Change-UpDecember 29, 2010
Jim Rice, Chet Lemon, and How I Think About Wins Above Replacement
By Patrick Sullivan

Inspired by Mike Axisa's new Twitter feed, @WARGraphs, I have been playing around with a new tool, or at least one that's new to me. As you may know, WAR Graphs is a Fangraphs feature where you can compare up to four players by Wins Above Replacement. Once one enters their desired search, three graphs appear. One shows how the players compare in their nth best seasons. The second shows how they compare year-by-year over the course of their careers. The final one shows how they stack up by age.

It's a neat tool, and a handy one when like-minded folks are looking to settle a quick dispute. For instance, as a Red Sox fan, a pet issue of mine has been the travesty that is Jim Rice's Hall of Fame enshrinement while Dwight Evans never amassed more than eight percent of the vote. Anyway, here are two of the three WAR Graphs for a Rice and Evans comparison.

Dewey%20%26%20Rice%201.png

Dewey%20%26%20Rice%202.png

Because the topic is something of an obsession for me, I tweeted my findings from this WAR Graphs search last night.

Jim Rice & Dewey were similar, if you ignore Dewey's 35-40 seasons when he hit .283/.387/.470 (133 OPS+) http://is.gd/jFVpAless than a minute ago via TweetDeck

When he saw this, Dave Cameron responded with the following:

@PatrickSull My favorite - run Jim Rice against Chet Lemon; pick up jaw.less than a minute ago via TweetDeck

And sure enough, here is the WAR Graphs comparison of Rice and Chet Lemon.

Rice%20%26%20Lemon%201.png

Rice%20%26%20Lemon%202.png

Chet Lemon and Jim Rice are more or less indistinguishable. Chet. Lemon.

**********

All of this was a long and graphical way of setting up the point of this post, which is to articulate a coherent way to think about WAR in the context of Hall of Fame voting. Jonah Keri has done a really nice job advocating for Tim Raines in a more visceral way than Rich Lederer has for Bert Blyleven. Rich has gradually won over voters by reminding them time and again of Blyleven's statistical dominance. Keri, on the other hand, will make his case with stats, but also with well-supported assertions along the lines of had Rickey Henderson never come along, Raines may well be regarded as the finest lead-off man ever. That resonates more than a WAR Graph with many.

To take it a step further, not only is something like WAR altogether unpersuasive to some, but when many see the WAR Graph above of Jim Rice and Chet Lemon, their gut may be to write off the statistic itself altogether. In other words, it's not that the graph shows that Rice and Lemon were comparable. No, the graph shows that WAR as a statistic is moronic.

But here's the thing about WAR. It lines up with so much of what we understand to be true, even before we start in on any sort of advanced statistical analysis. Here's a list of the top-10 position players by B-Ref WAR:

1. Babe Ruth
2. Barry Bonds
3. Ty Cobb
4. Willie Mays
5. Hank Aaron
6. Tris Speaker
7. Stan Musial
8. Rogers Hornsby
9. Eddie Collins
10. Ted Williams

The next five on the list are Mickey Mantle, Lou Gehrig, Honus Wagner, Rickey Henderson and Mel Ott. We are talking baseball royalty. It's not as though Nomar Garciaparra or someone crept into the top of the list because of some quirk in the statistic. It actually aligns beautifully with a list your grandfather might furnish you of the very best baseball players of all time.

Here are the top pitchers:

1. Roger Clemens
2. Walter Johnson
3. Tom Seaver
4. Pete Alexander
5. Lefty Grove
6. Phil Niekro
7. Greg Maddux
8. Gaylord Perry
9. Warren Spahn
10. Randy Johnson

The next five? BERT BLYLEVEN, Christy Mathewson, Bob Gibson, Nolan Ryan and Steve Carlton. Nobody is saying that this is the definitive list of the best pitchers of all time, ranked perfectly in order. Peak matters, for instance, and I don't want to speak for anybody else but I don't think you'll find too many stat heads saying that Niekro, Perry or Blyleven were better than Pedro Martinez or Sandy Koufax. But the point remains the same: that's a pretty darn good list in terms of how it compares to common baseball wisdom of the very best pitchers ever.

There are single-season examples, too, of the visceral or instinctive aligning with analytical conclusions. Growing up, I heard non-stop stories from my father and grandfather of how great Carl Yastrzemski was for the 1967 Impossible Dream Boston Red Sox. We would listen regularly to the WHDH-produced soundtrack to that season, including the ragtime adaptation a song whose chorus went "Caaaahhhhrrrll Yastrzemski" over and over again. Later in life, my father in law, a Long Beach, California native who studied law in Boston during the 1967 season, would tell me one story after another about how incredible Yaz was. This is a man who is no Red Sox supporter, and as prone to hyperbole as anyone you could meet. Given everything I had heard throughout my life about Yaz in 1967, you'd have thought he had one of the very best seasons ever. Having bought in more and more to advanced statistical analysis, I just assumed all of this was overblown.

Well you know what? Yaz did have one of the very best seasons ever. Go on and check it out. Aside from three insane Barry Bonds seasons, Yaz's 1967 stands as the finest year by a position player since 1958. All of that wonderful stuff I had heard about Yaz, all of what seemed like folklore, it ALL lined up perfectly with what WAR would tell you about Yaz's heroics in 1967. It was one of the truly great single seasons in baseball history.

When I see a graph like the one above of Lemon and Rice, I don't immediately assume Lemon was better than Rice or even that Lemon was the same caliber of player Rice was. I'm more skeptical of defensive data than offensive, and I have a ton of respect for what Rice did at his peak. But that's not how WAR is supposed to work, or at least it's not how I think it should work. Instead, I believe it should be your first pass.

Oh, I see here that Blyleven ranks 11th all time and Morris 119th. I probably would be wrong to vote in Morris then, and not Blyleven.

Huh, look at this: Tim Raines ranks 55th all time and Lou Brock 121st. Maybe I need to think a little differently about Raines's candidacy?

In the Rice and Lemon case, it just shows that maybe we've thought a bit disproportionately about both players. Rice is in the Hall of Fame while Lemon, well I hadn't even thought about Chet Lemon in over a decade. That doesn't seem right to me anymore now that I have taken Dave Cameron's suggestion to run the comparison.

WAR is not perfect but it cannot be ignored, either. My hope is that more Hall of Fame voters will look to the stat to help frame their decisions. If a certain player amassed many of his Wins Above Replacement in exceedingly favorable conditions, no problem. Dock him. If WAR sells short a player like Morris or Rice for whatever reason, you can make that case too. All I ask is that voters recognize how well the statistic holds up to everything we understand to be true about baseball. More often than not for the attentive baseball fan or writer, a quick pass at WAR will serve more as affirmation than an eyebrow-raising contradiction. That being the case, when it does not quite align with pre-conceived beliefs, it merits further investigation and not immediate write-off.

Comments

With apologies to my friends who are Red Sox fans, Rice is a marginal HOF candidate. He was great hitter at his peak, but he couldn't sustain it, and his cumulative statistics show that. Bob Ryan in this morning's Boston Globe has a piece deriding Palmeiro this morning-if he had stuck to steroids, I would have gone with it, but he found Palmeiro's stats unconvincing, even though they are no worse than equal to Rice's, and Palmeiro hit milestones that, not only used to be universally considered an automatic ticket to Cooperstown, but that Rice didn't even approach. Here's something to chew on for Rice fans. After his three peak years (77-79), in which he truly was a great player, for the last ten years of his career-about 5500 plate appearances (60% of his career totals), his cumulative slugging percentage was under .500, his OPS was about .820, and his OPS+ was about 116. His career OPS+ of 128 puts him at 197th all time. Good enough for the Hall Of Fame? Maybe, but by no stretch of the imagination is it compelling, and the BBWAA votes over the years show that diffidence. Boston writers have talked about Rice being "feared"-perhaps that's true-at his peak, he certainly was formidable, but it's a mushy way to get Rice in while justifying the exclusion of others who were statistically superior. The knowledge you get from your own eyes is important, but the data is more so. In the end, every voter gets to choose with his or her own biases intact. Jon Heyman just wrote a piece in SI which was remarkable in its ability to contradict itself. But if you parse out all of the self-justification, it really comes down to feel for Heyman (and, I would guess, for many HOF voters). Jim Rice, Jack Morris, Burt Blyleven, Tim Raines? We aren't talking about about Ted Williams, Sandy Koufax, Walter Johnson, or Rickey Henderson. There is a qualitative difference-and that's where all the fun comes in.

Brian Giles Nth best season WAR graph is pretty close to Rice. Granted there are dozens of such non-HoF players at that level or better who have virtually no shot at being enshrined [before maybe some vet committee ballot after their death]

http://www.fangraphs.com/graphsw.aspx?playerid2=990&playerid3=1010897&playerid4=&playerid5=

Thanks for the article, Sully.

A couple points. One that you certainly know is that there are two popular WAR implementations and others exist too, making it hard for the concept to go mainstream. That's an inevitable thing, so I'm just noting it rather than bemoaning it.

Second, your point that WAR can create a reasonable list of the top 15 batters and therefore is useful to differentiate at a first pass among Dwight Evans, Chet Lemon, and Jim Rice is not very persuasive. The top MLB position players are spread out way on the far right tail of the talent distribution curve so that any decent statistic identities them. For example, when I sort by career runs produced (R+RBI-HR), a statistic few statheads would endorse, the top 7 batters all are listed in your top 10 WAR list. It's not a very persuasive argument.

Eventually, looking at WAR and other advanced statistics will gain more mainstream adherents, but it'll be a very long, gradual process.

A few points, Detroit Michael:

1) WAR incorporates defense.

2) WAR assigns appropriate run values to batting events like walks, stolen bases, caught stealing, GDP's, etc.

3) A first pass WAR look at Chet Lemon and Jim Rice tells you they're darn near identical. That's not interesting? A first pass look at WAR tells you Dwight Evans was a way better player than Rice, even though Rice is in the Hall and Dewey was off the ballot after three tries. That's not illuminating?

Finally, I am aware of the different WAR variants.

Thanks a lot for the comment.

Sully,

I agree with all of your points.

What I was trying to write (with poor results apparently) was that your argument that WAR must be pretty good because the top 10 or 15 batters ranked by career WAR match our expectations of the best batters is a weak argument. A garbage stat like Runs Produced that makes no adjustments for ballpark & era and ignores tons of stuff also can rank the top batters. If someone is skeptical about WAR, saying it works well for the top batters in history is not persuasive.

Great article. I wish I was as eloquent in my arguments as to why Rice (and Palmiero for that matter) are not HOF material.

Many call Ryan a dominating pitcher, and they usually refer to his 7 no-hitters or his world record strikeouts. However, the objective evidence shows that, year in and year out, Ryan did NOT dominate his peers, and was in fact inferior to Blyleven. Given that a pitcher’s goal is to give up as few runs as possible, one way to judge consistent dominance is how often a pitcher ranks among the league leaders in fewest runs given up. If Ryan were truly dominant, then he would consistently finish in the top 10 in his league in ERA+. We're not talking leading the league, just Top 10. Let's see what the objective evidence says.

Ryan finished top 10 in his league in ERA+ a mere 6 times, or only about every 3 years or so. In other words, in 2 out of every 3 years, AT LEAST TEN pitchers had a better ERA than Ryan. Compare that with Bert Blyleven’s 12 times finishing in the top 10 in ERA+.

Ryan and Blyleven pitched in the same league in 8 seasons. In those 8 years, BLYLEVEN HAD THE BETTER ERA 6 OUT OF 8 TIMES. So, when comparing apples to apples, Blyleven was a better pitcher than Ryan 75% of the time, even with Blyleven pitching in more hitters’ parks.

1973 is a great example that debunks the Ryan “dominance” myth. Most Ryan fans remember from that year Ryan’s all-time record 383 strikeouts. However, Blyleven was better than Ryan in Ryan’s most “dominant” season. Despite setting the record in Ks, Ryan had a far worse ERA than Blyeven (2.87 to 2.52). Why? Ryan fans don’t mention his 162 walks (4 per start!) compared to 67 for Blyleven. Walks kill. So although it looked like Ryan dominated, Blyleven actually dominated more in what matters: giving up fewer runs. The same thing happened the next year. Ryan led in Ks but also he also led in BBs with 202! (that’s 5 walks a start!), and thus again gave up more runs than Blyleven. In fact, year after year, Ryan struck out more than Blyleven, and gave up fewer hits than Blyleven, but gave up more runs than Blyleven. Why? WAY MORE WALKS. So, while Ryan “should have been” one of the best pitchers ever, he was NOT even one of the top 10 in his own league in ERA in most years. Walks kill.

Because Ryan walked tons of batters, while Blyleven was a control artist, Blyleven outpitched Ryan year after year.

More Statborg propaganda--Blyleven, a compiler if there ever is one, Mr. 12-5/14-9 lifetime, is your idea of a Hall of Famer. Pay no attention to that guy w/ the 20 win seasons and the 250-odd wins, we can't mention him in the same breathe. His K-W ratio just isn't as pretty as Burt's.

CHET LEMON? Chet Lemon was to Jim Rice what Johnny Damon is to Albert Belle. PLEASE get a grip here. Did you actually see any of these guys play?

A very good article. My issues:
1) When using cummulative WAR, it favors players that had long careers, even those that 'hung on' with mediocre numbers their last years. Maybe we could also include something like WAR/500 AB.

2) Since Defensive stats are very imperfect, especially looking at older players, maybe instead of WAR we could use oWAR and dWAR.
Also, why not include career ERA+, OPPA and OPS+ and wOBA along with WAR, when looking at HOF chances? I know D in not included, but again, D stats vary greatly.

Great idea. Tim Raines vs Tony Gwynn is entertaining - as is Alomar & Larkin. Thx!