Ichiro Suzuki, Hideki Matsui and BABIP Aging

Two players I have always rather enjoyed — Ichiro Suzuki and Hideki Matsui — are in the twilights of their careers. And not only that, they are both having relatively miserable seasons at the plate.

At present, Suzuki is at a career-low 77 wRC+ and .281 wOBA. Matsui — playing in a reserve/pinch-hitting role for the first time in his career — has a disastrous 18 wRC+ and .195 wOBA. But both of these players have unusually low BABIPs and it is hard to know for certain if there has been a change in true talent levels or if this is random variation coupled with only mild aging.

In his reaction piece to the Ichiro trade, Eric Seidman rightly observed that Ichiro’s numbers should improve as he transitions from what has become an offensive deadzone in Safeco Park to the lefty-friendly grounds of Yankee Stadium, but how much can we expect his BABIP to improve? And what can the Rays expect from a 38-year-old Matsui, who may be needed more than ever with Luke Scott back on the DL?

Recent history suggests that both players are having abnormally bad BABIP, and they should improve if given consistent playing time through the rest of the season. But whether teams want to — or should — take that risk is another matter.

Biases Confessed

It is important to that readers understand my bias: I am a fan of both Ichiro and Matsui. I have a soft spot for aging veterans as well. It is important these biases are made plain because I hope to convince the reader of the legitimacy of my findings without the suspicion of self-serving motives.

When I began this study, I did so under the honest desire to find truth, not to trick myself, not to trick readers, and not to manipulate data in any way. My findings have flaws, my buckets have holes, and my data has problems, but none of these are extraordinary, none of these problems — in my understanding (which is apt to change) — is beyond what any other normal study has. Moreover, I think the findings are simple and loose and intuitive, so I hope few people will bristle at my suggestions.

Data Selection and Survivor/Confirmation Bias

First of all, let’s look at aging. From 2000 through 2011, there were 59 players aged 36 or older who put together a full season at the plate (a “qualified” amount of PA, which in this case has a minimum of 507 PA). This will be our Old Guys Bucket.

One of the great difficulties of any aging study is picking the right group. Especially in the MLB, there is the rampant problem of survivor bias and confirmation bias. The survivor bias comes via the players available — only good players make it to old age, and only players who are good in their old age get to keep playing. Very few players get to Chipper Jones their way to retirement — most of them Vladimir Guerrero their way into oblivion. And so my buckets will invariably include players who have done well — but hopefully it will also include a share of bad last seasons. For instance, Edgar Martinez had a -0.2 WAR season in 2004, and at age 41, he’s thankfully in the bucket. This helps balance things out because we don’t want to look at just the tails of the data, the extraordinary players that no one could have predicted — but for the purposes of this research, survivor should hopefully not be an issue. We are looking within careers, not really comparing across careers.

Confirmation bias makes an appearance in the playing time issue. Let’s say Ichiro has just been crazy unlucky and his skills haven’t eroded at all. Well, too bad. The Yankees have enough depth to play Ichiro according to his strengths, which likely means reduced playing time. And because we know all players eventually decline in their ability, we fully expect at least part of Ichiro’s drop off to be aging related — if not all of it. So, when he struggles, it confirms our expectations.

When Giambi struggled in 2009, and his BABIP dropped to .228, it was easy to believe that the veteran slugger has simply lost his hitting skill. He was 38 and had a 97 wRC+. Since then, however, he has not posted a BABIP below .284 (he has a career .297 BABIP) and he has been a useful hitter from the bench for the Rockies. Similar glowing things could have said about 34-year-old Jim Thome in 2005, who’s BABIP dropped to .256 with a, 89 wRC+. (Thome has not had a BABIP, by the way, under .300 for the last four seasons. Wow.)

So both of these issues — survivor bias and confirmation bias — are swirling around and playing with the available data. In the ideal aging study, every single MLB player gets 1500 PA per year until their 50th birthday, but alas, we do not have such luxuries.

What we do have are these 59 players — guys who have played with modern medical science, who have played in roughly the same run environments (these players are 2000 to 2008 heavy) and who did not suffer the penalties due pinch-hitters and bench players.

Of theses 59 hitters, they had a combined 112 split seasons aged 36 or older. We have 45 seasons of 36-year-olds, 31 seasons of 37-year-olds and the final four ages have groups of 15, 12, 6 and 3, respectively.

This, as if it needs emphasis, is not a big group. Remember that fact (I say this to myself as much as to the reader) when looking at the findings.

In the Old Guys Bucket, the average and median drop in BABIP (season BABIP minus career BABIP) was only 11 points. At the same time, it appears that most players, when looking at the distribution of those bad BABIP seasons, had their BABIP fall at least a little:

NOTE: Not all 112 season names can fit on the horizontal axis, but rest assured, they are all there. Click to enlarge the image.

We can see the majority of 36-or-older players cannot replicate their youthful BABIPs. It is worth noting here that the end of the aughties and beginning of the 2010s was the beginning of a reduced-BABIP era, and all these players are moving towards that world. Still, as I said before, this group is pre-2009 heavy. I think most of what we are seeing here is a decrease in true talent levels — guys making weaker contact and legging out fewer infield singles.

But if we break these differences down by age, we see the heaviest BABIP drops coming in the latter ages:

Click to enlarge.

From age 36 through 39, the group averaged BABIPs 8 to 16 points worse than their career numbers, with 8 and 9 points making the most appearances. This surprised me the most, considering we have 2003 Rafael Palmeiro, 2001 Cal Ripken, 2008 Kevin Millar and 2006 Frank Thomas included in this set.

These are old dudes — at the ends of their careers — with BABIPs barely shifted from their normal career numbers. Here we must interject the final confirmation bias warning. These hitters would not reach 500 PA if they looked lost in Spring Training, if scouts were seeing red flags in batting practice, if they were truly and undeniably kaput. I think few hitters really reach that level outside of catastrophic injury, but if and when they do reach it, they are unlikely to cobble together 250 PA, much less 500.

So these are hitters who, presumably, passed a modicum of the eye-test (at least among team officials) and who had success (to a degree) the preceding year. But nonetheless, we are seeing hitters aged 36 to 39 with BABIPs quite close to their career numbers, and that is worth remembering.

Ichiro and Matsui

This brings us full circle to Ichiro Suzuki and Hideki Matsui. Ichiro’s BABIP is 69 points below his career BABIP, and Matsui’s is 128 points down. Neither player is likely to get full playing time (or even 500 PA, for Matsui) through the end of the season — but, if they hypothetically did, and if their BABIPs remained constant, they would rank No. 2 and No. 1 among the worst BABIP drops in the whole Old Guys Bucket.

Maybe Matsui is really finished? It’s hard to say from a statistical standpoint as his 103 PA have come largely as a pinch hitter and in sporadic stints as the DH. Both roles come with considerable offensive penalties. Add in the fact his season started in late May, and we have to wonder if his extended off-season hasn’t resulted in a possibly temporary loss of bat speed.

Scouts might be able to offer insight into whether or not he has lost bat speed, but his contact rate (78.1%) is easily at a career low. Let us say he is not finished, that his skills have depreciated, but that he’s not an 18 wRC+ true-talent hitter.

Let us then presume he can at least match the worst possible BABIP difference of -57 points. That would mean that Matsui, with a .240 BABIP, would have approximately a .260 wOBA (according to FI wOBA) — which is great for the best defensive shortstop in the history of the universe, but not so great for a DH.

If his BABIP managed to hit the average decline — minus 11 points — then he would still only muster a .289 wOBA. If he matched his career BABIP (not going to happen, in all likelihood), he’d be at .296 wOBA.

The core of this issue becomes clear, then: His plate discipline numbers are askew. Let’s say — with some regular playing time, as with before — he sorts out his approach and at least matched his 9.6% walk rate and 14.4% strikeout rate from 2011 (not great numbers, but not terrible either). Suddenly, his wOBA (with just a .286 BABIP) jumps to .316 — which would rank him the fourth-best hitter in an offense-starved Rays lineup.

So is Matsui’s usefulness expired? Perhaps — but if he can find his old plate discipline, even with a career-low home run rate, he could realistically provide the Rays with solid production. But there are a lot of if’s there.

Ichiro, on the other hand, needs to change little. One of the curiosities of his hitting profile over the last few years has been the rapid decline in infield singles. Though he continues to steal bases at a considerable pace and though his defense and range continue to rate well, he seems to be losing the skill for infield singles. Here is a look at his hit types per plate appearance:

However, from 2002 through 2009, Ichiro averaged only 5.8% infield hits. From 2010 through 2012, that number has decreased to 5.1% — that’s a pretty small change overall. Meanwhile, his non-infield hits have gone from 19.1% of his PAs to 17.1%. That’s a more troubling decline.

Still, his doubles and homers are in line with his career rates, and his triples have ticked up in recent years. And he has been successful (in 2009 — 113 wRC+) with a lower non-infield singles rate. Moreover, his stolen base rate over the past few years is almost unchanged from his career numbers.

Let us again play with the hypothetical and say Ichiro’s rest-of-season BABIP climbs to the -11 point difference we found prevalent in the Old Guys Bucket. Ichiro’s wOBA jumps from .281 to .329 as his BABIP goes from .279 to .337 — that’s a huge BABIP jump, but it’s also a testament to how far it has fallen this season.

One other difficulty with the preceding findings is that all those career BABIPs we were looking at already have incurred the low-BABIP final seasons — so they were downward biased, which means the player’s difference between career and seasonal BABIP looks less dramatic.

Since these guys have played into at least their late-30s, most of them have thousands of PA, so it shouldn’t make much of a difference, but, hey, let’s say the penalty should actually be 25 points of BABIP. That puts Ichiro at a .320 wOBA — and with his defense, which continues to be Gold Glovish — he makes a considerable outfield asset.

Conclusion

The biggest information we can draw from this is that players given ~500 PA from 2000-2011 did not see their BABIP drop more than 60 points (57 points to be precise) from the career numbers and averaged a drop of only 11 points. Whether this applies to any modern player probably depends on the situation, but it should make intuitive sense — players’ skills decline with age, but with steady playing time, they should be able to at least best pitchers, who sport BABIPs in the .220s. Give these veterans less than consistent playing time, and they may not best the pitcher BABIP.

For the specific players of Hideki Matsui and Ichiro Suzuki, the results are a little more divergent. Ichiro’s decline appears all too sudden, especially given his speed-rich skillset, and if the Yankees choose to give him steady playing time in New York — which would unfortunately come at the expense of some of their other talented outfielder hitters — it could be a worthwhile gamble (especially considering Ichiro’s change in home stadiums).

Matsui, meanwhile, has plate discipline numbers that suggest greater problems — including a low contact rate. If he continues to receive only partial playing time and cannot improve his plate discipline numbers while in that capacity (or any capacity, really), then the Rays will have little else to do but release him. If he gets steady playing time, we can expect his BABIP to increase, but it will do his output little good if he cannot also improve his walk rate, strikeout rate and/or home run rate.

For those who want to toy around with the same data I used — presumably for the purposes of mercilessly mocking me — here are the two leaderboards I used to gather the date:

Old Guys Bucket: Seasons | Career




Print This Post



Bradley writes for FanGraphs and The Hardball Times. Follow him on Twitter @BradleyWoodrum.


32 Responses to “Ichiro Suzuki, Hideki Matsui and BABIP Aging”

You can follow any responses to this entry through the RSS 2.0 feed.
  1. tvotr says:

    Ichiro Suzuki, Hedeki Matsui, and BABIP Asians

    Vote -1 Vote +1

  2. Sean says:

    This article is racist.

    Vote -1 Vote +1

  3. Justin says:

    Love the graphs. Thanks for the article.

    Vote -1 Vote +1

  4. DSC says:

    Gee, since when did the Hall of Fame become oblivion?

    Vote -1 Vote +1

  5. Mike says:

    Would be interesting to see how BABIP and Contact rates correlate in aging players. Even if their BABIPs don’t drop too much, if their contact rates drop enough, that could still make them unproductive. For example, Matsui’s contact rate is 7.5% lower than his career average, while Ichiro’s is actually a bit higher than his career average. So I would expect Matsui to continue to decline, while Ichiro could see his numbers go up if his BABIP normalizes.

    Vote -1 Vote +1

  6. Basil Ganglia says:

    My observation regarding Ichiro’s infield hit rate is that he at least one – if not two – steps slower getting to first base from the batter’s box. Ground balls that were infield hits in his prime are now routine outs. Previously you could see infielders hurrying as fast as they could to complete the play on many routine grounders. Now plays to first on Ichiro are handled pretty much like any other “fast” runner to first.

    And, while Ichiro’s strikeout rate is down this year, it has been trending higher, and last year was the highest of his career by one full percent.

    Vote -1 Vote +1

  7. Jesse says:

    What program do you use to create your graphs?

    Vote -1 Vote +1

  8. Phantom Stranger says:

    I think there is a widespread belief that BABIP will not normalize to career norms for a hitter, if the hitter has truly fallen off a cliff and does not have the same talent level of prior years. There might be a point where no amount of adjustment or preparation can compensate for the total decline in physical skills, leading to permanently impaired BABIPs. As you said, it’s very tough to suss out of the data because most teams let players go at that point.

    Vote -1 Vote +1

    • rory says:

      This article could only have been written by someone who’s looked at the numbers and not watched the games.

      Ichiro used to beat out routine grounders for infield hits all the time. In a sense, there was no such thing as a “routine grounder” for Ichiro, but he’s slowed down.

      Look at his ISO .091 .vs a .095 career mark.

      Look at his K%. 9.4 .vs a 9.5 career mark.

      His doubles and home runs are about on pace for his career mark too.

      What’s missing are those infield singles… That’s why his BABIP is down. He’s the same player who can no longer make it down the line as fast as he used to, and unless the air in New York makes him run faster, there’s no reason to expect a significant rebound in New York.

      Sometimes people get lost in the numbers and forget that there’re real people on the other side.

      Vote -1 Vote +1

      • Matthias says:

        @rory: I am generally skeptical of a claim based on the eye test. I think both eyes and stats are needed, and here are some stats that may or may not support your conclusion.

        2012: 14 infield hits per 178 ground balls (7.9%)
        2011: 36 of 361 (10%)
        2010: 53 of 335 (15.8%)
        Career (to 2009): 338 of 2666 (12.7%)

        His drop from the 12.7% career rate to his 2011 and 2012 rates is statistically significant . However, over those same years his rate of reaching on errors per ground ball has gone from 3.3% (88/2666) to 2.4% (13/539), which is NOT statistically significant.
        (despite that those years were selected post-analysis and subject to confirmation bias)

        I would venture to say that if his BABIP were .340 right now, and he was hitting .315 ish, that you wouldn’t think twice about his speed to first. It turns out his loss of infield hits has cost him between 8-9 hits on the season. That would bring his BABIP up to .303 and his average up to .284. That still falls well short of his .325ish pre-2011 average and .350ish BABIP. Where did the other 50 BABIP points go?

        I would guess that Bradley is arguing that–without being able to explain where the BABIP went–we should assume some regression back toward similar players that were allowed to continue their careers late into their 30s.

        Vote -1 Vote +1

  9. Sandy Kazmir says:

    You’re asking the wrong question. The correct question, in my opinion, is, “When is it time to hang ‘em up?” To that end I looked at the final season with at least 50 PAs for all batters 35 and older from 1980 – 2011. This yielded 407 players. It can be useful to check out the workbook:
    https://docs.google.com/spreadsheet/ccc?key=0AhdYS83t3IB7dElaRHBxeWUydC1OZ1NvOF9nc3VVVXc#gid=0

    Final season BABIP for the entire cohort was a weighted average of .271 while the weighed career BABIP was .296 leading to a drop of -.025 or more than double what you found. Additionally, the range runs from -.185 for Milt Thompson to Darryl Strawberry’s .151. Other players that had larger last season BABIP differential include John Vander Wal, John Jaha, Harold Baines, Jerry Mumphrey, and Rondell White.

    It is not uncommon for batters to post BABIP well below career norms in their final season. Perhaps we should be using BABIP to isolate when a player has hit the wall rather than to make excuses for why they are not performing at career levels.

    Vote -1 Vote +1

    • OldStyle says:

      What you described has a heavy bias towards limited poor performance and possibly isolated at bats (not to say that it doesn’t have some merit). I do not have the numbers off the cuff, but I believe a pinch hitter performs significantly worse and an end of career player/low AB would be more likely to see pinch hit at bats (not sure if it is babip though). 2nd, a baseball player isn’t going to be forced into retirement if he is performing well (essentially ruling out the lucky players). I think it would be far more interesting (separate topic) to look at rookie seasons and see babip change without quite the same bias (Trout was given another chance after his 0.247 babip last season).

      Vote -1 Vote +1

    • “When is it time to hang ‘em up?” is a good an valid question — and it’s perhaps almost the same question except from just the player’s perspective.

      However, it’s also much more broad — as is your sample. Not only have you included a lot of run environments in your data (which, hey, that’s not too bad), but by limiting your samples to only 50 PA, you are automatically incurring — as OldStyle noted — PHing and part-time penalties. And even then, with all of that in play, a 25-point penalty is STILL much less than what I would have guessed off-hand — and it’s much smaller than what Ichiro is at.

      Vote -1 Vote +1

      • Sandy Kazmir says:

        I’m less concerned with run environments as I’m comparing to the players career rates. Isn’t that the exact role that Matsui is being used in? Standard deviation of .044 means that Matsui is just outside of two standard deviations from the mean (-0.114, 0.063). The beauty of using a weighted average is that those that were used in a traditional role, or were performing, most likely end up with more plate appearances. This survival bias, as you stated, means that it’s likely that our mean is a tad low for a player that is actually finished.

        For what it’s worth, a simple average yielded a difference of -0.030. I maintain that BABIP may be useful in identifying players that have lost the edge necessary to play the game at the highest level.

        Vote -1 Vote +1

  10. Doug B says:

    Ichiro’s career numbers:

    Home
    .320/.366/.417

    Road
    .324/.366/.419

    Vote -1 Vote +1

    • For some reason, 2012 Safeco Park has killed hitting. And beyond that, Yankee Stadium is undeniably lefty-friendly.

      Vote -1 Vote +1

      • Ralph Malph says:

        Unlike the rest of the country, Seattle has had an unusually cool summer. And with a change in weather has surely come a change in wind patterns. That surely (along with too much Miguel Olivo) accounts for 2012 safeco.

        Vote -1 Vote +1

      • Doug B says:

        Yankess Stadium is not lefty friendly for old guys who beat the ball into the dirt and can no longer run as fast as they used to. It’s still 90 feet to first base.

        Vote -1 Vote +1

  11. Tom says:

    The other thing that might be worth looking at is batted ball profile (LD/GB/FB) and xBABIP to see if some of the aging trends have batted ball profile trends mixed in as well.

    Looks like Ichiro’s LD rate is significantly up (which would normally help BABIP) , his groundball rate is down (which would hurt BABIP) and FB rate is about where it has been. Of course that is potentially noise but it suggests that the BABIP decline is not due to a change in batted ball profile.

    Vote -1 Vote +1

    • Matthias says:

      Coincidentally I wrote a little piece about this very thing the other day. If Ichiro were able to maintain his career averages on line drives, groundballs and outfield flyballs with this season’s batted ball profile, he would be hitting 0.315.

      Whether it’s bad luck or bad skills I cannot say, but I can say that his batted ball profile isn’t the thing hurting him.

      http://sodomojo.com/2012/07/06/ichiro-and-his-approach/

      Vote -1 Vote +1

  12. Jeff Mathis does Steroids. says:

    posted on another thread, edited a bit…

    Anyone remember when a certain writer wrote about how well Ichiro was going to do this season? http://www.fangraphs.com/blogs/index.php/ichiro-suzuki-will-rebound-a-final-postulation/

    Carson Cistulli: Ichiro will regress towards his career means! BABIP lol!

    Fangraphs community: 38 year old will not regress toward career means.

    Looks like Carson is losing that round. Now all we need is another Fangraphs writer to write a trade deadline post saying “Chone Figgins could be a valuable piece for a team! He is going to start playing towards his career means any minute now! BABIPlol!!!!”

    Vote -1 Vote +1

  13. rory says:

    Why would Ichero’s hitting improve if he moves to another stadium?

    Is it not still 90 ft to first base in New York?

    Ichiro has been declining because he’s not as fast as he was. He used to beat out infield singles with his speed out of the box, but at 38 he’s starting to slow down.

    That’s why his BABIP is down. Sometimes you have to look past the numbers and look at the skills of the player.

    Vote -1 Vote +1

  14. BlackOps says:

    Thanks for the article and research, I enjoyed it.

    Vote -1 Vote +1

  15. Hurtlockertwo says:

    I find it interesting that Ichiro is still considered by the Yankees as a speed guy and marked inprovement in the field. Even though his hitting is fading, he still has considerable baseball skill.

    Vote -1 Vote +1

  16. Matthias says:

    Bradley,

    I’m confused as to your calculation of infield hit percentage. Fangraphs seems to define it as IFH/GB, but those aren’t the numbers you used. What was your denominator?

    Vote -1 Vote +1

      • Matthias says:

        Oh, that makes sense. Any reason you used that particular stat, and not IFH/GB or IFH/IFGB?

        (I feel like IFGB could estimated by simply subtracting the groundball hits found in a players splits from his total groundballs, and adding back in the infield hits)

        IFGB might isolate the speed factor a little betterer. Maybe.

        Vote -1 Vote +1

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>