If you’ve ever solved a 15 puzzle or its more visual variations, you’ll understand how we arrived at the decision to enshrine Jeff Kent in the Hall of Miller and Eric. I thought for sure that we wouldn’t take Kent, but we had to balance performance, position, era, and his competition while also looking toward the future.
• Performance: Kent may well not have been the best player available by performance; he might not have been the fifth best.
• Position: Kent played second base, and there will be no more second baseman of note eligible until at least 2021 (unless Chase Utley plays next year, then it’s 2022).
• Era: Kent fits nicely chronologically as we ensure that modern players get a fair shake
• Competition at Kent’s Position: At second base, his main competition was Cupid Childs who appears to have been a superior performer, but who played during an overstuffed time
• Competition at Other Positions: Sam Rice played a position that was a little overbooked and his era underbooked. Heinie Groh played an underserved position but one rich in soon-to-be eligible players; he also played in an underserved era, but we’re missing a lot of important info from that time (same for Rice). George Gore has similar things going on.
In other words, there’s a lot of ins and outs here. Instead of going into all the decisions that led to Jeff Kent, HoMEr, I want to use the situation to talk about the nature of this process as it relates to statistical inputs.
WAR (Worried About Regrets)
If you do a process like this, you will feel bronzer’s remorse—the sense that you should’ve honored that other guy instead. But it’s not like it’s all that clear to begin with! Here’s how the last few players (including two HoMErs) looked after I customized their BBREF WAR with my own set of adjustments to create my eqWAR. The table includes their career eqWAR, their CHEWS (just like JAWS with my eqWAR instead of BBREF’s WAR and slightly more peak-oriented weighting, and then each of their seasons from most to least valuable.
NAME A B C D E F G eqWAR 56.2 57.4 61.9 56.1 54.7 58.4 54.4 CHEWS 49.9 48.9 47.7 47.0 46.6 46.5 46.4 =============================================== 1 8.3 7.4 5.9 6.5 7.5 7.2 7.4 2 7.2 6.7 5.4 6.1 6.5 7.2 6.1 3 7.2 6.0 5.2 5.9 5.8 5.2 6.1 4 6.2 5.8 5.1 5.8 5.6 4.9 5.9 5 5.8 5.7 4.9 5.2 5.4 4.4 5.8 6 5.5 5.6 4.9 5.1 4.7 3.9 4.7 7 4.5 4.6 4.6 5.0 4.5 3.9 3.7 8 3.5 3.1 4.6 5.0 4.5 3.7 3.7 9 3.4 3.1 3.7 4.7 3.4 3.3 3.6 10 2.0 2.9 3.6 3.3 3.0 2.9 3.4 11 1.8 2.6 3.3 2.2 2.8 2.7 2.0 12 0.8 2.1 3.1 1.1 0.7 2.3 1.7 13 -0.1 1.7 2.1 1.0 0.6 2.3 0.3 14 -0.1 2.0 0.8 0.3 2.0 0.0 15 1.5 -0.3 0.0 1.3 -0.1 16 1.3 -1.2 -0.7 0.9 -0.1 17 0.3 0.2 18 0.3 19 0.3 20 0.0
Those guys, the whole lot of them, are awfully close together. The outlier is Mr. C, who has the flattest-shaped career. Mr. A is clearly our peakmeister, and the rest fall somewhere in between. If you knew nothing else about them, you could draw names from a hat and feel good about it. Or maybe your own predilections in the old peak vs. career debate would lead you toward one fellow or another.
When you reach a place like this, you still have to make a decision, even though all the players appear to be roughly equal. You can no longer trust the uberstats by themselves. This is where using them has gotten you. Even so, most of the uberstat makers will tell you that WAR is accurate within 0.5 to 1.0 wins, and that the further back in history we go, when we have less data, the less likely we are to have assessed the player accurately. It’s the best we can do.
An assumption that we probably all make is that the directionality of any errors within a single player’s career is mixed enough that the final evaluation is very close to accurate.
But what if it isn’t?
Let’s say that Mr. A’s metrics were all off by a win in his top ten seasons, overstating his value. Similarly that Mr. C’s were off by a run in the other direction in his top ten seasons.
NAME A C eqWAR 47.9 71.9 CHEWS 42.3 56.1 ================= 1 7.3 6.9 2 6.2 6.4 3 6.2 6.2 4 6.2 6.1 5 4.8 5.9 6 4.5 5.9 7 3.5 5.6 8 2.5 5.6 9 2.4 4.7 10 1.8 4.6 11 1.0 3.3 12 0.8 3.1 13 -0.1 2.1 14 2.0 15 1.5 16 1.3 17 0.3 18 0.3 19 0.3 20 0.0
Mr. C is suddenly an obvious HoMEr, and Mr. A is about as good as Tony Lazzeri or Johnny Evers. This is an extreme scenario, of course, but just three WAR one way or the other, in the right seasons, could be huge for one of our six examples above. Give Mr. F three more WAR, one each in seasons 1 to 3, and he’s got Mr. A’s original peak combined with Mr. C’s career value. It doesn’t take a lot. Case in point: Bobby Doerr. I mentioned the other day about how Bobby Doerr hit into a lot of double plays. Without taking that into account at all, Doerr shows up in my system as the missing quadruplet for Messers E, F, and G. But we’re missing two years of Doerr’s DP data, and BBREF only computes Rdp data for his last three seasons, which means that he’s missing runs in his WAR calculation. If Doerr’s got another -15 to -20 runs of uncalculated rDP value lurking in his record, he’s falling behind all of the guys above. That’s why we cast him aside. Maybe when that data is finally accounted for, we’ll have a clearer picture. But you can see that when you reach the point of needing to select one or two players among six equals, this kind of thing spells doom for a player.
The Big Reveal
So who are those six guys?
A. Cupid Childs
B. George Gore
C. Sam Rice
D. Bernie Williams
E. Sal Bando
F. Jeff Kent
G. Heinie Groh
We invited Kent and Bando to the dance and passed on the rest. You may have noticed that Bando was the only fellow who didn’t get any bolded type above. None of his nth-best seasons were the best among this group. But within the context of the AL of the 1960s and 1970s, he ranked higher against his peers than Childs, our peakiest example, did in the NL of the 1890s. These are the kinds of evaluations that become increasingly important as the field narrows. It’s why we have our SABERHAGEN list of questions.
You might ask us, Why not Bernie or Gore when we don’t have enough centerfielders? Because Griffey, Edmonds, and Andruw Jones are reaching the ballot soon. All of them are superior centerfield candidates. Why not this guy or that? Because the weight of thousands of small decisions led to an outcome that makes the most sense to us.
That’s not to say that Childs, Gore, Rice, Williams, and Groh are forsaken and doomed. As more information becomes available to us through Retrosheet play-by-play data (and we truly hope they will release a lot more ASAP), we will be able to better evaluate earlier players, and our view of them may change to their benefit or that of their competitors. Or if the Hall of Fame goes on an election boondoggle and sucks up a whole bunch of VC selections, we too may find ourselves dipping into history for a few more names. But for now, they’ll just have to wait, while Jeff Kent gets to be the first among several equals.