s amazing season and understand that he’s still young enough to replicate it, the way he accomplished it was unsustainable.
The amazing plate discipline has arrived, but there’s nothing there that indicates he’s actually a .300+ ISO hitter besides his surface statistics.
Actually, I think he did more damage to the belief in the metrics you quote. Or he was just powerball lucky.
He was powerball lucky. See my post below.
Hm. Well, his hard contact% was 40.9%, which was I think #2 in all of MLB.
…perhaps we don’t understand exit velocity…other than the obvious, that is.
I mean exit velocity still thinks he has power on a per batted ball basis. He’s in the top 1/4 of hitters in the more relevant power categories (exit velo, fly ball exit velo, fly ball distance).
What the low average home run distance indicates is that he had a bunch of fluke, short distance home runs. Which was probably to be expected when a guy hits 40 home runs. But basically my point is that he doesn’t have power like Chris Davis or Nelson Cruz, even if his HR numbers seem similar.
In addition to “lucky”, possible answers could include:
a) Good horizontal aim: Pull vs. opposite field, etc…
b) Good vertical aim, aka exit angle.
c) Park effects
Of course it would be the subject of some conjecture how much control Harper might have over (a) and (b)
“What the low average home run distance indicates is that he had a bunch of fluke, short distance home runs.”
or what that means is that he could have hit a whole bunch of flyballs that went for HRs that would have been routine flyouts for a normal hitter.
Or it could mean that when he hit a flyball, he either got under it or got all of it, with little middle ground.
I’m with Lester on this one. Doesn’t Podhorzer have a statistic that he monitors: SDD – standard deviation of fly ball distance? Guys that are high in it hit more home runs.
I’d much rather see a chart of the biggest wRC in a season compared to their previous peak wRC. Saying Babe Ruth “broke out” at age 31 because he went from 133 wRC in an injury plagued year to 216 (when he had surpassed 216 multiple times in his career already) is kind of silly.
In that metric Bryce Harper had a 60 wRC increase (from his 137 two years ago). I would imagine under this new metric Harper’s actual BREAK OUT would still stand out very well, but be in more just comparisons.
That’s what I thought as I read this article. The method here isn’t measuring “breakouts” but just drastic year-to-year disparities. “Breakouts” as a concept are really far, far more interesting.
So I ran the query myself and Harper’s jump of 60 points of wRC+ in 2015 from his *previous high* ranks as the 30th greatest of all time, using Jeff’s 300 PA requirement for both seasons. Al Kaline’s is the biggest and Josh Hamilton’s jump falls off the board once you include his 130 wRC+ seasons in 2007-08. Babe Ruth is no where to be found since 1926 wasn’t even his best season.
Cool stat, I would love to see the list!
https://docs.google.com/spreadsheets/d/1iF8gb76VmxzXUUjLIBs8dbkzWlVn1vSlknrkJOQCees/edit#gid=1255995009
There is good reason to believe that Harper isn’t nearly as good as his superficial stats suggest. His BABIP in 2014 was .369, the 13th best of 300 players with 250 PAs, although he was only #110 in LD%, #230 in IFH%, and #266 in Oppo%. In other words, he got very lucky on BABIP. His HR/FB% was ridiculously high, 27.3% (#5), but he was only #33 of 284 in fly ball distance, so here too he got lucky. But the most striking thing is his poor performance in high-leverage situations throughout his career. I know that there is a tendency not to believe in clutch or choke hitting, but in Harper’s case the numbers are so extreme that it’s essentially statistically impossible that it’s due to chance. In 1659 PAs in LL and ML situations, his slugging percentage is .530, while in 172 PAs in HL situations, it’s only .395. If you do a statistical test on these numbers, the chi-square value is 11.34, p<.001, which means that such a distribution would come up only once in more than a thousand cases if the distribution were due to chance. There is clear evidence that he simply chokes in important situations, and this has been a consistent pattern through his career. Overrating Harper is one of the reasons that people generally overrate the Nationals.
Harper hit 42 HR. How does his likelihood of doing that again have any bearing on whether he broke out this season?
Also, name one person who rated Harper as more than a 10 WAR player going into the season. So, who exactly over-rated him? And if no one over-rated Harper, how is he the reason the Nats were over-rated?
I’m not arguing about whether he had a breakout season, I don’t know of any statistical measure of ‘breakoutness’. I’m just arguing that he isn’t nearly as great as his superficial statistics this year would suggest. Specifically regarding the HR data, there is objective reason to believe he did make a significant improvement in 2015, as his average fly ball distance went up from 287 feet to 299–if you want to call that a breakout, okay, it was a significant improvement, but it was still only #33 in the major leagues, and it isn’t nearly enough to explain the radical increase in home runs. By the Nats being overrated, I wasn’t referring to the situation before this season, I was referring to the discussion during the season–as it became apparent that Harper was having a fantastic season in terms of superficial statistics, there began to be a lot of talk about how could the Mets be keeping up with the Nats, it must be luck, because the Nationals are clearly a better team, and Harper was the main reason people were saying that. I’m saying that this talk was to some extent misguided because Harper wasn’t nearly as good as he appeared to be.
Superficial stats like wRC+?
39th best all-time. If you consider only a player’s best season for such purposes, Harper’s season is 17th best…
His poor performance to-date in high leverage situations might be striking, but there’s zero reason to believe it’s anything more than meaningless noise resulting from a small sample of ABs. I mean you’re talking about outcomes from only 30-50 ABs per season – maybe 2-3 ABs per week that qualify as high-leverage. Your “statistical test” is testing a sample of data that’s simply to small to believe it has any predictive value.
Much more likely that his poor high-leverage performance will regress toward his (higher) true-talent level, as evidenced in his far more meaningful sample of 1658 other career ABs, or as evidenced in his projected .564% SLG% for 2016.
The entire point of statistical significance testing is to see whether a sample size is big enough, and in this case it clearly is, given the enormous difference in HL as opposed to ML and LL situations. If you take a college-level statistics course, you will understand this. Or you can just go ask someone who teaches statistics. If you show them this data, they’ll tell you that it’s so significant that it’s ridiculous to think it’s due to chance.
Harper probably won’t repeat his BABIP, but it could drop somewhat, to say, .330-.340, without having a huge impact on his stats. Also, there is some irony in your arguing, on the one hand, that his BABIP must have been the result of a highly improbable chance, while claiming, on the other hand, that his high leverage stats don’t reflect chance. The 172 PA is a small sample size by any definition, I guess without looking it up about a third the size the sample of BIP that his 2015 BABIP is based on. So if his BABIP is not very reliable as an indicator of future performance, his HL SLG must be even less reliable.
I can accept that your analysis might have shown that there is very likely a significant difference between his HL performance and his LL/ML performance, but it doesn’t follow that the actual values of .395 vs. .530 accurately reflect this difference. Your analysis can only show that there is some difference, that if .530 in 1600+ AB is a pretty good estimate of his SLG in those situations, then if he had performed equally well in HL situations, .395 would be too low for 172 AB (or PA, whatever it is). But that kind of analysis does not show what his true HL value was. Maybe it really was .395, but it could have been somewhat higher, say, .450, with the actual .395 resulting from chance. IOW, your analysis indicates it isn’t .530, but it doesn’t say how much lower than .530 it has to be.
When you further consider that this is a sample over four years, the first three of which he was still developing, and much of the time dealing with injuries, we might wonder how relevant it really is. Even if Harper really performed less well in HL situations over this entire four year period, you can’t be sure that most or all of that difference didn’t result in the first three years. His performance across the board this past season was better than in any of the first three years, and that might include situation hitting. The sample size for 2015, for any single year, is simply too small to tell.
You’re right, we have no way of knowing if the real leverage difference, given an infinite amount of data, would be .135. It might be bigger, it might be smaller. But there is no statistical evidence for it being smaller or larger.
I don’t see how dealing with injuries would affect his relative leverage performance. Variation being different leverage situations occur within single games, and presumably within each game his injury status would basically stay the same.
Here are HL/ML/LL slugging% for each of his four seasons:
2012 393/475/493
2013 387/582/426
2014 355/438/424
2015 426/671/677
As you can see, this has been the pattern in every year of his career, and although obviously his overall numbers went way up in 2015, the difference between HL and ML/LL went up even more, to about 250 points. Of course for any single year, the difference might be accidental because of the small number of HL ABs, but when all the data are included, the statistical evidence is undeniable.
The statistical evidence is undeniable? Give me a break. He’s a 23 year old kid who just put up a 10 WAR season. If he’s pressing too much in clutch situations, he can figure that out and stop it. Just like he stopped running into walls and getting hurt.
Or maybe he’s just taking garbage pitches and walking or hitting singles rather than putting them over the fence.
OK, I took a look at his splits, and though the sample size as we all agree is small, there does seem to be a trend there. It seemed possible to me that in HL situations pitchers would pitch around him even more than in other situations, so he might take more walks, but losing patience sometimes, swing at poor pitches and make weaker contact.
The data do bear this out. In HL situations, his career BB% is 17.3% vs. 12.9/12.1 in LL/ML. I didn’t run any tests, and don’t know if that’s significant, but it’s obviously quite suggestive. However, his HL BA as well as SLG is much lower, with the result that his OBP in HL situations is lower than LL/ML, .352 vs. .378/.401. Again, I didn’t run any tests, but I doubt that difference is significant. But even assuming it’s not, we can conclude that in HL situations, at best he maintains his OBP–with greater welking compensating for fewer hits–while his SLG suffers, and thus his OPS, wOBA, and wRC+, etc. Thus his career wRC+ is 98 in HL vs. 145/162 in LL/ML.
This year, however–and again, the caveat that we’re dealing with a really small sample size–there was no significant difference in BB% in the three situations, 18-19% in all cases. As with his overall career, his BA was lower in HL, with the result that his OBP was about .100 lower in HL than LL/ML, and his wRC+ was 112 vs. 208/206.
So my tentative conclusion is that Harper sees more pitches out of the zone in HL–certainly a reasonable thing to expect–and when he does swing at them, is more likely to make weak contact. This year,when he was walking a lot even in LL/ML, he might have reached a saturation point, i.e., his patience reaches a limit at a certain BB%, with the result that he’s no longer walking more in HL than in LL/HL. Coupled with continued weaker contact with out of zone pitches, his OBP and overall hitting performance suffers even more.
I don’t know if I’d call that choking, but it has the same result, poorer performance. Assuming this view is correct, the only thing he can do is take even more walks.
For comparison, I looked at Trout. For just this year, his BB% looks a lot like Harper’s career: 12.3/13.9 in LL/ML vs. 19.6 in HL. However, his BA was also higher in HL, with the result that his OBP was higher. His SLG was also much higher. Overall, his wRC+ was 227 in HL vs. 178/152 in LL/ML.
He only had 56 PA in HL, though. What about his career (261 PA)? The same pattern is fairly evident. In HL, his BB% was 16.1 vs. 12.4/12.0. His BA was also slightly higher, with the result that his OBP was higher, .429 vs. .396/.391. His SLG was about the same, .568 vs. .574/.536, and his wRC+ also about the same, 170 vs. 173/159. Probably none of these differences is significant, but overall, the conclusion seems fairly safe that Trout performs about as well in HL as LL/ML situations. The fact that the numbers are so close, moreover, gives me a little more confidence that the differences for Harper are real, though I’d still like to look at other hitters.
So while Trout probably gets pitched around at least as much as Harper in HL situations, when he does swing, he seems to make at least as good contact as in less critical situations.
My guess is that he may be trying to do too much in clutch situations. Teams are pitching around him and he’s swinging at bad pitches anyway. It would require a lot more detailed study to figure this out.
The reason he hits worse in high leverage situations is likely to be that high leverage situations are generally at the ends of games when he will be facing the toughest left handed reliever or closer available to the opposing manager. The base line comparison should not be to his low leverage at bats, it should be to how he does against similar pitchers. Like most left handed batters, he does have a pretty pronounced platoon split. This isn’t because he isn’t clutch, it is because left handed pitchers are actually more difficult to hit for left handed batters.
The HL/LL comparisons probably have nothing to do with being clutch or not clutch. They have to do with facing tougher pitching in those situations. Harper is the hitter that other manages reserve their best high leverage relievers for.
What would Harper’s fly ball distance be if his home field was Coors? Or Chase field? Or Yankee Stadium?
Aha, who is being superficial now.
Love that you know what a chi-square value is, but you don’t know how to remove extraneous bias from your samples.
It is very welcome to see inferential statistics applied here. However, the Chi-Square test is not appropriate for slugging percentage. This is because it is, of course, not an actual percentage.
Sorry, I meant 1658 ABs, not 1659 PAs.
Well, I was on board when it was 1659, but now with the lower PA total I’m skeptical that it’s a meaningful sample size.
Only 99 ribbies? You get to be the MVP for that now?
This is exactly my point (see above). He’s simply a terrible clutch hitter.
It’s also worth mentioning that Harper’s fWAR was the 5th best all-time for an age 22 season, trailing Williams, Collins, Musial and Cobb. His rWAR (BBRef) was second all-time, trailing only Mr. Ballgame.