Hanley Ramirez and Batted-Ball Data
It seemed like this post was practically going to be able to write itself. Hanley Ramirez has been hot at the plate, and he’s tied for the big-league lead in homers, with 10. There are hundreds of hot streaks by so many players every single season, but this year we have the treat of new data, and Ramirez’s has seemed particularly remarkable. I thought this would be simple and straightforward, but instead we have something more complicated and kind of boring to what I assume would be the majority of people. Keep reading, though! There’ll be some .gifs. You love .gifs.
If you’ve paid attention to Gameday, you’ve probably noticed that we’ve started to get some early-season batted-ball data. It hasn’t been complete, but it’s been fairly consistent, as one of the first signs of the rolling out of StatCast. It can be tricky to find and preserve that information, but thankfully for the masses, there’s Baseball Savant, which I feel like I must link in every post. There, for the first time, we can sort hitters by batted-ball velocities. The industry has had HITf/x for years, so this isn’t progress for them, but it’s progress for us, on the outside. And we all love a new toy.
So, play with the toy. I sorted all the hitters in baseball by average batted-ball velocity, setting a minimum of 30 balls hit fair. It’s early, for every statistic, but that doesn’t mean you can’t be curious. And the results already seem to be decently sorted. Around the bottom, in average speed: guys like Billy Hamilton, Ichiro Suzuki, and Alberto Callaspo. Around the top: guys like Alex Rodriguez, Josh Donaldson, and Giancarlo Stanton. That’s all intuitive and encouraging. That suggests there’s something real being measured.
At the very top is Hanley Ramirez, with an average exit velocity of 98.7 miles per hour. As a good hitter and the co-leader in dingers, one shouldn’t be surprised to see Ramirez ranked so high. But you also notice his lead — almost two full ticks over second place. More than four over fourth place. I’m interested in big leads on leaderboards, and Ramirez got my attention.
I then decided to measure something related: rate of balls hit at least 100 miles per hour. It’s an arbitrary cutoff, but an appealing one. I kept the minimum of 30 recorded hit baseballs. Again, toward the bottom you find Hamilton, Ichiro, and Callaspo. Callaspo’s actually the one guy in the sample without a single such hit ball. At the other end, Stanton, Paul Goldschmidt, and Jorge Soler. Again, it’s all looking good. But there’s Ramirez at the top, and by a considerable margin.
The top of the list:
- Hanley Ramirez, 59%
- Giancarlo Stanton, 50%
- Paul Goldschmidt, 47%
- Jorge Soler, 45%
I can buy Ramirez being in the group, but something didn’t sit well with me. I wanted to check for video confirmation, and while there’s no such thing as accurately eye-balling batted-ball velocity, you should at least be able to get some sense of any egregious data errors. So I went to MLB.tv to check balls in play against a spreadsheet. I looked only for balls said to have been hit at 100mph or higher. I started at the beginning.
I was immediately confronted with a groundball, although to be honest I could see this one being super fast off the bat:
So that didn’t disprove anything. Maybe that really was hit at 100+. It’s not like only line drives are well-struck. But, I’m pretty certain this one wasn’t 100+:
Ditto this one:
And this one:
And this one:
I stopped there because I didn’t need to see more. I didn’t want to look at every single ball Hanley hit between the lines — I just wanted to see if I could trust the data. And I think most of the data is fine. But it sure looks like there are also errors, and the samples at this point are small enough for those errors to make a difference. The raw data says Hanley has hit the ball harder than anyone. He’s certainly hit the ball hard — that’s how you get dingers — but it appears the numbers are influenced by some exaggerated readings.
Maybe this was all unnecessary. Maybe all I had to point out was this: according to the data, on April 10 Anthony Rizzo bunted a ball toward Nolan Arenado at 103 miles per hour. That’s insane and stupid. I did look at the video, just to be absolutely sure. Yeah, no. It’s insane and stupid. It’s a mistake.
There are mistakes. That’s the point. This is the second post I’ve published today focusing on data errors, and that wasn’t intentional. Nor is the material of broad interest. This isn’t the FanGraphs mission, and I don’t want to make a habit of this, but I do think it’s worth noting when something misleads. Especially something new, that everyone wants to play with because they’ve never had access to it before. Just, be careful. When the numbers first started showing up, Dave warned all of us to be careful, because it’s early and the system could have glitches. You’d think it’ll be smoothed out over time, but just don’t put complete faith in the numbers. If something extraordinary is brought to your attention, it’s not a bad idea to confirm whether it’s actually true.
As for Ramirez himself — I’m sure his true average exit velocity is high. The data we see blends both errors and legitimately well-struck drives. He’s always been strong, with a quick bat, and you don’t share a dinger lead with Nelson Cruz if you’re rolling over on the ball. But the numbers I wanted to use in this post: they seemed hard to believe, and, it seems they were too good to be true. So we can’t yet write about Ramirez in the way we’d like to in 2015. We have to write about him in the way we would’ve in 2014. At least until there’s more data, and more reliable data. There’s no reason for us not to be patient.
Oh, and, Hanley has 10 homers and zero doubles or triples. And he’s been worth just 0.3 WAR, because he’s been so bad in left field. It’s been a weird month. A weird and instructive month.
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.
So you write this whole thing about bad data and then you conclude with his low WAR based on 19 games of fielding data?
also based on literally everyone who has been watching Red Sox games
As someone who has been watching many Red Sox games, I’m not sure anyone actually told Hanley he’s playing left field. Someone just blindfolds him and walks him out there every half inning and lets him figure out what to do from there. He’s still not quite caught on yet.
A thought experiment: Who would be better at LF? Hanley Ramirez, or…
if you were to take a vaguelly athletic person who has never heard of baseball, and put them in left field and told them simply to try to catch the ball and throw it back towards where it came from.
I’m not sure of the answer. But it’s great to have Manny back on the team.
Agreed, it was a much less amusing team without him
Has he Hanley visited MannyLand yet? During a game, I mean.
At least the Red Sox fans get his production at the plate. Reds fans get his defensive doppleganger who 1) can’t stop swinging and 2) can’t stop missing.
Apparently it takes a larger sample for someone to believe that a bad shortstop/third baseman who has never played left field is also a bad left fielder.
My favorite Hanley stat this year: he’s hitting .293 with a .233 BABIP.
I’ve seen Ramirez lose an unusual number of well hit balls up the middle to infield defenses that must have been biased that way–not bad luck?– and line drive outs to CF–bad luck.
I wish Napoli would look at the position of Ramirez’ hands at contact. He might be able to hit a ball above the belt.
That sort of makes sense. He loses 10 “hits” because homers aren’t included in BABIP, and he hasn’t really hit any other XBHs.
indeed.
turn those 10hr into doubles off the monster and his babip is .343.
I am not convinced by the gif’s you have posted. Just because a ball is hit at a steep downward angle does not preclude it from coming off the bat at 100 mph. Such a ball will likely lose a lot of speed after hitting the ground, so you can’t learn much about the speed off the bat (which is what is being reported) after the first bounce. And there is very little trajectory to see between impact with the bat and first bounce to help in assessing visually the speed. But Trackman has the ability to measure over that short path.
Now, you may well be right, but I am not convinced that you have proved it with the gif’s you showed.
While it’s absolutely true a groundball can be extremely well-struck, what troubles me is that we’re seeing Hanley make relatively poor contact, in that he’s not squaring these grounders up, for the most part. They’re glancing off the bottom of the bat, yet the reading is that he’s hit some of these grounders harder than he’s hit some of his homers. I’m definitely not the expert on this, so maybe I’m wrong — and I would *love* to be wrong! — but then, if nothing else, there’s the matter of the Rizzo bunt. If this was 103 miles per hour, then there is a lot I don’t understand.
They all seem to be of a similar type — not rollers, exactly, but ground balls drilled right into the dirt 5-10 feet in front of home plate. Either way, I think the conclusion holds, which is that batted ball velocity is not the only thing we should care about. I am much more interested in it when combined with launch angle, which unfortunately is not fully available yet.
I agree there are errors which are clear to the naked eye, but comparing Rizzo’s bunt to these grounders is a bit of a straw man.
Take a look at that 3rd gif. The ball is chopped almost straight down, yet the second hop is almost at the shortstop. Is this 100+ MPH? I don’t know. Is it plausibly 100+ MPH? I believe so.
If a ball leaves the bat at 100+ MPH at an elevation of +25 degrees, that’s a HR. If it leaves the bat at the same absolute angle, but negative, there’s a good chance it’s a groundout. Similar to one of the gifs above.
Assuming a standard margin of error on a swing, intuitively there would be an equal amount of 100+ MPH groundouts as there are HR.
I like that argument!
Rizzo’s 103MPH bunt isn’t a straw man. It might be a red herring, but straw man doesn’t really fit in this case. It’s excellent evidence that there’s something going weird somewhere in these numbers, but it doesn’t necessarily mean that Ramirez’ own numbers are incorrect. Ramirez’ may very well be hitting some crazy-hard grounders, although it seems like there’s good reason to take a step back and question whether these measurements are totally accurate. Not to take away from Ramirez – he’s certainly crushing the ball.
Let me amend my previous comment. I like the argument, but just a reminder that there is not up-down symmetry in the problem, since the ball always approaches home plate on a downward trajectory. So I wouldn’t argue that +25 deg and -25 deg are exactly equivalent…but they are approximately so.
If these grounders are actually leaving the bat at over 100MPH, I think the conclusion is that MPH off the bat is useless, since you can achieve a high average speed without hitting the ball well. That might be the case, but that would be super disappointing. Here’s hoping these grounders are data errors.
But there’s the matter of the swing path. Home runs tend to be hit about optimally — the bat is going up, meeting the ball going down, and the ball comes off following the swing path, at some angle above horizontal. In order for a groundball to be, it doesn’t hit the bat in the same place. It seems to follow, then, the ball will come off a little weaker, not being met square. But! It’s not like I’m the physicist in this thread. So I could be completely out of my league.
There is literally no reason to think that with all the factors in play that a groundball couldn’t be the result of a ball coming off the bat at 100+ MPH.
Bad contact is still bad contact, but the resulting speed off the bat has nothing to do with squaring it up.
Jeff…I read your article quickly and missed the part about the Rizzo bunt. On that one I would agree with you. In fact, I would have expected the ball to come off the bat at less than the threshold speed for Trackman to measure it. But I still remain (mostly) unconvinced about the Hanley videos you posted.
Dr. Nathan, this comment concerns me a little. What use is Trackman if a bunt is below its minimum detectable velocity? Do they not have sophisticated clutter processing?
I agree with Alan. I absolutely believe the balls exit his bat at 100 MPH +. No reason not to. This is why we don’t use the eyeball test. Jeff sees relatively weak contact. I watch the same clip and to me he’s drilling them. But he’s drilling them right into the ground and they lose their momentum.
I agree with his conclusion – exit velocity is only a small part of the equation – exit velocity + trajectory is much more important.. 103 and into the ground isn’t as good as 98 with loft that finds the seats in LF…
Pedantry time… Velocity includes trajectory.
Back to the content-related discussion. “103 and into the ground isn’t as good as 98 with loft that finds the seats in LF…” Absolutely agree, but what would we use the extra information for anyway? We know that the ball went over the fence or on a couple hops to the SS. We even know which were classified as LD vs. GB vs. FB. In that sense batted-ball speed tells us nothing new.
It would seem that batted-ball speed would be useful if it were predictive. If it’s giving us a stronger prediction than what LD rate gives currently, then great. The Venn diagram of {what batted-ball speed tells us} vs. {what LD rate tells us} probably overlaps by a lot. We obviously don’t know this yet. But it’s entirely possible that batted-ball speed – even when hitting the ball directly into the ground – tells us more than LD rate, or maybe tells the same thing at an earlier point in time.
Jeff, the best city with the best velocities is never going to be topped. It’s Chinatown, braj.
This might be off topic, but can Hanley Ramirez please get a helmet that fits so that we don’t have to wait every single time he swings for him to tuck his pretty hair back under the helmet? You’d think with the $700M or whatever the actual figure is that they’re paying him he could afford to not look like a joke every time he whiffs.
I’m sorry, but this analysis seems to fall woefully short.
Story time: I was looking to invest heavily in Ramirez going into the fantasy season. I liked the idea of him not having to deal with infield bumps and bruises, his bulked up body, and the like. But most importantly, I liked the look of his spring at bats. He looked like what I’ve come to expect from him when he’s on.
But there was a final piece that made all the difference: I was able to attend a spring training game! I saw two Ramirez at bats. The results: 0 for 2. On paper, that was bad. But scouting is a wonderful thing…he CRUSHED both those outs. Not just in terms of ball flight…the ball jumped off his bat, and the sound was different, the crack that comes from someone who is really transferring his swing energy efficiently. One ball was mashed to the opposite field, but right to the RF, and the other was a VERY hard GB between 1B and 2B. I came away noting that if his OUTS were so good, so hard hit, that I could expect to see more of the good stuff come the regular season.
Small sample sizes and all, his April has validated what I saw. And then I saw your post, and the read the early portion of it. Surely by my observations Ramirez is going to rank highly on here, yes? Oh, he’s at the very top, and by a wide margin? No wonder his helmet keeps falling off. The batted ball data confirms both the results AND the scouting. I’ve watch most of Ramirez’s at bats this year…I’ve seen GB an liner outs that are hit harder than most peoples’. He’s simply squaring up the ball, even when the trajectory is bad.
And so comes the question: why did you not accept this result? More importantly, how did your investigations partly invalidate the results in your view? If there is distortion, contact included as “100+ mph” that shouldn’t be, don’t you think that EVERY hitter would see similar inclusions? Perhaps the readings are inflating the percentage of balls that are counted as hit that hard, but it shouldn’t influence the rankings themselves, as everyone should see similar levels of inflation, with exceptions made to very unique hitters.
And as an above fellow wrote, where are you not seeing high velocities off the bat with the above clips? Look how often the fielders have to move back or hold their ground on the balls, rather than attack them. Look how quickly the balls get there. Look how high some of them bounce.
It’s funny how analysis works. Had you asked me to guess which hitter has the highest batted ball velocities so far this season, I would have guessed Ramirez with a relative amount of confidence, and the data completely supports my expectations. You seem to be committing the Derek Jeter fallacy: if the system says Jeter isn’t a good fielder, the system itself must be broken. There’s NO WAY my view of Jeter could be improperly colored, no sir…
P.S. I bought into UZR and DRS years ago when I looked at the OF leaderboard, saw Suzuki and other usual suspects at the top, the expected poor fielder low down…and Manny Ramirez dead last, but a hefty margin. I saw some unexpected results, but the field as a whole confirmed my general expectations, and I was sold. Just because there were some outliers to my expectations didn’t mean I should trash it.
P.P.S. If you’re troubled by the Rizzo bunt, take my above point to heart. Don’t trust any one single reading, given the unknown chance of error that we currently have, but trust the data as a whole. If you hit the ball harder in general, your entire data set should reveal that reality.
P.P.P.S. Go back and look at those Ramirez GBs again (I just did). Instead of watching the ball…watch his swing. Then on the next take watch the swing into the ball and then the balls initial path. That is NOT what most MLB hitters look like when hitting GBs, even the power guys. Ramirez is swinging hard, and even when he is a little off he is on target enough to get major energy transfer.
Come back to bed, honey.
Thank you for not in any way replying to my argument.
Thank you even more for implying some kind of bias. How does bias factor in my thought that the MEASURED DATA about Ramirez’s hard hit balls is accurate? Since when is hard hitting some kind of great thing? It’s a trait, not a direct measure of success. I simply have observed in multiple ways that it is accurate to assess Ramirez as having hard contact, and incorrect to question that assessment in the manner of this article.
As for bias: did you miss the part when I said I bought into UZR and DRS because it viewed Manny Ramirez as dead last? As in, I knew from objective means that Ramirez was a terrible fielder, and approved of defensive metrics that reflected that reality. As in, no rose colored glasses here.
So I guess because I have “red sox” in my username that I can’t sound off on anything Boston because that automatically makes be biased. Fine, be an ignorant little turd.
It’s too bad you didn’t bother to read my whole post (clearly you and about 10 other people didn’t, as NOT reading it is the only way to conclude that I was being biased, given the reasoned and objective points made). You might have learned something.
Not sure I agree with everything that you’re saying, but Ramirez really has been hitting the ball hard so far this season (at least in home games), and I find it believable that he’d be leading the league in exit velocity.
I’ve seen a bunt back to the pitcher characterized as a line drive at times. Maybe the so-called bunt by Rizzo wasn’t really a bunt. Have you seen the actual play to verify it? Of course, I’m not saying the speed is correct, just wondering what else they messed up in the data.
Yes, I watched it.
Let me add another comment about the Rizzo bunt. For the Doppler radar to “find the ball”, the speed along the ball-radar axis has to be high enough so that the Doppler frequency shift is above the “ground clutter”. The latter is the low-speed noise due to all the other stuff that moves in the field of view of the radar. I don’t know exactly where the measurement threshold is. But a bunt is likely coming off the bat at a speed of 20 mph or below and is likely below the threshold. I am surprised that anything was reported at all.
One of the things I learned about PITCHf/x is that there were a lot of manual logging errors in the early days. The measurements were correct but they were being attributed to the wrong event. A classic case occured on Barry Bonds’ record-breaking 756th home run, where the initial pitch speed was some ridiculously low number. As it turned out, the home run was preceded by a throw-back, where the pitcher was requesting a new ball. It was the throw-back that initially got recorded as the pitch.
If the Doppler can’t see the ball, does someone manually add a value? Or is it a null event? I would assume the latter is more appropriate in the case of a bunt, given the intent is to dramatically reduce the exit speed. I love your work, by the way!
Wait a minute. Is it because it’s covered in the clutter or is it because it picked up something else and aliased it back into the observable range? Do we have any kind of technical specifications for the radar itself?
I think its urgent that we include incoming pitch velocity data along with all batted ball velocity data, right off the…er…bat.
i don’t see how we can possibly glean accurate information from it otherwise.
Just curious: why do you think we can’t glean accurate information otherwise.
i’d be shocked if incoming pitch velocity didn’t have a very significant effect on ecit speed…but i guess it would be just as instructive to have that proven wrong.
either it seems to be something we really shouldn’t ignore.
on a related note, i think i’m actually more interested to find out the quality of contact allowed by pitchers rather than the quality of contact made by hitters. I think the stats we have tell us a pretty good story about hitters already.
Quantitatively, for a well-hit ball about 15% of the batted ball speed comes from the pitch speed, the other 85% from the bat speed. The latter is far more important. For example, a 10 mph change in pitch speed leads to about a 2 mph change in batted ball speed, all other things equal.
ah cool.
15% seems to be pretty significant though, rightm
Jeff, I’ve been trying to figure out where int he Gameday XML data you can find Hit FX/StatCast data this season. I’ve dug through several games and been unable to find it. Do you have a link to an example of one game from the Gameday XML that shows batted ball data?
The Statcast data available is only batted ball speed and distance traveled on select plays in a random JSON file under “Feed” in Gameday. He’s almost assuredly using Baseball Savant’s database that scraped said JSON files, unless there’s access to Statcast that Fangraphs isn’t telling us about.
Would you be able to provide a link? Can’t find it in the Gameday directory with all the XML files, etc.
Wouldn’t it be safe to assume that it would be balanced out by balls that were hit over 100 mph he didn’t get credit for due to errors and by others suffering from the same subtraction of 100 mph contact so that he likely still leads?
I think the issue is not whether the list of leaders is accurate. At least for me, the most important issue is how reliable are the batted ball speeds reported by StatCast. And not the accuracy for the average over many batted balls but the accuracy for any individual batted ball. Other than the Rizzo bunt, I have seen no evidence thus far that brings the reliability into question for any particular batted ball. Please note my careful choice of words: “no evidence thus far”.
Wow, thats just so disappointing. Like you say, its not like this is all that new, MLB has had hit f/x for several years. Of course, they have a monopoly on its data and until now we have not got any of the data to validate.
I see no evidence statcast is going to be more than a marketing tool for MLB. I doubt we get much out of it, except what they want to give us, and now we know we can’t even trust that
Posted as reply instead of stand alone, sorry Alan.
But replying to your comment now, its seems the burden of proof should be on MLB to show their data is accurate. One can’t see at night on the dark, and MLB has not turned the lights on. Just a flicker here and there
I’ve watched most of his at bats and he’s had more than a handful of ROCKET shots that turned into outs. I can’t recall a guy hitting the ball as hard as consistently as he does. Virtually every swing so far has been spot on the ball.
Teixeira ht a HR last week that Statcast had as 430 ft, yet ESPN which is supposedly using hit trackers methodology to estimate HR distance had it at 370 something ft. It was an upper deck shot, although down the line, so I was inclined to believe statcast, but one of the 2 is way off (if ESPN is wrong they are not using hit trackers methodology but just going by the distance the ball travelled till it hit the seats and not extrapolating the flight path which is a cheat)
The integrity of these stats is in question