That’s an interesting thing to play around with but just strikes me as being a numerical toy. It effectively pretends that the difference between Rice’s best year and Williams’s best year is quite a bit greater than the difference between their fifth-best years, which doesn’t strike me as being meaningful or informative.
Whether Williams was or wasn’t a great Hall of Fame choice, he was a pretty substantially better player than Jim Rice. Being able to maintain a high level of production for more than five years means a lot.
Maybe so, but I’m sure I can find cases of greater peak coupled with shorter career (and thus fewer career WS) whom you might rank higher than Rice in comparison in Williams. The best casse I’m familiar with Koufax as compared to, say, Suttton–what I’d be looking to do is quantify that superiority, which most observers share. IOW, since Sutton had far more WS lifetime over Koufax, but Koufax was elected to the HOF easily (Sutton, not so much) and every observer with two eyes would clearly rank Koufax higher, how might we describe this in a formula? Or, to take other teammates playing the same position, Dale Murphy and Brett Butler both had very similar numbers of career WS (294/295) but Butler peaked at 27 while Murphy in a much shorter career got into the 30s several times and won multiple MVP awards. I think we’re all agreed that Murphy deserves consideration for the Hall and Butler less so. This would be a way to quantify that sentiment.
I can’t quite dismiss it that easily, especially if the samples are groups. Imagine a 20 year career broken down into 4 year groups, then these groups ranked according to productivity, like this-
I’m thinking this is generally how it would break down (an occasional season misplaced due to injury aside). Then give this guy WinShares (averaged for each year sample)
4th-24
2nd-30
1st-32
3rd-26
5th-18
Total=130 (remember this is only 1/4 actual WS’s)
Then you weight the 2 best slightly upward, the middle as neutral, and the 2 worst slightly downward, using the notion that peak performance is 10% more valuable than average performance, and that tailing off years are 10% less valuable.
Then you take a player with the same total, but averaged the same number of WinShares (26) every year.
[261.10]+[261.05]+[261.00]+[260.95]+[26*0.90]=130
Multiply them both by 4 to get Peak-Weighted Winshares-526.8 vs. 520.
So the guy with the defined peak is rewarded a shade for concentrated performance, while the perfectly consistent guy isn’t himself penalized. (I guess he is penalized via omission, but I think everyone subconsciously weights peak performance in this way to some degree.)
It is a mathematical construct, but properly grouped and weight, it could actually be a fairly accurate quantification of people’s general bias toward high-peak players vs. very consistent players. Note that I am not saying it accurately reflects the merits of the respective players.
ETA-PRR, I wasn’t throwing off on your explanation, just saying that as I understand it right now, multiplying WinShares 3X shows a level of caprice on James’ part that I don’t really like to see in people constructing algorithm’s with subjective constants. I’ll get the book and inform myself.
I’ m not that skilled numerically to follow you very far, but essentially James is saying that the 3X stuff derives from a lack of meaningful distinctions if he just used a 1=1 W/WS ratio. That is, there would significant distinctions between some players with a identical WS number, so to get that extra precision (and still use whole integers) he made up this 3X stuff. If he’d gone to a 10X system (as some have urged) then the opposite problem would occur–it might look as if a single Win Share were meaningful, when his formulas aren’t quite that precise, so it would imply a false level of confidence in his precision. He explains better than I do, in the book, though the formula takes a few dozen pages to explain. If you understand it any better than I, maybe you could tell me what some of the technical stuff is all about. It’s capricious, but it makes a certain amount of horse sense to me.
The other advantage to the “three shares per win” is that the numbers are just nice to work with. Win shares as a stat tracks nicely with the standards (or what used to be the standards) for home runs - 12-15 is about normal for a regular player, 20 is very good, 30 is MVP territory, 35 you might lead the league, 40 you probably will, 50 a historic year, and the modern record is 59 (Honus Wagner in 1908.) 3 is also a good, wholesome baseball number.
You could switch is to 5 per win, 2 per win, anything you wanted, by just multiplying the number. Wouldn’t make any difference in relative terms.
I sort of understand his hedging against imprecision, but resorting to multiplication and limiting it to integers (in a sport whose followers are veritably steeped in the meaningful difference 1/100th makes) just seems a very odd choice to address this problem. Why not just give the numbers and a disclaimer, like ‘plus or minus .1666667, or 1/6th of an actual win’ (which seems to the level of precision he claims)
Okay, having read my own disclaimer, I sort of see the value of the 3x approach. I still don’t like it though.
And RickJay, that is an interesting correlation…do you know if it was part of the design or just a bit of lucky happenstance?
And PRR I got carried away with myself there, but I was basically saying weight someone’s best 4 years by +10%, second best 4yrs by +5%, average 4yrs +0%, second least 4yrs -5%, least 4yrs -10%.
Probably not very useful as far as ranking players, and you run into problems pretty quickly trying to scale up, but it might reflect people’s biases toward the high-peak guys if done correctly. A quantification of opinion, if you will.
Let me stress that this is mostly for my idea of the HoF, not a general ranking of players. The unit that gets overlooked in HoF rankings, in my opinion, is the season. Too many ballplayers, especially in an age of extended careers and 31 teams, play for five or ten years after their predecessors would have retired, which is fine, except that their stats get skewed. A .280 hitter who might have retired at age 32 with 2200 hits in 1958, is retiring at age 38 with 3000 hits fifty years later, and is presenting a case for the HoF with no real improvement other than the extra six years of 133 hits per seasons, i.e. mediocre play. If we were to emphasize a player’s 10 best years, both players would be more nearly equal, and those 10 best years in both cases would be what the player owes his fame to, not for accumulating stats because he has a reasonable long-term contract and there are so many teams who need some extra bats around.
The season is the unit, not the career, a tight finite period in which players compete for the pennant. You don’t really want to win 85 games every year and never make the post-season–every player should rather win 90 some years and make the playoffs and 80 other years and not make the playoffs. Individually, players should want to have successful years, and not just content themselves with padding their lifetime stats a little at a time. You should be able to say of any HoFer “He was the best [second-baseman, basestealer, leadoff hitter] in the game from Season X to Season Z,” and not just say “I never noticed how good he was when he was playing but because he played so long and so durably, he’s got better numbers than anyone despite never having excelled in any one year.”
To be specific, let’s look at players like Maris and Mattingly–they’re borderline HoFers, and probably will never get in, deservedly, but we want to consider their cases because they were world-class excellent ballplayers but just not for quite long enough. With another spectacular year or two, we might have to put such players in the Hall–because we recognized when they were playing that they were something special. But it’s absurd that another ballplayer who accumulated the same number of win shares in a longer career even be seriously considered. Chet Lemon, who retired with more WS than Mattingly, or Bill Buckner, who retired with more WS than Maris, are not even worth considering for the HoF, because they lack the peak seasons that Mattingly and Maris briefly put up. We look at Maris’ '61 and Mattingly’s '86 and we go “A HoF year for sure, that’s the kind of player we want in the Hall–but did he have enough of those years where he might have been the best player in baseball? No, fraid not.” When you look back at Lemon or Buckner, do you have an equivalent year anywhere? No, you just have a long career with impressive lifetime stats.
Isn’t this the wrong approach? Sounds more like you’re trying to find a way to prove your opinions are correct rather than trying to find an honest measure.
Incensed, to answer your question first, probably happenstance. At a glance, one Win Share’s worth about three runs, give or take. I think that’s a pretty good interval, personally. When the folks at BP get into thinks like saying Player A has 54.2 VORP and Player B has 53.1, there’s just no way you can tell me that the measure is accurate enough to say A is materially better than B. It’s a ludicrously tiny difference that could be entirely negated by the most pedestrian of unmeasured acts. Three shares per win is, I think, a pretty good midrange - large enough to constitute a possible difference in quality, but not so big as to be too clumsy.
I’m scratching my head over where the problem lies. The thing is that there just aren’t any players like what you’re describing. I can’t think of a single example among position players of a player who was just pretty good, then held on as an average player long enough to get the counting stats, and then was inducted into the Hall of Fame. There are a few long-career questionables, like Rabbit Maranville or Bobby Wallace, but almost all of them were inducted based on defensive prowess, which isn’t easily measurable, and even most of those were inducted back before there was a lot of objective study of this sort of thing. I can think of one pitcher, Sutton.
The great, great, GREAT majority of questionable Hall of Fame selections are the exact opposite; players with relatively short careers whose peak value has been exaggerrated by either statistical illusions (Hack Wilson, George Kelly, Chick Hafey, Freddie Lindstrom) or sheer stupidity (Rick Ferrell.) Players who play a long time and are just pretty good generally don’t get much HoF support. Al Oliver didn’t last long on the ballot.
I mean, if you want everyone to agree Chet Lemon’s not a Hall of Famer, I think we all agree, but so what? Nobody’s saying he is. Where’s the issue?
I agree with Yookeroo’s point that you seem to have arrived at your conclusion and are now trying to devise a way to statistically represent what you’re already decided, which is going to simply produce an arbitrary number constructed to confirm your own opinions. In which case, knock yourself out; there’s no argument against any metric you construct, unless it’s mathematically illogical. That’s not really a wise use of statistical evidence, and to be honest, I don’t precisely understand what the point of it is. I find it puzzling you want to do it. Who is arguing for the inclusion of Vada Pinson in the Hall of Fame?
Besides, didn’t Bill James do this already, in the Historical Abstract? He had a variety of ways of rating players according to a mix of career and peak value.
Just trying to quantify an explanation of why Mattingly’s 263 is actually superior to Chet Lemon’s 265. Win Shares looks illogical if you’re going to argue, as I might, that Donnie B. deserves HoF considereation but Lemon is a non-starter, unlesss you find some across the board way to account for peak performances.
That’s silly, and you couldn’t be more wrong. I would actually argue against Mattingly, Murphy, and most of the players I’ve listed as being elevated by privileging peak performances. I’m not even sure that it does enough for Koufax, who I would argue is deserving of his HoF status, but it’s a method to apply to ALL MLB players, the ones I would argue for and the ones I would argue against.
I’ve heard it, and it’s not a bad argument. Vada had some terrific years, not enough of them, and did have a pretty long career, not quite long enough. His (early and brief) peak was very impressive: over 30 WS twice, with several others in the mid twenties. On what basis do you make a case for Jim Rice (282 WS) that excludes Pinson (321)? If he’d been able to tack on a few more year mediocre years at the end of his career (he retired at 36) he would have well over 3000 hits and no one could keep him out of the Hall. Is this so important, that someone (with a long-term contract) lingers a slow death, gets some at-bats (in a DH league) and compiles a hundred hits per year until the contract expires, so we put him in the HoF? And why not discuss this issue BEFORE you get a prominent example IRL?
So your point is what? James tried to invent such a system, which is fine, but when I offer some refinements or variations on James’ system, I’m a self-justifying asshole who needs to be ignored?
There are problems with James’ system, mainly that it’s not very systematic. In the Historical Abstract, James was focusing on being entertaining at the expense of any system, hence the subjective points he awarded Williams and anyone else who he felt at that moment needed some extra points, and to his credit he pretty much says so. He’s still working on it, and so am I. Sorry for breathing.
My problem with your basic premise is two-fold. One I find Win Shares to be a bit of a junk stat. Equally talented teams can win 94 games or 84 games just based on random chance. The players on the latter team will come out uniformally worse as compared to the former, despite not having any less talent.
Now, you might say that we want to measure value and not ability. A player who’s hits happen to come at key times, even if it is just random chance, produced more value to there team, and should have a better case towards Cooperstown. Such an argument would not be without merit. However, Win Shares doesn’t measure that. Win Shares doesn’t care if you hit a home-run up 10-2 or a walk off game winning shot. It measures how well a player happened to play and how many games a team happened to win. It does not measure how helpful a player was towards winning.
Win Shares is meant to answer an interesting question, but it doesn’t answer it. For that you would need to get into win probability analysis, something that we probably won’t have historical numbers of for a while. Win Shares is a fun toy, but it doesn’t really tell us anything of note, and I don’t think should be used to argue the merits of a hall of fame case.
My other major problem with your approach is that your total dismissal of a non-peak value. If I asked a Bulls fan if they would rather had a player who was 60% as valuable as Jordan, but played for twice as long, they would turn me down in a heart beat. That teams would probably consistently been knocked out in the first or second round of the playoffs, and never won a championship.
Baseball isn’t basketball. In the modern times the key isn’t the five wins between 95 and 100, rather it is the 5 wins between 85 and 90. The Cubs wouldn’t had much better odds if they were slightly more talented. The Phillies however, with slightly less talented, wouldn’t have made the playoffs. You don’t get any extra bonus for being great. You just need to be good enough to make the playoffs. A team that alternates 101 and 81 wins a year will win less championships then won that wins 91 games every year.
You can look at this couple ways. One is to look at World Series teams and see how many hall of fame seasons they had. The Phillies had a couple, but were by no means a great team. The Rays didn’t have a legitimate MVP candidate. Go back a year and the Rockies are another prime example of just get in and take your chances. No likely hall of famers on that squad. The Red Sox did have some candidates like Manny, Ortiz, and Schilling. Know what they have in common? All of them were not nearly as good as they were at there peak, but still provided enough value for the Red Sox to win a championship.
Another way to approach this, if your thesis was correct, would be take a look at historic player seasons. How many championships has Bonds won? Griffey? Arod? How many all time great seasons have they had, and how many have led to a title? In a game with 25 players per team, and one that everyone has to take turns, even a historic season can’t determine an outcome of a team. It might in football, basketball, or hockey, but it doesn’t in baseball. There just isn’t that much added value if a players value be concentrated in a peak. You are often better off having a player’s value spread out, if you can get 85 wins worth of talent elsewhere.
This highlights the entire problem. Donnie’s 263 isn’t superior to Chet’s 265. The stat is intended to measure actual value[sup]1[/sup], not a players ability. You are trying to make the statistic do something it was never intended to do. It’s just meaningless masturbation. The fact that Lemon played longer on more winning teams is a fact, and that’s why he has more WS. Your initial presumption that more WS somehow indicates a better player is fundamentally flawed, and no amount of statistical gerrymandering to promote elite seasons will fix that problem.
1 - I have significant reservations with the stat so I don’t think it does this very well, but I think it’s safe to say that this is the goal and this is the limit to which it can be meaningfully used.
Heck, you don’t have to get too complicated; Mattingly’s superiority could be very quickly assessed by just figuring his Win Shares as a function of his playing time. Mattingly played 1785 games, while Lemon played 1988, a difference of more than ten percent. That’s actually a pretty big difference (and it still undersells Mattingly’s peak.)
How you’d factor Mattingly’s higher peak in there - well, I dunno, pick your formula. Sorry, I didn’t mean to sound jerky, I was just struggling to understand.
Anyway, I don’t support the “Multiply the best year by 5, second best by 4” formula just because it just put a huge, huge weight on that one year. That would exaggerate players with one enormous fluke year, the Norm Cashes and Ken Caminitis, and understates the peak value of the guys with prolonged peaks. I think I like Incesed’s idea more, but it looks a but too LIGHT on peak emphasis.
I’d be more inclined to a Jamesean approach of weighting different numbers - allocate X percent for career win shares, Y for best season, Z for best 3 seasons, A for best five, something like that. Then you get a nice, general approach that weighs together several different definitions of peak.
James’s argument here, and I think it is correct, is that “Talent” isn’t the point. The purpose of a baseball team is to win games, and a team that wins 94 games had a better year than one that wins 84. The players who makes up that baseball team were 10 wins better than the other and so in judging their actual performance, the only logical thing to do is to allocate 10 wins’ worth of credit to them. A measurement system that were to pretend 84 wins is the same as 94 just isn’t grounded in reality, and to be honest, I think he’s right. You would of course also be quite correct to use other statistics to guess that one team might have just gotten lucky and will not continue to play that well in the future - and in fact I am going to use an example very soon - but James is rating performance, not future prospects.
James uses as one of his examples of a bad system the Linear Weights ratings of the 1915 A’s, who were one of the worst teams of all time (43-109). James notes that according to LW, the A’s should have scored 143 fewer runs than they allowed, but in fact scored 343 fewer runs. His position is that the individual ratings cannot possibly be right, or even close to being right, if the team total is off by 200 runs. That missed their actual performance by about 20 games. I don’t see how he can be wrong; if your rating system is describing a completely different team, it’s not right.
The point to Win Shares is philosophically solid; value in baseball means winning. The number of games you win is the point from which all measurement of value should begin. A 94-win team had more value than an 84-win team and those wins have to go somewhere.
That’s not to say Win Shares does not have its problems. A good example would be this year’s comparison between Albert Pujols and Lance Berkman. Look at their numbers in baseball reference and tell me who you think had more Win Shares.
Check?
Okay; Berkamn beat Pujols in WS, easy, 38 to 35. But looking at their individual numbers, it’s impossible to guess how that could happen. Pujols’s offensive numbers are much, much better, and there’s no park differential or defensive difference that could explain that. So why does Berkman get more WS?
Because the Astros scored fewer runs but won just as many games.
Weird, isn’t it? Berkman scores as better because his team was, in fact, VERY lucky, winning 86 games while scoring fewer runs than they allowed; so each “marginal run,” by the Win Shares system, must be worth more. Pujols, playing for a team that won the same number of games but deserved to win about that number of games, gets fewer Win Shares because the Cardinals weren’t lucky.
So luck does play into it, but, bizarrely, here luck gives Berkman more win shares even though his team actually didn’t win any more games than Pujols’s team did. This is unusual, but it does happen. But if you want to “Take away” win shares from Berkman, who do you give them to? It doesn’t make any sense to pretend Houston didn’t win 86 games.
It’s not a perfect system, to be sure.
Incidentally, I agree with your argument that “pretty good” seasons have a lot of value and can be used to bolster a Hall of Fame case, but I don’t think anyone’s saying they can’t - PRR isn’t advocating that Pete Reiser be elected to the Hall.
I have a very hard time accepting any stat that uses a teams wins to somehow issue merit to a single player, it’s just far too complex. Your Pujols and Berkman example highlights this clearly. It’s true that wins are the ultimate measure, but it’s foolish to think that you can meaningfully assign one players share of that. I don’t know the numbers, but I wager that the all hitting, no pitching Texas Rangers of the beginning of the decade who won 71 games or so every year have some suspect WS numbers for their hitters who were very very good. What were A-Rod and Teixiera’s WS stats for those seasons, I’m guessing they were dramatically lower than they were once they moved to teams with competent pitching staffs and ownership, what sense does that make? Their hitting stats are very similar but somehow the fact that their new teams spent twice as much money and had quality pitching increases their value somehow? The entire endeavor is flawed.
I don’t have a problem with the giving extra credit to players whose actions help win games even if said players were lucky. The problem with Win Shares is it doesn’t actually do that. Did Berkman come through in the clutch leading to the Astros over performance. I don’t know and win shares doesn’t tell me. Just because the Astros overperformed doesn’t mean each individual player overperformed equally. If a player was great in the clutch, but his teammates were terrible should he punished? Win Shares does just that.
My point is that a concentrated peak isn’t terribly important in baseball, because teams don’t need peak seasons to win a pennant. A player who’s value is more evenly distributed can be just as valuable. Thus, i favor a low peak multiplier.
They can be more valuable as far as winning pennants go because of the new economic realities. Those white-hot peak guys tend to sign $160 million contracts and cripple most teams ability to field a complete team. Steady performers can be gotten cheap and add up to a greater whole from a Team Wins perspective.
I agree that the system is very complex, but suppose you were to examine the method and felt it was reasonably accurate - would you be okay with it? Complexity carries more chance of error, but it doesn’t PROVE it’s wrong.
And after all, too much simplicity carries a chance for error, too. Remember, James has never said the Win Shares system is a perfectly accurate rating system. It is not meant to be.
No, that’s not true at all. Hitters who are legitimately good consistently post good WS numbers. When A-Rod joined the Rangers in 2001, and they went 73-89, he posted 37 Win Shares, an excellent total - only one fewer than the league leader (Jason Giambi) and one MORE than the league MVP, Ichiro Suzuki, whose team won 116 games. If the system favoured players on winning teams, you’d think the guy on the team that won 116 games would be way ahead. No starting pitcher on the Rangers, however, had more than 8 Win Shares, which is what you would expect.
My favourite team for this is the 1972 Phillies, who went 59-97, pretty terrible. Here are their team leaders:
Steve Carlton - 40
Greg Luzinski - 16
Willie Montanez - 16
Larry Bowa - 12
Everyone Else - Sucked
Carlton, who of course had a monster season, is rightfully given an assload of Win Shares (This was the last time a pitcher got 40 Win Shares in a season, by the way.) He’s not robbed by his crappy team.
Or the 1991 Orioles, who at 67-95 sucked pretty bad:
Cal Ripken - 34
Mike Devereaux - 15
A Bunch Of Bums - Less
Ripken is seen by the system as the best player in the league, even though his team blew chunks. If a player is legitimately first rate (as Ripken was) on a lousy team, he just ends up getting a really huge slice of the smaller pie.
James put a huge amount of work into ensuring good players would not be penalized for being on bad teams and his work seems to have been effective. A-Rod posted big time WS numbers throughout his run with Texas that appear, at least at a glance, to be perfectly consistent with the real quality of his play. He was a great player on a cruddy team.
Ten wins is a LOT of wins–is it possible that a 94-win team could be worse than an 84-win team? Well, lots of things are possible, but this is on the order of you getting run over the next time you step outside your door. Not the way to bet.
.
It;s as close to meritless as you can get, sheer bullshit that you hear bruited about by people who don’t want to troubled to look into it. Clutch hitting either doesn;'t exist at all, or is so small a factor we can safely attribute it to chance and not ability. If it’s luck, it should even out over the course of a career, which is the length we’re conisdering here.
I['m not arguing that HOF voting should be based on WS. I’m trying to find a formula that would make Win Shares more useful in looking at HoF voting. If you want to argue that a player with fewer WS gets into the Hall above one with more Win Shares, which happens sometimes, then you’ve got to argue particulars–I think the largest particular is usually peak value.
You’re kind of all over the map here. I don’t totally dismiss non-peak. I DO think that it muddies up the waters of some HoF arguments, and that it shouldn’t play a big part in those arguments, but obviously every positive thing a player does in his career has value. You need light-hitting backup catchers with mediocre fielding ability on championship teams–but you would never (or should never) make the case that a year that someone caught 45 games and batted a weak .240 for a championship club is some kind of qualification for a HoF consideration, though his championship club teammates probably were glad to have him aboard. He gets his 4 or 5 Win Shares for that performance but IMO it shouldn’t move him an inch closer to Cooperstown. That’s not what we’re honoring there. You can do that for an entire career, a very long career, and it gets you zero support for the HoF, as it should.
Anyway, **RickJay **is right–such arguments in principle tend to go nowhere. Find me a case of a player with lesser peak value than I like in a HoFer, and show me why you think this an HoF player.
My comment was intended to state that baseball and the outcome of a game, let alone a season, is too complex to reduce to a statistic. An individuals impact on a given game is impossibly to quantify meaningfully. A given game has a certain butterfly effect quality. I know that James’ stat is very complex and I don’t dismiss it because it’s complex as a rule, I would be interested to dissect it and understand what he’s trying to do and I think it’d be informative. My issue is that a player’s impact on a team’s season is infinitely complex and that any formula James creates is going to be a crude approximation at best.
I see what you’re saying with the fact that great players tend to be “accurately” represented but my instinct is that this is the result of a lot of creative tweaking by James. You just can’t convince me that Ripken’s worth is dependent on how good the pitching staff was. Had the O’s landed Jack Morris in that off-season they might have won 5-10 more games and Ripken’s shares would have gone up by 2 or 3. Logically that is just silly to me.
Oh, no doubt you’re right. I think pretty much everyone acknowledges that all statistical measurement of baseball is incomplete.
Win Shares aren’t meant to be a perfect, bang-on thing, they’re meant to do two things; approximate a player’s true performance, and do so in a way that eliminates statistical illusions (the era, park effects, stuff like that.) It’s meant to be a tool for studying other things. For instance, one suggestion James makes is that you could use Win Shares to figure out how well a team drafted in a particular year. You’d have to agree that while a 36-share season isn’t necessarily better than 34, it’s certainly true that if one team’s 1993 draft pumped out 1700 career Win Shares among its players and another team’s got only 700, the first team probably had a better draft. You could then move on to ask other questions - what kind of players are better draft picks? Did adding Canadians to the draft increase the talent coming out of Canada? Are draft picks more or less economical than Latin American amateur free agents? It’s a heck of a lot easier to answer those questions if you have a handy stat that quickly gives you a solid estimate of a player’s contribution.
That is not how the system works. I encourage you to read the book. Had the Orioles landed Morris and won 5 more games because Morris was better than the bozos he replaced - which actually is quite possible, as Baltimore had one crappy pitching staff - Ripken would still have had 34 Win Shares. The extra Win Shares would have gone to Jack Morris (he had 18 that year; more than any two Oriole starting pitchers combined) assuming his performance was the only difference in the alternate-universe 1991 Orioles.