Ichiro Suzuki not only should not win the MVP, he should not get one single vote for MVP. He hasn't been one of the 10 most valuable players in the American League, which is what one needs to get a vote.

Even more telling, Ichiro has not been the most or second most or even third most valuable player on his own team. The Mariners have had a lot of great performances this season, so being that club's 4th most valuable player is not a knock. But league MVP? That's nuts!

Keith Woolner's VORP is, in my opinion, the best tool in measuring value. It's based on the idea that a player can do one of two things to add value to his team: create runs for your club or prevent the other team from scoring runs. Add those together for each player (minus what contribution a replacement-level player would have contributed at the same position) and you have a player's VORP (Value Over Replacement Player). The numbers are, of course, adjusted for ballpark.

Here are the 14 most valuable players, so far, in the AL in 2000:

1. Jason Giambi 92.9

Again, Rich, I feel compelled to point out that while I regard VORP as the single best measure of offensive value, it does not factor in intrapositional defense. While it's extremely unlikely that Ichiro catches Boone because of this, he could easily be number two on the Mariners. If he's 10 runs better defensively than the average RF, that puts him even with Edgar, who's obviously not better defensively than the average DH. Similarly, he could pass Manny Ramirez or Derek Jeter. He's nowhere near A-Rod or Giambi, of course.

VORP is not the end-all be-all that you make it out to be.

Ichiro is a very good right fielder. But the evidence is not very strong that he is a great right fielder. He is currently 5th in the American League in Zone Rating among RF's; and despite his vaunted arm, he is now 7th in assists among AL RF's.

Even though VORP apparently fails to distinguish on a defensive quality basis - and thus overrates someone like Derek Jeter and underrates someone like Jason Giambi (who really is a superior first baseman) - Ichiro is not so great in the outfield that he should be moved up many spots in the VORP rankings.

Why do we even need to watch any of these guys play? Let's just let the computers pick our MVPs for us. That way we'll get the right choice every time.

And if that works, then we can assign championships to teams which have the best regular season run differential. Besides, it's going to be too cold for the World Series this year anyway.

More seriously, what would be interesting would be to have the players vote for the MVP as well as the writers. They may not come up with choices we would agree with any more than the writers do, but their votes would certainly have more standing.

And if the players voted Ichiro the MVP, I could even live with that.

I'm not clear why the players' choices should or would have any more standing than those of the writers.

Andy, my opinion is that the vote for MVP should be based on the best evidence available. As much as possible, I would like to leave personal prejudice out of the vote. David's reply above is a good one, because it directly addresses the "evidence" I presented. By whining "why don't we just let a computer pick the MVP", I infer that you don't like the evidence I presented, but that you can't refute it.

I really don't see why players would make the best interpreters of evidence. It's true that they witness a lot of games -- and that is valuable in making a judgment. But an American League player sees only 13% of the games played by all AL clubs. If he does not use statistical evidence to make his vote, then he will be casting his vote while ignoring 87% of the games his peers played. And there is absolutely no reason to believe that baseball players are the best prepared to interpret statistical data.

Unfortunately, the baseball writers do make the judgment. And all of the arguments against players (or coaches) casting a sound vote fit most baseball writers. Not only do some writers have antiquated (and incorrect) notions of value, but they tend to be personally biased because of the fact that, for example, a writer who covers the Indians will see a LOT more of Roberto Alomar than he will of Brett Boone.

I happen to see a LOT of A's games - about 15 in person this year and the rest on TV. I also have seen a LOT of Giants games, but only 1 in person, this season. I have also seen all the other 28 teams play on TV this year. Does that make me a qualified voter? Of course not. I have not seen the vast majority games played in baseball. So my making a decision only based on witnessing games and seeing a lot of highlights would inevitably be a biased decision. It just so happens that my bias - to vote for Giambi and Bonds, respectively - fits with the best statistical evidence we have.

So who should vote? Should it be a computer? I really don't have a problem with having a computer making the pick, as long as Most Valuable Player takes into account these factors: 1) ballpark effect, 2) how many runs a player creates with his bat, 3) how many he creates drawing walks and beans, 4) how many runs he creates running the bases, 5) how many runs he prevents with his glove, 6) how many he prevents with his arm, and 7) how many runs a pitcher prevents (over a replacement player).

Rich...

I like VORP, but we should also keep its limitations in mind:

I'll go through each of your points:

1) There isn't general agreement as to whether one should use single-year or multi-year park factors.

2) 3) and 4) The various RC and XR formulas use an average value for an event. These are not context specific. You may say that there is a lot of luck involved and that the XR or RC is good for measuring abilities - I can agree with that. But what we are talking about here is the MVP and that middle letter refers to value. It is certainly quite possible for a player to get hits or walks in key situations over the short term and thus for the value of his contribution to be greater than is reflected in his component stats. That is how teams exceed or fall short of XR run projections.

5) 6) VORP does not account for this excecpt to adjust for positional values (again should you use long-term or single season position adjustment factors). Defensive value can vary a great deal. We don't know how many bases runners have taken on Ichiro, though we do know how many runners he's thrown out. We don't know how many balls an average RF would catch playing in Safeco RF. The same concerns about context apply. Events which take place in close games late in the game have greater importance than in blowout games.

7) Finally, we don't know what a replacement-level player is. There is no agreed upon definition of replacement level. A pitcher's ERA or RA is heavily influenced by the defense behind him. There isn't really a clean way to separate the two.

With all these limitations, it would be foolish to simply consider one measure like VORP and rank the players accordingly.

Each player has only seen Ichiro or Giambi or any other player between six and nineteen times this year, but I did not suggest that any one player be allowed to choose the MVP. In fact, I only suggested a kind of parallel award by the players, and this would of course be voted on by ALL of them, who COLLECTIVELY would have seen ALL the games. Sort of like your basic invisible hand theory of Adam Smith, elaborated on by the Austrian school of economics. And I make no apologies for suggesting that the average player, whether trained in the fine art of statistical interpretation or not, is more qualified to detect "value" qualities in his peers than the average writer or fan. And to take it even further, I would have even more respect for the players' collective choice than a computer's choice. I doubt if the Yankees, for instance, would fall into the statistical trap of thinking that Mussina was more valuable to them this year than was Clemens. I have, however, read a few fans and writers armed with run support averages try to advance this very claim.

BUT, I realize the pitfalls of any choice made by human beings acting on incomplete information, so my ideal compromise solution (never to be acted upon, of course) would be for ESPN and the newspapers to give full publicity to the new wave of statistics, so that the players and writers (who do watch ESPN and read the papers, if not much else) would have supplementary information with which to enhance their judgments. Much in the same way I would compel HOF electors to pass a knowledge test (including awareness of the types of statistics discussed on this site) before granting them the ballot.

And we can still get together after all these votes are taken and argue how stupid the choices were. It's the American way. As far as I'm concerned, anyone who doesn't vote for Giambi and Bonds is a chowderhead of the worst order, and probably on the payroll of the East Coast Media Elite Conspiracy.

I might be tempted to vore for Alex Rodriguez. True, his team is not in contention, but that cuts no ice with me (one player does not a team make).

He has been the best player in the AL for several years are has definitely been short-changed in past votes.

P.s. I am not a chowderhead - note Alex plays for Texas, not an East Coast team.

And Andy, why would you automatically assume Clemens has been more valuable than Mussina ?

ABdy...

One more note.

A pitcher's won-loss record is perhaps the most potent statistical trap there is. Short term pitcher wins and losses have very little significance. And yet time and time again writers have awarded Cy Youngs to pitchers with impressive W/L numbers - how else to explain Bob Welch, an average pitcher having a decent year winning the Cy Young back in the 1990.

It's no use asking the Yankee players who has been more valuable - they read the papers and know the won-loss records of both pitchers. They are as subject to misconceptions about the significance of won-loss records as the average fan is.

Wins and losses are a function of team play and not a picther's performance.

A pitcher's job is to prevent the opposition from scoring runs. Period. Full stop. Only evidence directly relevant to that task should be used to evaluate pitchers.

A pitcher's W/L is not only influenced by how many runs are scored by his team, but also how effetive his bullpen is. A good bullpen can help erase potential losses by holding the opposition until the batters get going.

Also, there is no evidence that a pitcher gives up more runs when his team is way ahead than when the score is close. If you want to claim that Clemens gives up runs in blowouts because he relaxes, then provide some hard evidence for this.

Bob...

Since you make me reconsider my opinions all the time, I would hardly call you a chowderhead. Kindly allow me a bit of rhetorical excess now and then, even when my attempts at parody descend into self-parody.

I think that Alex is unquestionably the best player in the

Bob...

When I wrote the above, I had only seen your earlier post. When I get back from work, I'll attempt to address your second one.

Maybe his teammates consider him more valuable, however, as a Yankee fan, for Game 5 against Oakland in the Bronx, I'd rather see Mussina on the hill. For a Game 7 at Safeco, or Enron, I'd rather see Mussina on the hill. But for my money, the best pitcher in the AL is Freddy Garcia.

I'm not sure how VORP is computed, but just taking runs, innings, and Robert's park factors I get Freddy Garcia at 69.83 RP (through Thursday afternoon), Mays at 64.32. This is using replacement at 1.25*league. Garcia deserves the AL Cy Young Award, and in my opinion, he will become this generation's best pitcher, this group of 23-27 year olds. He won't even be 25 until early October. He scares the hell out of me in the playoffs this year, and I think he'll be a household name by the time October is over.

His career is on a nice parallel w/Greg Maddux, except while Maddux was getting pounding in the big leagues as a 20-21 year old, Garcia was growing in the minors. Both had unexpected great years at the age of 22, and have been going forward ever since.

22 ERA+ Maddux 114 Garcia 123

Just updated my numbers, now I have Garcia at 68.16 and Mays at 69.78. Going back to the 1983 Abstract, I used Bill James formula for determing RP based on W-L w/respect to run support. Garcia shows up at 65.73 that way, Mays at just 55.52, meaning his record should be much better than 16-13 given his support and runs allowed. James suggested weighing these 50/50, but I think 70/30 would probably be more appropriate (James used this weight in the 1982 Abstract).

I'd vote Garcia, Mays, Clemens if I had a vote (Clemens 20-1 is not explained just by better run support, his record is still better than expected). But I'd rather have Mussina on the hill than Clemens. This gets back to your situational concerns a few posts back Robert.

The top 15, weighted 70/30:

1. Clemens 67.77

Robert - I am replying to your comment that VORP is imperfect (and you gave several examples of how that is so). I concur. But what I like about VORP is the approach that it takes. When I wrote above all of the components that I think should go in to "value", I did not consider that to be limited to VORP or exclusive of some other considrations.

Side note: someone above commented that maybe the reason Ichiro has comparatively few outfield assists is because, due to his reputation, no one will take a chance and run on him. That very well may be true. It very well may be false, too. It seems hard to prove one way or the other. Jermaine Dye, on the other hand, has a reputation as a great right-fielder with a cannon for an arm. Everyone in the American League knows how good Dye's arm is. Yet, that has not stopped some players from trying to advance against Dye, as he has gunned down twice as many runners as has Ichiro this year.

Rich..

Raul Mondesi is another good example of what you are talking about. But his assist rate has definitely tailed off in the second half because many runners will not challenge him.

Is there anyway to determine the denominator here, the number of guys who try to run? This is the missing piece, but I'll be hanged if I can think of a way to track this.

Get those STATS Inc. guys who are tracking Zone Ratings to monitor it - how many runners had a reasonable opportunity to try to advance when a ball was hit to a particular outfielder. It'd be subjective, but there's no completely objective way to measure it.

Robert,

Mondesi also has a great reputation for his arm. And he leads all AL right fielders in assists, with 17. That happens to be Raul's best season ever for assists, though he's had 16 a few times in the past. Ichiro has 7. Mike Cameron, the M's CF currently has 8.

I certainly agree, though, with the importance of the "denominator". It very well may be the case that against a lesser arm, runners would have attempted many more bases than they did against Ichiro this year.

Bob...

To pick up where I left off....Unfortunately, I don't save my BB Weeklys, and haven't been able to find a site which gives pitchers' game-by-game logs, or even box scores, so I'm afraid what follows is based on little more than intuitive memory. But...

My recollection is that this year Mussina has mostly been his old self, with occasional flashes of downright brilliance. However, he has also been knocked out early on quite a few occasions, effectively taking the Yankees out of the game before they had much of a chance to score some runs.

Clemens, on the other hand, has pitched well enough in nearly every game to keep his team in the fight. Sometimes this has meant holding the opposition close until the Yanks had a chance to rally; sometimes this has meant getting a big lead and letting the other team get close but no further. There is no question that his W-L has been inflated by his run support, but there is also no question that in 2001, Clemens has been the Yankees' stopper, the one who brings losing streaks to a halt. I can hardly recall a bad outing of his since May, or at worst early June. Mussina has had at least four or five games within that period where he let his team out of the game early. For this reason I'd say Clemens has been the more valuable of the two up to this point.

Two caveats. First, as others have pointed out in other threads, Clemens has not faced the calibre of opposition that Mussina has. This is a point in Mussina's favor, no question. And like Scruff, I'd probably rather have Mussina facing Oakland or Seattle in a game 5 or game 7 in post-season than Clemens. But this is based more on my memories of Clemens prior to the LCS and WS last year, and in those two series he was completely dominant in pivotal games.

So on balance, I'd vote for Clemens for Cy Young, since this is based on the regular season. Now if he can bring it against Oakland and / or Seattle in the playoffs, he'll really have a season to remember.

"Get those STATS Inc. guys who are tracking Zone Ratings to monitor it - how many runners had a reasonable opportunity to try to advance when a ball was hit to a particular outfielder. It'd be subjective, but there's no completely objective way to measure it. "

In "The Baseball Scoreboard," Stats, Inc. has a "chapter" called, "Which Outfielders Know How to Hold 'Em?" They use their play-by-play database to find out which outfielder allowed the lowest percentage of extra bases to be taken on them. Last year, Raul Mondesi led RF with 30.1%, followed by Vladimir Guerrero at 40.3%, Bobby Abreu at 41.3%, Paul O'Neill at 43.8% and Albert Belle at 47.3%. Of course, this ignores the runners who were thrown out trying to advance, which is even better than preventing the advance.

To really get at this issue, you need to use the base/out table and include advances, non-advances and assists. A player who has intimidated the opposition into not trying for the extra base will probably come out behind one who gets a lot of assists, because an out is worth quite a bit more than a base.

Andy: "Unfortunately, I don't save my BB Weeklys, and haven't been able to find a site which gives pitchers' game-by-game logs, or even box scores, so I'm afraid what follows is based on little more than intuitive memory."

Andy,

You can find the game logs on BigLeaguers.com and on espn.com.

"My recollection is that this year Mussina has mostly been his old self, with occasional flashes of downright brilliance. However, he has also been knocked out early on quite a few occasions, effectively taking the Yankees out of the game before they had much of a chance to score some runs.

Quite a few occassions? Well, three occassions.

"Clemens, on the other hand, has pitched well enough in nearly every game to keep his team in the fight."

In fact, Clemens has had pretty good luck. In 10 of his outings, he has recorded an ERA from 4.50 to 7.94 for each game - yet he only lost once. But on the other hand, Clemens has had a huge number of games this season in which he pitched good enough to win with average run-support.

Here is the game log for Mussina and Clemens (IP, ER), listed in order from best performance to worst:

Mussina~~~~~~~~~~~Clemens

I don't know what happened to my table above.

Here are the logs in paragraph form for Mussina and Clemens.

Allowed 0 runs - Mussina 7 times, pitching 53.7 innings; Clemens 4 times, pitching 28.7 innings.

Allowed 1 run - Mussina 6 times, pitching 44.7 innings; Clemens 3 times, pitching 20.3 innings.

Allowed 2 runs - Mussina 3 times, pitching 23 innings; Clemens 10 times, pitching 67.7 innings.

Allowed 3 runs - Mussina 6 times, pitching 41 innings; Clemens 4 times, pitching 29 innings.

Allowed 4 runs - Mussina 3 times, pitching 16 innings; Clemens 3 times, pitching 18.7 innings.

Allowed 5 runs - Mussina 3 times, pitching 19.3 innings; Clemens 6 times, pitching 37.7 innings.

Allowed 6 runs - Mussina once, pitching 5 innings.

Allowed 8 runs - Mussina 2 times, pitching 6 innings.

Rich...

Thanks for the BigLeaguers.com link. Where have I been all these months?

After charting Clemens vs. Mussina, though, it's still fairly clear to me that Clemens has been the more valuable of the two this year. Certainly not to the extent that their respective W-L records would indicate (Bob is right about the often-deceptive simplicity of that figure), but still there is a distinct Clemens edge. While I realize that there is room for argument in this game or that, Clemens has had at most 5 or 6 starts where he either lost or was somewhat lucky to escape, and one of these was the day after the All-Star break where he volunteered to pitch on 3 days rest and just a few hours sleep, when the Yankees had no other starter available. Mussina has had about 9 starts where he either gave up a blg cluster of runs fairly early, or (in one case) he let the first run be scored against him, then another, and then another, in a game the Yanks lost 4 to 1. I didn't count that game among the 9. He also had one "Clemens"-type game (which I also didn't count among the 9) where he blew most of a 6-0 lead, but the Yanks exploded and eventually won 15-5. Clemens, by the way, has had only one bad start since May (other than the 3 day rest one, which he won but was lucky to), an August 20 game against Texas where he was ineffective but had no decision. Mussina has also been fairly consistent since the All-Star break, though still with two bad starts to Clemens' one. All in all, two pretty good pitchers to carry into October, and with Pettitte and Hernandez, I wouldn't bet too much against this team. My biggest worry is the bullpen, even Rivera, whose ERA belies a rather troubling number of blown saves, most recently just last Friday against Baltimore.