Bleed Cubbie Blue: An SB Nation Community

Navigation: Jump to content areas:

Around SBN: DTN Interviews SI's Austin Murphy Bar-right-arrows

Projecting Kosuke Fukudome's Stats to MLB

Hitsugaya140x140_tiny by Ayralin on

The great debate of Japanese stars making the journey from Nippon Professional Baseball to MLB is just how well their skills and abilities will translate in a new league. As with any argument, each side has their successes (Ichiro) and failures (Kazuo Matsui) to point to. For the rest of us, there are statistics. (Warning: Numbers are to follow. Save yourself while you can.)

As with any translation, nothing is exact. And no matter how many players make the move and no matter how many at-bats they accumulate, you'll never be able to forecast what someone else will do from their numbers. Much as I admire Rob Neyer, I don't have his flair for finding the right way to explain how the numbers work, and you're probably tired of "Yet Another Kosuke Fukudome Diary" as it is. So, without further ado, I give you this Excel spreadsheet.

First off, no pitchers were included. This is strictly relating to Fukudome, meaning hitters. Second, I didn't include every Japanese hitter to see time in the Majors, nor did I limit it to only those with significant time in both leagues. I simply chose the names most likely to be recognized with at least one full season of recent playing time (apologies to Tsuyoshi Shinjo.) Statistics were accumulated from ESPN as well as Japanese Baseball and The Baseball Cube.

This is a FanPost and does not necessarily reflect the views of SB Nation, Bleed Cubbie Blue, or Al Yellon, editor-in-chief. FanPost opinions are, however, valued expressions of opinion by passionate and knowledgeable baseball fans.

0 recs | Comment 98 comments

Story-email Email Printer Print

Comments

Display:

First of all...
...very nice work on this; it looks pretty nifty.

That said...


  • When projecting future performance, it's best to use three years of data, generally as a weighted average (5/4/3 is a pretty common weighting for such things).
  • No adjustments are made for league or home park; all of these numbers should have been "normalized" to a more specific offensive context before deriving translations.
  • Seven players is a very small sample size to work with; you can derive a larger sample size by looking at major and minor league U.S. players who went to Japan as well.

If I can find some more detailed MLEs (Major League Equivelencies) for the NPB before the entire midwest is covered in ice, I'll post them here.

FREE CARMEN PIGNATIELLO!

by cwyers on Dec 10, 2007 7:32 PM CST   0 recs

Cool!
If Fukudome puts up a .800-.850 OPS with above-average defense in RF, he will be worth $10M a year to his team.  There is a shortage of lefthanded-hitting players with any power.  
Free the upper deck!

by zambranofan on Dec 10, 2007 7:46 PM CST   0 recs

I think you've got it.
It appears that for most Japanese players, SLG drops more than BA or OBA. This makes some sense just thinking about the differences in size between Japanese men and Western men (and yes, I am generalizing here -- I know there are wide variations), and the differences in the sizes of the ballparks AND the differences in the style of play.

Given the numbers, in particular, that have been put up by Ichiro and Hideki Matsui, who have probably the most AB in the USA by any Japanese position players, I think your projection for Fukudome is probably pretty close to what we can expect.

"That's my opinion and if you don't like it, well, I have others." ~ Groucho Marx

by Al on Dec 10, 2007 7:52 PM CST   0 recs

I disagree...
Given the numbers for Ichiro and Matsui (both Matsuis, actually), you should expect a 40 point dropoff in career OBP and a 130 point dropoff in career OPS.  If you just look at their last few years in Japan, the dropoff is much more dramatic.  If you apply that simple adjustment to Fukudome's stats, you get a .290/.360-.370/.440 statline.

What this analysis has done is taken advantage of the fact that some of the lesser-known transfers seem to not have had the same dropoff as the "big three."  I tend to agree with the Ichiro/Matsui comparison (which isn't favorable for Fukudome) and not this analysis, but honestly either approach is justifiable.

I'd guess he'll be somewhere between .790 and .850 in OPS.  Which end of that OPS spectrum he falls (and whether his OBP is .360 or .380) will make a big difference in his value, obviously.

by SouthernCub on Dec 11, 2007 6:59 AM CST to parent up   0 recs

Have you taken into account
that the Japanese players, in their first year here, have an average OPS drop of 100 pts when compared to their "home" stats.
Wait 'til next year. And the next. And the Next. And the next after that too.

by TheEman on Dec 10, 2007 7:58 PM CST   0 recs

No special considerations...
...are made in the spreadsheet, only a straight compilation of stats between the two professional leagues calculated out to an average which is then applied to Fukudome's career stat line.

As cwyers mentioned, there are a lot of secondary effects I could have taken into account (park factors, considerations for National/American league splits for the players' MLB stats, weighting applied to particular years of performance), but I was only aiming for a basic translation.

The number of players I used is small, and about 80% of the MLB statistics were accumulated in the American League, but I don't think the additional factors would skew the result much more than 10-15% except when the most exteme factors were applied.

by Ayralin on Dec 10, 2007 9:43 PM CST to parent up   0 recs

Awesome work!
I asked something like this the other day, but I was trying to relate contract info.  This is great, thanks.

Dan

Evey Hammond: Vi Veri Veniversum Vivus Vici. V: By the power of truth, I, while living, have conquered the universe.

by dtpollitt on Dec 10, 2007 8:09 PM CST   0 recs

You need MS Excel
To open the spreadsheet. OpenOffice would probably work as well.

by Ayralin on Dec 10, 2007 11:24 PM CST to parent up   0 recs

If you don't have Excel...
... you can download a free viewer, which allows you to view Excel files, here.
"That's my opinion and if you don't like it, well, I have others." ~ Groucho Marx

by Al on Dec 11, 2007 3:30 AM CST to parent up   0 recs

Hmmm...
... still didn't work after a reboot.  And I have Excel (I don't know what I would do in my life without Excel, quite frankly!).

Can you shoot me an e-mail with it?  cpm023[at]comcast[dot]net.

Thanks!

This is who I think the Cubs #1 should be...

by initram on Dec 11, 2007 10:03 AM CST to parent up   0 recs

Powre never seems to make the transition
I think they were discussing this on Chicago Tribune Live tonight.  Tad Iguchi hit a bunch of homers when playing in Japen. No power here.  However, if he keeps doubles power he is well worth it to the Cubs.  We'll have plenty of guys hitting the ball out of the park, we need someone on base ahead of them.

by Nibbles on Dec 10, 2007 10:38 PM CST   0 recs

Check this out...
http://www.anothercubsblog.net/2007/11/02/kosuke-fokudome/

I'm having a very hard time how you're projection system is projecting such a small decline in his AVG and OBP.  

by Maddog on Dec 10, 2007 10:43 PM CST   0 recs

Ours projections put weight on different factors
The projection system on that blog takes the changes in each player's statistics and averages them out. What I did was total sets of statistics and average those. In essence, they weighted each player equally, while I weighted each at-bat equally.

Their projection is a lot less clinical (and in all honesty, probably more accurate for it). The blog posits Fukudome and Matsui as similar players (which they are), while my spreadsheet calculates every at-bat - whether it's from Matsui, Ichiro, or Taguchi - equally. They're not, of course, except as a direct interpretation of how stats translate from one league to the other, which is all my spreadsheet takes into account.

by Ayralin on Dec 10, 2007 11:21 PM CST to parent up   0 recs

In this case...
...the underlying cause is the same thing; the projection expects Fukudome to continue to hit for a good average.

The question is, how accurate are either of these projections? And which of them is better modeling the underlying reality?

I'm going to go ahead and use a data set that's much less recent than the data set here, but has the benefit of being mcuh more detailed. [Seamheads is supposed to be coming out with a much more detailed set of analysis on recent seasons soon, and has a little bit of background on how these things work.]

Let's go ahead and ignore park effects here (even though the Chunici Dragons seem to play in a pitchers' park, which could further distort the data; this is all back-of-the-envelope stuff.]

Instead of simply weighting AVG, OBP, and SLG, Albright weights every main hitting event:

Hits .932
Doubles .980
Triples 1.577
Homers .604
Walks .805

What we see here is that batting average and (to a lesser extent) walk rate should hold up pretty well. If we weight those events like that, his MLE line from last year was .274/.371/.441. Which looks pretty close to what you're predicting, actually.

But two caveats here. One, looking at OPS is supposed to be a quick-and-dirty way of modeling offense; if we're going to go to all the trouble of building MLEs in the first place, we can go to at least a slightly more sophisticated tool, and weight OBP accordingly (weight every point of OBP at 1.8 per point of SLG, or 1.8*OBP+SLG). Fukudome gives you 1.102 GPA, versus (say) Murton's 2007 1.069 GPA. That very healthy difference in OBP is something to consider.

Second, predictions using only one year of playing time are not as good as predictions factoring in three years of playing time. So let's take a weighted average of his last three seasons, 5/4/3. Once again, we work with the counting stats (hits, 2b, 3b, HR and BB) and then figure out AVG, OBP and SLG; it's a bit more work than just averaging the AVG, OBP and SLG but more sound mathematically.

What do we get? A (very rough) projection of .306/.380/.515 for Fukudome next season. That's an .895 OPS, or 1.199 WPA. Very nice numbers. We probaly should whack off a bit for an aging curve, but we should also probably credit him a bit for going to the NL, a better offensive environment than MLB as a whole.

FREE CARMEN PIGNATIELLO!

by cwyers on Dec 11, 2007 12:43 AM CST to parent up   0 recs

Before I get pilloried here...
...yes, I know there are flaws in the model. I think I pointed a few of them out myself. I do think the slugging percentage is more than a little optimistic, if you were wondering.

But I will point out that I'm only projecting 13 or 14 home runs, which is not too far out of line with what you'd expect -- it passes the smell test. What's powering that slugging projection is his high batting average and his high doubles rate.

FREE CARMEN PIGNATIELLO!

by cwyers on Dec 11, 2007 1:31 AM CST to parent up   0 recs

Weighting recent seasons
Placing weight on recent performance makes a lot of sense, but you can also end up with one season skewing your results. Sometimes career years indicate a new ceiling for a player (Derrek Lee, 2005), sometimes they are pure aberrations (Adrian Beltre, 2004), but most often it's just a very good player getting a little luckier.

I tried to emphasize counting stats as opposed to averages, just as you do, and the projections come out expectedly similar. I think the discrepancy in SLG can be attributed to Fukudome's career year in 2006, in which his HR rate was very good (16 AB/HR, just shy of his career-best 15.2 in 2004) and his doubles rate spiked to a career-high 10.6 AB/2B.

by Ayralin on Dec 11, 2007 7:43 AM CST to parent up   0 recs

On a second reading of your post...
...it occurs to me: the main difference between your projection and this projection is that you pull a "can't possibly be as good as Hideki Matsui" adjustment out of your ass.

Would've saved me a lot of math if I'd picked up on the importance of that to your arguement.

FREE CARMEN PIGNATIELLO!

by cwyers on Dec 11, 2007 1:52 AM CST to parent up   0 recs

But it's true.
Hideki Matsui was doing things at a younger age than Fukudome was and he as doing them better.  He has been twice the player Fukudome has been in Japan.  Accepting a projection that projects him to be a better hitter in the United States is not rational.  It would be like accepting a projection that showed David Eckstein to be a better shortstop over the next 5 years than Derek Jeter.  Eckstein has never been as good as Jeter at any point in his career and it's an almost impossibility that he will be better than him over the next 5 years.  The same is true with Fukudome.  

I like projections systems.  I think they do a good job, if used properly, of saving teams from making stupid mistakes, but the one thing they lack is common sense.  And it's frustrating too.  

What this spreadsheet has done, by weighting by at-bats, is essentially weight the career of Ichiro more than anyone else, who hits for an unbelievable average and gets on base at a high rate as a result of that high average.  Even Ichiro was a significantly better ballplayer in Japan at a younger age than Fukudome.  Not to mention, the skill sets of the two couldn't be more different.  So we've weighted Ichiro more than anyone else (who happens to be the most successful Japanese player in MLB history) and he has very few points of similarities with Fukudome.

Furthermore, it projects Fukudome to be just as productive offensively as Hideki Matsui who just murdered baseballs in Japan starting at a young age while Fukudome failed to even have one season as good as multiple years that Matsui had.  It takes a player who had a career OPS some 60 points lower and projects him to be the same player as him at the MLB level.  

There aren't enough Japanese offensive players in baseball to do a thorough analysis that creates a quality projection system.  The only way to do so is to calculate the average decline in various rate stats of all the players who have played here and even in doing that, we end up with a line that Fukudome is simply not likely going to reach.  Doing so shows him to be about as equal as Matsui, which is just not going to happen.  Matsui was a better player at 22 than Fukudome was at any point in his career so far.  

You can't weight the system by at-bats, which automatically weights Ichiro (the best Japanese player of all-time) the highest with Matsui right  after him (the 2nd best Japanese player of all-time).  

by Maddog on Dec 11, 2007 9:24 AM CST to parent up   0 recs

Does it make any more sense...
...to give Ichiro and Hideki Matusi equal weight to players with only one or two seasons of MLB experience (Iwamura/Johjima)?

In his first MLB season, Matsui posted averages that he has since surpassed in every category in every season since (with the sole exception of BA, which was .285 in 2007, .002 lower than 2003). It is a line similar to Matsui's 2003 that Fukudome is expected to most closely produce, in my opinion.

by Ayralin on Dec 11, 2007 9:44 AM CST to parent up   0 recs

One thing you're forgetting...
...and cwyers, too, is that Fukudome is 30 right now. Kaz Matsui was a year younger than that and Hideki was two and Ichiro was three years younger.

If you think a baseball player's general decline begins around 28-30, then it doesn't make sense to project Fukudome  like he was coming over as a 28-year-old.

I really think this is the biggest mistake when thinking Fukudome using Ichiro and Matsui et. al as comparisons. I think you need to take any under-31 projections off of Matsui, Ichiro, etc.

by tyger1147 on Dec 11, 2007 9:59 AM CST to parent up   0 recs

Peak years and decline
True enough that Fukudome is older than the other players who made the transition. However, decline is a beast which can only be generalized. Every player is different in this regard, no matter how similar their stats are. For every Roberto Alomar, who fell off a cliff after age 30, there's a Chipper Jones, maintaining a high level of performance into his mid-30s.

That said, I don't expect Fukudome to get much better than he is now. The questions are how well his current performance will translate, how long he'll be able to stay at that level, and how fast the decline will be afterwards. It's hard enough agreeing on the first question without throwing the other two in the mix.

by Ayralin on Dec 11, 2007 10:21 AM CST to parent up   0 recs

So...
...you just ignore the statistics relating to decline because there are exceptions? No real projection does that. It's pointless to project if you won't even consider this. It almost always matters. Is it possible that it won't? Yes. But I didn't know projections dealt with, "Well this unlikely thing is possible, so that's what I'm considering."

You really should do a projection just based on age-31 seasons and over. It might vary wildly, but it would at least be interesting instead of using a 27-year-old Ichiro to project a 31-year-old Fukukdome.

by tyger1147 on Dec 11, 2007 10:26 AM CST to parent up   0 recs

Decline
No, I omit the factor of decline because my projection is intended for a single year, and impacting factors are limited for the purpose of maximizing the available sample size. The statistics are already limited enough without restricting the pool to just the age 30 vs age 31 seasons of Japanese players. And by your own reasoning, it wouldn't make sense to use the later years, because then we'd be using 34-year-old Ichiro to project 31-year-old Fukudome. I try to balance the fact that the player's pre-31 stats are included by also including the post-31 stats, which is hardly an ideal mitigation, but I'm open to suggestions that don't limit this to a comparison of 1500-at-bat slices of less than a half-dozen careers.

I can tell you right now, using the post-31 seasons of Japanese players would certainly skew Fukudome's projection downwards. After all, Ichiro suffered a 69-point drop in BA in his age-31 season, and Hideki Matsui posted a 50-point drop in OPS (or was injured, depending on where you place the age cutoff date). However, I've already said that I don't expect Fukudome to get any better than he is now. I also admit there are many factors that most definitely will affect his stats that I have not included, and the projection you see in my spreadsheet is more of a projected peak than an indication of how I think he will perform over the life of whatever contract he signs.

by Ayralin on Dec 11, 2007 11:40 AM CST to parent up   0 recs

I didn't use Ichiro or Matsui...
...as baselines, other than what data from them crept into the MLEs I used.

Maddog is right when he says that there's not enough Japanese players coming to America to create a sizeable enough data set. The answer to that, though, is that we don't NEED more of them. The study I linked to used American players who went to Japan, and saw how their stats increased, as well as Japanese players taht went to the US.

And I didn't forget the aging affect, because I mentioned it. A generic aging curve for those projections would be something like .006 less. He's old, but not THAT old.

FREE CARMEN PIGNATIELLO!

by cwyers on Dec 11, 2007 12:01 PM CST to parent up   0 recs

Yes it does.
By weighting playing time as you have, you've immediately assumed (as a result of doing it this way) that Fukudome is going to be among the better players to make the jump from Japan to the US.  

By taking the average decline of the players who have already made the jump, you're not making that assumption.  The only assumption you're making is that the game is more difficult here than it is in Japan, which is undoubtedly true.  

He just might beat the odds and be one of the better ones to move here, but you can't assume that.  All the players who played here from japan were good.  None were as good as Ichiro or Hideki Matsui.  It's no coincidence that they have turned out to be the 2 best players from Japan among the hitters.  They were the best 2 in Japan of the bunch who came here.  They each, however, had huge declines in their numbers because MLB is tougher than Japan.  

The only accurate assumption at this point is that MLB is tougher than playing in Japan.  There simply is not enough data to make any other assumptions.  

by Maddog on Dec 11, 2007 12:06 PM CST to parent up   0 recs

Where in my data...
...do you get the impression that I'm saying playing in MLB isn't more difficult than playing in NPB? Other than BA and the inconsequential (for this debate) CS%, all stats do show a marked decline when players make the transition between leagues.

You said yourself, both our ways of weighting have their advantages and disadvantages. Mine skews it towards players with more playing time (ergo, more talent), while yours skews in in the opposite direction.

I prefer my method, because while it does give more weight to Ichiro and Hideki Matsui and their good statistics, that road travels both ways. If you compare the career stat lines, the players with the best NPB stats also suffer the largest disparity between their NPB and MLB stats (with the exception of Iwamura, who has only played one MLB season so far). Does that mean that MLB is harder for players the better they are? In some ways it is, I suppose (no pitching coach is going to make a game plan centered on stopping So Taguchi), but it's more an indication that, like any player facing inferior competition, their numbers have farther to fall when the playing field evens out. And despite the fact that they suffer sharper declines in their statistics, they will always be better than the players they outperformed in Japan.

by Ayralin on Dec 11, 2007 12:47 PM CST to parent up   0 recs

I'm interested to see on what grounds...
you think Fukudome will outperform (or even match) Hideki Matsui in the major leagues as an offensive player.  Their OBP was similar, average was similar, power was in favor of Matsui, speed was in favor of Fukudome.

And if you use the last 3 years in Japan as the more relevant reference point, the argument swings even further in favor of Matsui.  Matsui had a better average, OBP, and power in his last 3 Japanese seasons than Fukudome did.

by SouthernCub on Dec 11, 2007 9:30 AM CST to parent up   0 recs

I never said that he would...
...but I'm not declaring that he won't and using that to justify invalidating my own model of what I'm doing.

All I did was apply MLEs to the data set we had of Fukudome and did a weighted average of the last three seasons. That's about as bare-bones as you can get with a projection system for NPB players, really.

But if you don't like the result you get when modeling a problem, what you should be doing is looking for ways to improve the model. There's absolutely no point to the sort of Kentucky windage Maddog was applying.

(The other point -- we're all very good about understanding park and league differences around here, so I'm suprised that more people don't grok this; even if Fukudome puts up better-than-Matsui numbers next year, it won't necessarily mean he's better than Matsui. You have to account for things like the fact that Matsui plays in a tougher league, probably in a tougher park, and then you have to baseline everything out at league average if you want to compare players from across years that well.)

FREE CARMEN PIGNATIELLO!

by cwyers on Dec 11, 2007 12:10 PM CST to parent up   0 recs

What would those numbers have projected...
for Matsui based on his last 3 years in Japan?  Because he was much better than Fukudome (by ~ 100 OPS points and ~20-30 OBP points) over each of their last 3 years.

I'm just wary of projections that suggest he'll put up better numbers than Matsui.  Especially when you consider that Matsui plays in a LH hitter's haven in NY, and actually has better numbers at home than on the road.

I hope I'm wrong and that Fukudome comes in and tops .850 OPS.  I just don't see it, based on those previous players' stats.

by SouthernCub on Dec 11, 2007 12:17 PM CST to parent up   0 recs

Three-year weighted projection...
...for H. Matsui in 2003 is .307/.395/.533. He didn't do that in 2003, but he did pretty close to that in 2004.
FREE CARMEN PIGNATIELLO!

by cwyers on Dec 11, 2007 12:39 PM CST to parent up   0 recs

And then never again in any other season...
so we've got Matsui falling short of those projections in every season in the big leagues (coming very close in 2004).  But other than that year, he's been about 25 points lower in OBP and 50 points lower in SLG.

I'd say that Matsui's 2004 is the high-end outlier, and his 2003 was the low-end outlier.  His 2005 and 2007 are remarkably similar to each other, and to his MLB career average.

So based on that, I'd say your estimates for Fukudome are too high.

by SouthernCub on Dec 11, 2007 12:42 PM CST to parent up   0 recs

There's every reason to do so.
When the model spits out numbers that are so clearly wrong as yours did, you make adjustments.  You don't just say, "well, this model says this so that's what he's going to do."  You apply some common sense and make adjustments to the model.  We know for an absolute fact that Fukudome is not as good as Hideki Matsui.  He never has been close and he never will be.  Why on earth would you settle for a model that spits out numbers that says he's going to be just as good?  That makes zero sense to me.  All you've done is essentially waste your time with a mathematical exercise.  Without applying some common sense to your model, it has no value.  

You may as well try to tell me that David Eckstein is going to be better than Derek Jeter.  It's as wrong as saying Fukudome is going to post the same numbers as Matsui.  

I love stats and think they are far more valuable than most people who visit this site, but you're applying a significantly flawed model to a player from Japan and the result is he ends up being as good as someone he's never been as good as.  I understand the need to defend what you've done, but to defend it while attacking someone else who made these common sense adjustments is just plain silly.  

Your system is wrong.  The projection is wrong.  The answer is to fix it...not regurgitate it as if it's right when you know it isn't.

by Maddog on Dec 11, 2007 12:25 PM CST to parent up   0 recs

You didn't adjust your model!
You just substituted a hunch for the results of the model.

And your "absolute fact" could well be wrong. You've yet to demonstrate that it's a 100% certainty that Matsui is better than Fukudome. Nobody's going to argue that conclusion is unlikely, but you have a LOT of work to prove it's a certainty.

Oh, and then you ignore all the stuff I wrote about league and park effects. Yay!

FREE CARMEN PIGNATIELLO!

by cwyers on Dec 11, 2007 12:35 PM CST to parent up   0 recs

I think he's saying...
that you should have adjusted your model, given that the results look surprising.

by SouthernCub on Dec 11, 2007 12:39 PM CST to parent up   0 recs

Yes, but that's not what HE did.
And what I was specifically calling out. You can't just call the model out on a specific instance without trying to get to the underlying cause. And people who are unwilling to give the model a chance to present arguements to the contrary face the danger of not doing analysis at all, but cherry-picking data to fit predetermined conclusions.

To the larger point -- to improve the model I need more data, really. Park effects, more recent MLEs... all of which is difficult to come by. But it's very dangerous to look at eight players and start making generalizations about their performance; the sample size just isn't what it should be.

FREE CARMEN PIGNATIELLO!

by cwyers on Dec 11, 2007 12:49 PM CST to parent up   0 recs

Sure, more data is always helpful...
but his point is that if the model fails the common sense test, it's worthless as is.  And thus, it's somewhat pointless to present the results as an estimate.

I'm a numbers guy, and I spend way too much of my life in excel.  I'm all for quantitative analysis as a means of projecting future success.  But face validity test is the ultimate challenge of any mathematical model, and I don't think the model (as is) passes that test.  Thus, I don't buy those projections for Fukudome.  

Also, it doesn't seem to just fail the common sense test.  Given the small sample, the model should at least fit the data that ARE available somewhat well.  I have trouble believing the model fits any of the Japanese players very well.  Perhaps if it did, you could just argue that Matsui is an outlier on the low end.  But if it doesn't, then there's clearly something wrong with the model.

by SouthernCub on Dec 11, 2007 12:59 PM CST to parent up   0 recs

Once again...
...you're putting far too much weight on the performance of eight players and trying to build a model to fit that "reality," when the larger question is, do those eight players fit reality?
FREE CARMEN PIGNATIELLO!

by cwyers on Dec 11, 2007 1:02 PM CST to parent up   0 recs

Then the larger question is...
what's the value of your model?

The goal of any model is to predict (as closely as possible) what actually is going to happen.  The most reasonable way to develop a model is based on what we already know to be true (i.e., current data).  If the model doesn't fit the current data as well as possible, then it's value is greatly compromised to say the least.

In other words, why would anyone assume your model fits reality better than the 8 players who really have played?

by SouthernCub on Dec 11, 2007 1:14 PM CST to parent up   0 recs

Ok, let's try this again.
There WAS a problem with the model; I wasn't properly "normalizing" for plate appearances. (One could do the same thing for the model the OP used.)

.297/.378/.412

Doesn't significantly alter the AVG/OBP, but it does bring the slugging back into the range of expectations.

(Revised "projection" for Matsui: .307/.395/.425)

So, lessons learned:


  2. I was wrong.
  3. Fukudome's OBP still looks very healthy, and addresses a major team need.
  4. No Kentucky windage used.

FREE CARMEN PIGNATIELLO!

by cwyers on Dec 11, 2007 1:44 PM CST to parent up   0 recs

Kentucky windage?
Who provided a more realistic analysis of what one could expect from Fukudome?  You or me?  You did an analysis, presented it as fact, and failed to address to obvious issue that it was so far out of whack with reality and defended it.  

I did an analysis, presented the data, and challenged that data using one similar player who played in both Japan and the US.  No windage.  I compared the stats and presented a more realistic projection as a result.  

If you'd like to continue to believe that because you did something different, that it's better...be my guest.  You presented something that anybody with 2 eyes could see was clearly out of whack.  

No, 8 players is not a large sample size, but if your system is of any value at all, it should at least be more accurate than what you've come up with.  The most common error in projecting a player is playing time.  The players who do reach their projected level of playing time fall equally above and below (and equal to) their projections.  Not one player you did was remotely close.  

Again, 8 is a small sample, but you'd expect different results nonetheless.  I can live with it not having different results if the results you presented weren't so far from reality.  Your system projects these Japanese players to be superstars here in the US when the only one who has achieved that status is Ichiro (and that status is questionable in my opinion as it's more on hype than anything else).  

Most of the numbers you've presented here are like video game numbers.  Yet you attack a prediction from someone else?  Really?

by Maddog on Dec 11, 2007 2:29 PM CST to parent up   0 recs

I substituted common sense for something
that was clearly wrong.  

Hideki Matsui was better at age 22 than Kosuke Fukudome was at age 30.  He's by far the better player.  Is it possible that Fukudome could suddenly become the better player here in the US?  Sure.  It's also possible that I win the lottery.  It's also possible that David Eckstein has a better 2008 than Derek Jeter.  But any projection system that spits those scenarios out is wrong.  Just plain wrong.  Under no rational circumstances can a quality mathematical model tell us that Fukudome will be as good as Matsui or Eckstein as good as Jeter.  You may as well just throw the work out because it's just plain wrong.  

I don't even know how this is hard to understand.  Look at their stats.  Only under the most unlikely circumstances is Fukudome going to come to the United States and suddenly be as good as Matsui despite not ever being as good as him at any point in his career.  This is as much a fact as 2+2=4 is.  

by Maddog on Dec 11, 2007 12:46 PM CST to parent up   0 recs

But you're assuming that an .880 OPS...
...in the American League East in 2003 is equal to and .880 OPS in a neutral context in 2007+1. That ain't necessarily so. So just because the OPS figures look similar doesn't mean the projection is calling for Fukudome to actually be better.

I keep repeating this; maybe at some point somebody will start at least responding to it.

FREE CARMEN PIGNATIELLO!

by cwyers on Dec 11, 2007 12:53 PM CST to parent up   0 recs

NY is a hitters' park for LH hitters...
and I'd guess the AL East from 2003-2007 to not be substantially tougher on hitters than any other division.  It's substantially tougher on non-NY/BOS pitchers, but I don't think hitters (especially NY/BOS hitters) face that much tougher a test.

Thus, I don't think I'd argue that Matsui's MLB numbers have been deflated due to park/league effects.  Given that he actually has better home splits than road splits (and his numbers within the AL East work out to about the same as his overall numbers), I'd definitely say he hasn't been hurt by MLB park/league effects.

by SouthernCub on Dec 11, 2007 1:11 PM CST to parent up   0 recs

Why would I?
We're talking about points here.  It's been a few years.  There's been no significant change in offense in this span.  There's no reason to respond to it because it's of no value to this discussion.  We're not talking about 1940 vs. 2007 here.  We're talking about a 5-year span.  

If anything, there were more runs scored in 2003 than in 2007 or likely 2008 so this .880 OPS you project for Fukudome would likely be equal to an .890 one in 2003 making it even more silly.  

by Maddog on Dec 11, 2007 2:35 PM CST to parent up   0 recs

I kind of agree with Maddog...
minus the harsh tone (which goes both ways).

I'd be very interested to see what the MLEs would say Matsui should have done upon coming over.  I suspect that the projections are much higher than what he actually did.  That's the sort of "face-validity check" that Maddog is saying this analysis is lacking.

I've done a similarly-crude calculation of my own based on the numbers of Matsui, Ichiro, and Kaz Matsui.  Each of them had remarkably similar transitions to the major league, in terms of their dropoffs in OBP and OPS.  Applying those trends to Fukudome's stats come up with a number around .800 OPS.  Given that Fukudome wasn't as productive in Japan as Hideki Matsui, it seems like a reasonable output.

Of course, it doesn't consider park effects, either in Japan or MLB.  I'm not sure I'd expect the numbers to jump wildly in either direction though, unless the players played in fairly extreme hitters'/pitchers' parks.

by SouthernCub on Dec 11, 2007 12:36 PM CST to parent up   0 recs

My apologies for the poor tone.
It just irritates me that someone can take a projection that projects Player A to be as good as Player B despite Player A having been so inferior the rest of his career.  

by Maddog on Dec 11, 2007 12:49 PM CST to parent up   0 recs

You do realize the age difference, right?
Age at beginning of first MLB season:

Ichiro - 27
Hideki Matsui - 29 (turned 29 in mid-June)
Kazuo Matsui - 28
Akinori Iwamura - 28
Kosuke Fukudome - 31 (turns 31 in late-April)

So, even the closest guy came over a full two years younger than Fukudome. Knowing what you know about baseball players' decline "due to" age, do you still comfortable making that projection? If you're going to use the age 27, 28, 29 and 30 seasons as data, shouldn't you really make an "off-the-wall" adjustment like Maddog did? Either that, or just straight-up drop any data before the age-31 season, but then that wouldn't give you a lot to work with.

I think you guys are vastly overrating Fukudome. As usual, the Cubs are late to the game and doing it wrong. --sigh

by tyger1147 on Dec 11, 2007 10:09 AM CST to parent up   0 recs

Who knows?
I'd happily take an ops of around .850, and cwyers perdiction of .895 would make me as happy as can be.  I think he ends up closer to .800 than .900.  The predictions are great, keep them coming, but in the end we're just going to have to wait and see.

Don't forget to do predictions for Fukudome based on Wrigley and Petco, because we still have to wait and see about that too.......

I know that you believe that you understood what you think I said, but I am not sure you realize that what you heard is not what I meant. - Robert McCloskey

by pageian on Dec 11, 2007 7:52 AM CST   0 recs

Park factor
Disclaimer: Applying park factors to non-neutral statistics can skew results badly, as you'll notice quickly. I may dig around later today to see if I can find park factors for the Japanese stadiums to try and get some of the noise out my projections.

My projection for Fukudome (498 AB), with Wrigley Field's park factors applied to half:

158 hits, 33 2B, 6 3B, 14 HR, 65 BB
.317/.396/.492 (.888 OPS)

My projection, with Petco Park's factors applied to half:

141 hits, 27 2B, 7 3B, 11 HR, 56 BB
.283/.356/.431 (.787 OPS)

by Ayralin on Dec 11, 2007 8:29 AM CST to parent up   0 recs

All due respect,
you've done something wrong in your projection if it projects him to be a near .900 OPS hitter in any park in this country (you could include the home of the Little League World Series in that if you want to).  I have a better chance of winning the lottery twice this week than Fukudome has of coming anywhere close to a .900 OPS.  He's never been as good as Matsui and he's not suddenly going to become better when he switches to a far more difficult league.  If anything, based on what Matsui has done as well as other declines, it's not at all unreasonable to project that he fails to even hit for an .800 OPS.  

Furthermore, he'll be 31, which is the point where a player's decline begins more rapidly.  

This guy is going to get a shitload of money and not be very good.  I wonder how a Japanese person will do when they get treated the same way Jack Jones did, because it will happen...probably sooner than any of you think.

by Maddog on Dec 11, 2007 9:30 AM CST to parent up   0 recs

As I said in my disclaimer...
...applying park factors to non-neutral statistics can skew results badly. Which you will apparently not notice.

Placing factors on top of projections derived from other factors creates some whacky numbers. The "park factor" projections I posted essentially place half a season of MLB park factors on top of stats derived from a full season of MLB equivalent statistics with the original Japanese park factors not eliminated.

Not in my wildest dreams do I expect Fukudome to put up a .900 OPS in the Major Leagues. Considering his skills and age, the .850 OPS I have in my spreadsheet is pretty much a projected peak.

by Ayralin on Dec 11, 2007 9:51 AM CST to parent up   0 recs

OK...that I agree with.
I misunderstood what you were saying.  I, too, think an .850 OPS peak is what we could see.  I think it's more likely than not that he doesn't reach his peak potential as most players never do, but I'd agree with you.  

by Maddog on Dec 11, 2007 10:57 AM CST to parent up   0 recs

Japanese players are different.
They approach the game, and its physical training, differently than Americans.

Look at Ichiro, who is now 34. You see any decline there? It's been said somewhere here that Hideki Matsui is "declining", but I don't see that in his numbers.

Maybe they'll "overpay" by American standards, but they are getting something intangible as well. Yes, this is a "feeling" more than something you can measure.

Incidentally, if he fails in the USA, there's no doubt in my mind the Cubs could sell at least part of his contract back to Japan, where he'd be welcomed back as a hero.

"That's my opinion and if you don't like it, well, I have others." ~ Groucho Marx

by Al on Dec 11, 2007 10:25 AM CST to parent up   0 recs

japanese players declining?
pretty simple measure here, we'll use OPS+ again:

we'll judge from 2004 on since this is the year both players hit 30 (Ichiro and Hideki)

Year          Ichiro        Hideki
2004           130           137
2005           113           130
2006           106           128
2007           122           123

Ichiro's 2007 was buoyed by a .389 BABIP, if that regresses to the normal levels for him you're looking at an OPS+ in the 108-114 range

From a purely offensive standpoint both players have been in a pretty slight decline phase. It's a slight decline, but there is evidence of some decline being there based on their numbers

by DartmouthCubsFan on Dec 11, 2007 10:36 AM CST to parent up   0 recs

Loud, sustained applause!
Just because it's not a rapid fall from grace doesn't mean it's not happening. Just because someone has a good year, doesn't mean they aren't generally declining.

I'm sure we could all go find a few 36-year-old players that had a year similar to their primes and say, "See, there is absolutely no decline." That doesn't make it true.

by tyger1147 on Dec 11, 2007 10:38 AM CST to parent up   0 recs

Define "normal levels" for...
...BABIP here.

Ichiro's career BABIP is .359, so you're talking he only exceeded his career norms by 30 points there. He pretty much has to be considered a "freak of nature" when it comes to BABIP in general, because he has a consistant, measurable talent for exceeding his expected BABIP based on his line drive percentage.

I think 2004 is simply an outlier (a "career year"), and that Ichiro isn't regressing much at all.

What does this mean? Absolutely nothing; Ichiro is Ichiro, and there really is nobody else remotely like him.

FREE CARMEN PIGNATIELLO!

by cwyers on Dec 11, 2007 7:25 PM CST to parent up   0 recs

right
and 30 pts in BABIP is pretty significant...

and the 30 "extra points" in batting average would impact his OPS+ by somewhere between 5 and 10 points

by DartmouthCubsFan on Dec 11, 2007 8:02 PM CST to parent up   0 recs