The first two sections of this three-part study of All-Star roster selections and mis-selections have focused mainly on anecdotal observations of some major blunders, of which there have been no shortage over the past 80 seasons. Some of those blunders have been caused by bone-headed or hasty choices, given the scanty information available at the time; some have been due to various biases of fans, managers, or players; some have been due to how the roster-selection process is mandated (e.g., one rep from every team); some have been due to an undue emphasis on a very small sample of uncharacteristically good baseball; and several other causes. There is plenty of blame to go around, if you’re of a finger-pointing nature. (Personally, I regret having only ten fingers to point at other people with.)  But what can done to make the selection process less subject to these causes?

I’d like to propose tracking the best players in MLB on an on-going basis, something like what Bill has done with keeping a running tab on the "#1 Starting Pitcher in the Game," only ranking all the positions (and all the runners-up), including closer, set-up man, pinch-hitter, utility player, and so on.  Of course, I’m not (MarisFan—pay attention here!) proposing substituting numbers for our judgment in picking All-Star teams, but I do think that such a running tab would at least inform voters, who might otherwise be ignorant of things not apparent to the naked eye. If we could distribute a constantly changing list of the best outfielders in MLB going back a little bit beyond the beginning of the current season, for example, a voter would certainly still be free to vote for Joc Pederson (.300 batter, with a 1.096 OPS through May 1)  to start the All-Star game, but he would have to justify his choice, if only to himself, in the face of Pederson’s standing on our well-publicized chart as the 18^th-best outfielder in MLB over the last 162 games (or whatever ranking he would have had over whatever number of games our running tab would keep track of. For the record, from May 2^nd on, in Pederson’s 2015 season, in which he started the All-Star Game, striking out twice in two plate appearances, he batted .198 with a Ruthian .714 OPS. Good one, All-Star voters!)

This is perhaps the most obviously unjust part of the current system: the heavy weight given to a hot few weeks in April or May that are wildly out of whack with a player’s previous and future performance.  It’s just frankly nutty that some guy who barely belongs in MLB (and who is sometimes out of MLB, or out of a starting job, very shortly) can get chosen as the best player in the league at his position (assuming that that is the meaning of the All-Star selection process, as I am: to identify the best in each league at each position). It’s not only contra-logical, but it diminishes the honor of being selected for the All-Star team. If we were to develop a running-tab system, this injustice would occur much less frequently: Vic Davalillo, for example, would have to content himself with winning, perhaps, "The AL Player of the Month" or several "Player of the Week" awards for his excellent May of 1965, but several other AL centerfielders that year had better careers, and better 1965s, than Davalillo did.  (Paul Blair was just getting started, but Jimmie Hall, who was Davalillo’s backup, would have been the starter, unless Tom Tresh, who was left off the 1965 All-Star team entirely, had been chosen. Both Hall and Tresh had much better careers and previous seasons than Davalillo, who was in fact the AL’s starting centerfielder that July.) Not to pick on Vic, he’s just a convenient example of the phenomenon I’m trying to describe here, but he was a remarkably consistent .500 player for 16 seasons, mostly as a fourth or fifth outfielder. He had a lifetime .502 winning percentage, and exactly 16 WAR accumulated in his 16 years, never very far above or below .500. Except for those few weeks early in the 1965 season, he was never anyone’s idea of a starting centerfielder in the All-Star game.

Bill got me started on this whole approach to the All-Star game way back when, in some early Abstract where he wrote about Toby Harrah, I think, getting more All-Star support than George Brett on the basis of outhitting Brett early in some season. "Is it anyone’s opinion that Harrah is a better third-baseman than George Brett?" he asked, or words to that effect, and I think that’s the crucial question to apply, who’s a better player, not who’s hot just now. In retrospect, we can visit any year and find some mediocre player starting ahead of some established star, and while this is no terrible travesty of justice, calling for no Kevin-Costner-as-Jim-Garrison outraged cries to heaven, if we can force ourselves to make selections that we will be prouder of making in the future, why not do it?

OTOH, why not give the Vic Davalillos of this world a break? Guy plays for 16 years, why not choose him for one All-Star team in one of his better seasons, especially if the league is in a transitional period with no outstanding players at his position? (Mantle had just moved to left field, Agee and Blair weren’t quite ready.)  Is this a crime, or something? If so, who’s the victim?

This reasoning can be labelled the "It’s His Turn" logic, often applied in MVP voting, where the rationale for voting for a lesser player over a better player is that the better player has already won a previous MVP award, screwing over Stan Musial or Mickey Mantle or Willie Mays having his typical spectacular season in favor of some lesser star having the season of his life. Maybe we’ve improved in this regard, due to our heightened statistical awareness, I would argue—in that Barry Bonds has gotten 7 MVPs, while not quite being more than twice as valuable as Mantle or more than three times as valuable as Mays. Of course, the argument against this line of thought is that it’s not an "It’s His Turn Award." If someone happens to be a more deserving centerfielder than Vic Davalillo, then shouldn’t he start on that year’s All-Star team? As for a victim of this crime, if "crime" it is,  and if we had the equivalent of a "Cold Case" file for past travesties in MLB, I’d nominate Tom Tresh, who (or whose kids) is missing a trophy or knickknack or whatever the league gave for playing in the 1965 All-Star game.

Damage is also done to players who have great second halves, especially if they have them fairly consistently (by chance, more than any other factor) over a career with a short peak.  You rarely hear, in a case made for an All-Star candidate "Well, consider the second half of last season." No, that’s usually forgotten by voters, quite as if the second half of the previous year didn’t happen.  Taking my comparison of Glavine and Bottenfield from part 2, if you combine Glavine’s 3-7 record from the first part of 1999, when he wasn’t chosen for the All-Star game, with his record from after that same date in 1998, which was 12-4, his record for the whole season from June 5, 1998 to June 4, 1999 now stands at 15-11. A similar combination of Bottenfield’s entire year (different dates, but a complete second half/first half combo) turns his impressive 14-3 W-L into a less impressive 16-4. Now that we’re comparing a 16-game winner to a 15-game winner, the question becomes a little clearer, and when you compare Glavine’s entire previous career to Bottenfield’s, I think the typical voter (or the manager, actually) has to conclude, quite reasonably (and as it turns out, entirely correctly) "Well, let’s see if Bottenfield can actually sustain this kind of pitching or if it’s just a flukey year." It’s certainly not as tough a case to make as comparing Glavine’s poor start in 1999  alone to Bottenfield’s hot start—if we make the playing field a little closer to level, and we account for the entire year since the last All-Star game, I think we choose the rosters a little differently and a little better. This could be accomplished, not through cherry picking dates, as my example does, so much as just keeping a running tab on who the best pitchers are at the time we’re choosing players. I suspect Glavine would have been much higher than Bottenfield on a running tab whenever the 1999 All-Star team was chosen. If you still want to choose Bottenfield, you can, of course, but you’ve got to do it in the face of what the numbers say.

And since Bill actually does keep a running tab on starting pitchers, we’re pretty much there. I’d like to compare next year’s All-Star team on the basis of Bill’s running tab. I’m suggesting that if we were to keep such tabs at every position, we’d make better choices across the board.

Another problem with choosing players the way we do is the failure to recognize players who don’t get off to a fast start, especially in a short career, where justice won’t always even up. I’m thinking of someone who has a few peak seasons of All-Star caliber play, who has better second halves than first halves, and maybe who plays for a team with some perennial All-Stars, an example of which would be Bill Hands, who had a few very good years with the Cubs, contemporary with teammate HoFers Ernie Banks, Ron Santo, Billy Williams, and Fergie Jenkins, and several lesser stars, so Hands would never benefit from needing to find some Cub to play on the All-Star team.  Hands’ peak years were 1968-1970, when he averaged 18 wins  and 220 IP a year, but with starts of 6-5, 5-6, and 9-7 (and finishes of 10-5, 15-8, and 9-8) to each of those three years, he came up a tad shy of seeming as dominant in June as he would seem by season’s end. Over a longer peak, those numbers might have evened out, or Hands might have stumbled into a year that few established NL pitchers got off to dominant starts, or something, but his peak just wasn’t long enough.  If you ask someone who followed MLB in those years "Who never made an All-Star team, Bill Hands, Grant Jackson, Ron Reed or Woody Fryman?," however, I think you’d win yourself a bar bet or three. (Jackson, Reed and Fryman all made the squad instead of Hands between 1968 and 1970.)

An even better example of poor luck might be John Tudor, who won more than 13 games only once, in 1985, when he went 21-8. Unfortunately, he achieved this mark after going 1-7 in April and May, so you can fugeddabout him making that year’s All-Star squad, and the next season, when you might expect people would remember his previous year’s last four months, Tudor started the year at a measly 6-5 at the All-Star break, so fugeddabout that too. If voters were at least given a list like Bill’s of "Current Top Pitchers" you can be pretty sure Tudor would have ranked pretty high on that list, and probably gotten more recognition of his actual ability than he got from people who were looking at stats from only the first few weeks of 1986.

Tudor and Hands make their names onto a few lists of players with decent lifetime WAR records who never made an All-Star team, which I’ll include links to below, with a few comments where I have something to say. You can certainly make a team out of these players that would beat the living hell out of the team of worst players who did make an All-star team, but everyone knows that. (Did you know that you can save 15% on your car insurance, though?)   This squad can be arranged into a pretty strong batting order, with no one playing very badly out of position:

With a starting rotation of

Danny Darwin’s WAR puts him near the top of this list, but I don’t see where he ever had a real All-Star type season. There’s nary a HoFer or an MVP on this team (though a few came close to winning an MVP):  what these players mostly have in common is a short peak, during which their talents got overlooked when the All-Star teams were being selected, probably with the thought "Well, yeah, he’s good and deserving, but this year I’m voting for Player Y or Pitcher X, and he can have his turn next year," which never came. No tragedy, but I do believe that each of these fine players would have found himself at the top, or very close, to being the best player in his league at his position if we actually kept close track of such stuff.

Cumulatively, it strikes me as obscene that these players should have zer0 All-Star game appearances while All-Star squads consistently feature one player, and sometimes 3 or 4, close to the middle ranks of players. Obviously, we don’t really care that much about who plays and does not play in this exhibition game, which would be fine with me, or at least I’d be better able to ignore the All-Star game even more than I already do, if we didn’t elevate the results of this exhibition to helping determine the outcome of the World Series, which strikes me as absurd.

Back of the rotation candidates include:

Dennis Leonard  26.1

Jim Barr           30.5

John Denny         29.5

If you’re thinking at this point, "Hey, the All-Star Game happens in July, the teams need to be chosen by June, so voting needs to happen in April and May—what do you want, that voters should just ignore the baseball that’s being played as they’re voting?" let me propose a thought experiment: Who says the All-Star game has to take place in July? The NFL’s All-Pro game takes place right after the season.  MLB could hold its All-Star game in November in some tropical climate, couldn’t it? And then we could use the entire year to decide who really played at All-Star levels, right? But the thought experiment I’m actually proposing is to hold the All-Star two weeks into the season, in mid-April. If we did that, no one would be voting for the mediocre starting pitcher who won his first two starts, would they?  And no one (well, no one sane) would be arguing that some power-starved humpty who hit 5 home runs in his first 10 games has suddenly emerged as a dominating slugger, right?  We’ve absorbed the lesson that strange things happen all the time in small samples, so why do we do allow an only slightly larger sample, two months rather than two weeks, apply to All-Star choices?  What’s the point of understanding the dangers of small sample size, if we’re determined to ignore it just because it’s traditional to ignore it?

So that’s my positive, affirmative recommendation: let’s elect an All-Star team based on a more meaningful measure of games than the last month or two, and let’s get rid of the requirement that every franchise have at least one All-Star regardless of the level of that team’s actual players. Both of these ideas are feasible, and would lead to more, you know, stars playing in the All-Star games.