Unfortunately, the foundation post for the thread is in error.
The combined interleague winning percentages and totals through 2010 are:
1806 - 1652 = .522 (AL) -- .478 (NL)
There's a humongous difference between a .378 and a .478 winning percentage.
In point of fact, the NL has had only two seasons where its winning % was less than .450 - (2006 at .389 and 2008 at .409).
However, from 2000 - 2009, the Pirates had a 71-110 interleague record, (.392 winning %), lending credence to the notion that Pittsburgh is a AAA team. San Diego was next worst at .421 - but improved a little on that this year. As a comparison, Baltimore has posted a .441 winning percentage over the decade.
But, if you're going to build an argument based on a .387 winning percentage that is actually a .478 winning percentage, the argument kind of falls apart.
=== Dr D axed, Dept. ===
SABRMatt opined that the NL was "the game's elite minor league" and we tossed in a 9-foot-arc'er:
Real quick Matty, with AL being 100 on the index, where would you personally ballpark the NL, NPB, and AAA at the moment?
To which Matt replied,
Per the last decade of interleague play results, the NL is Pythagenmatting it at a .387 clip. A .387!! clip. For a whole league in ten years worth of interleague games. That's a LOT of games to be THAT bad.
If you called up the five best PCL teams and had 'em run shotgun with the AL for 162 games each...would they play .387 ball? It would be close.
I liken the gap between a major and minor league to the gap between a grandmaster and a master in the ELO ratings. A Grandmaster's 2400 will beat a master's 2200 about 2/3 of the time. If the NL is Pythagenmatting barely better than I'd expect the PCL to do...then yes...I call the NL a minor league and don't look back.
To answer Doc's question above...I would rate the leagues thusly:
AL: 100
NL: 92
NPB: 90
PCL: 90
IL: 90
Although a lot of stars have come out of the NPB, it's not significantly better than today's AAA if only because there does not exist a very deep player pool who can play with team Japan. Japan's best players can play with anyone...so could the best players in the Negro Leagues...unfortunately, the Negro Leagues were also minor leagues based on the kinds of statistics the stars routinely posted as compared to what they did in the big leagues when allowed to play.
Extreme, of course, and I'm not sure how much Sandy is going to appreciate the logic :- ) but we did know that Matt would swing away with some PythagenMatt facts.
Taro had my reaction: that even if the gap between the NL and NPB was as big as that, you'd think that AAA would index 80 or 85 or something. No? Taro sez:
You have to consider that a group of AAA stars would typically be a replacement level team at the MLB level. A group of NPB stars would lead the AL in Ws and a group of NL stars would break the MLB record for Ws in the AL.
The NL is significantly better than AAA and clearly better than NPB. NPB is also clearly far superior to AAA. Heck, I think the gap between the AL and AAA should be larger as well.
Matt produces five solid points, arguing that a PCL All-Star team would go .500 in the NL, and that Japanese baseball consists of a few stars and a bunch of weaker players:
A) I don't think a team of the best PCL players would be replacement level, no. I think you'd have a mix of up and coming studs and AAAA talents...I think that team might easily play .,500 ball in the NL...hence why I don't think the NL is that far above AAA.
B) And I know with pretty high certainty that the NL is enough weaker than the AL that the NL plays replacement level baseball against the AL. Internalize that for a moment.
C) And I also know that the conversion factor from NPB is typically worse than it is for the NL...so I know that the NPB, as a league would be worse than replacement level against the AL and probably about as good as the NL is against the AL...against the NL.
D) The difference between the NPB (and the Negro Leagues) and AAA...is that the NPB stars are significantly better than most of the rest of the players in that league. There isn't talent parity there...it's 30-40 really good players and then a whole bunch of really bad ones. The same is true of the Venezuelan Winter League, the Negro Leagues, Cuban Baseball etc. There's a reason that fringe MLB players can go to Japan and OPS 1.000 without breaking a sweat. The NL is probably better than the NPB since there is more talent parity there...but no, I don't think a team of NPB stars would break the MLB wins record...I think they would be competitive with the Phillies.
E) And I'm also fairly certain that so far, the NPB has produced ONE hall of fame level talent that stuck in the big leagues. Ichiro. No one else has come close. Not Godzilla, not Nomo, not Sasaki, Look what happened to Kosuke Fukudome...look what happened to Daisuke Matsuzaka...how about Kazuo Matsui...these were all players on the Japanese all star team, taro. EASILY!! And they all became run of the mill big leagues...average at best.
Taro points out that MLB vacuums up AAA players who could compete with them, and notes that the NL All-Star team would win 120 games in the AL, or something...
A AAA star team would consist of guys barely above replacement level. AAA is a farm system for the MLB and anybody good enough to be in the MLB usually is.
An NL star team would dismantle the AL. Theres no way its closer to AAA than MLB. If you were to theoretically NUKE NL Central and spread that talent over the rest of the NL, the AL-NL disparity would essentially vanish.
And in this comment, as well as this one, Taro noted that NPB stars didn't, as a rule, lose that much in coming over to MLB. Not after you "adjust" for the fact that they were 30-something players (who needed a year or so to learn new pitchers -- and then were sliding downhill physically).
Matt has a sophisticated reply to Taro's "MLB vacuums AAA" logic:
First of all, you say that 3 WAR players don't stay in AAA, and of course this is true, but AT ANY GIVEN TIME there are probably DOZENS of those kinds of players down there. ...I think an all-star team from AAA at amy given time could play .500 ball in the NL.
Second of all, your 3 WAR estimate for NPB stars is IMHO optimistic. The best of Japan's best might be 3 WAR players, but second level stars are likely to be RLPs. Using that logic, I would estimate that a Japanese AS team would be wortj 48+(10*3)+10...or about 88 wins. Competitive with the Phillies as I suggested earlier.
.
=== Dr. D's Take on AAA ===
Just quickly: hitters from AAA, as a rule, need adjustment time in learning new pitchers. I wouldn't be real confident about the 9 best PCL* hitters having a good first weekend in the Bronx.
But that isn't true of AAA pitching. Give me any 10 or 12 AAA pitchers I wanted in, say, May of a given year, and I'm not sure we couldn't get a top-3 staff ERA out of them.
When you're saying that 9 AAA hitters would have, as a group, a tough first month in adapting to NL pitching ... are you commenting on the quality of AAA? Or are you just commenting on pitcher-hitter adjustments? I don't know.
I'm sure Matt, Taro, and the entire civilized world would agree that you could pick 25 AAA players from May of any year, and in 1-2 years have a tremendous MLB team.
But how do you separate out the learning, vs. the skill differential? I don't know how you'd do that.
..............
Last chart I saw, Baseball Prospectus pegged the EqA of PCL baseball at something like .258 against MLB's .270 (or some ratio like that).
.
.
Comments
Crucial point Sandy. If I remember correctly, the AL/NL really weren't seperated that much in the early 2000s. It been in the past several years.
http://www.beyondtheboxscore.com/2010/6/28/1541053/interleague-records-t...
The author makes some incorrect assumptions IMO (assuming true talent .532 W% difference, when its still probably closer to .550), but the gap between an AL and NL team is more realistically around 8 Ws. Big difference, but this is still MLB baseball being played in the NL.
Sandy...look carefully.
I said PYTHAGENMATT!! W%...not W% straight. PythagenMatt...that's game by game pythag run differentials...not W%. Please...read what I said. THank you.
The Pythag (-enport, -enmatt, -orean) projections were .378 but the on-field records .478? Across 3,500 games?
If so, that's the most interesting stat we've ever had on here :- )
That's the difference between a 61 win team and a 77 win team. Matt should justify his numbers before anyone accepts them at face value. With that many trials, I don't know why you wouldn't just use the actual results.
Okay, yes, I missed that you were discussing pythag... win%. My bad.
But, given the basic usefulness of Pythag is that over large numbers it DOES match reality - so you can identify the flukes - I'd say you've discovered an arena where it (for whatever reason), is completely useless. If, over the entirety of the data population, the real vs. projected numbers are 100 points apart, then by my reckoning you don't have a correlation at all. And, if that's the case, the subsequent analysis of relative strengh of leagues is worthless.
I am actually a little baffled now that I see the difference between W% and PythagenMatt W%...I will need to take a second look at the spreadsheet I created specifically to address the question of the NL, however, I do recall that a previous analysis showed that the AL was outscoring the NL by 1.2 R/G from 2004 to 2009...so although that's over a smaller period, that would correspond to a roughly .415 PythagenPat W%. I may be pulling the .378 PythagenMatt from the same study and not the decade study (2000-2009)...I'll confirm this evening which number it actually was. *sigh* If I quoted the wrong year range for this sample, I am officially annoyed at myself...but it does not change my argument one bit regarding the weakness of the NL, since they have played considerably worse in the last half-decade than they did in the first and the last half-decade is closer to NOW...which is kind of what's relevant.
True, the dominance was more recent, (though '06 and '08 were the two SERIOUS outlier years).
One oddity which I hadn't realized until just now - (when looking at the '01 - '09 historical results on mlb.com), is that while the AL teams have all played basically the same number of games, (Tampa is short - missing one year of interleague play), the NL actual games played are not.
Thougth 2009, Washington had played the most interleague games (231), which is oddly more than the AL max of 230 -- while the Pirates have only played 181. So, the AL isn't playing the NL - they're playing a randomly distributed sample of the NL.
Mind you - Pittsburgh playing more games certainly wouldn't do the NL any favors. So, I'm not saying this is to blame for the skew - (the AL is better at the moment - no contest). But, attempting to compute precisely how much better gets more complicated when you've got a 50 game skew in games played among the NL teams.
Of course, the AL won its fewest games since the 2004 season THIS year. The .532 AL winning % this year is still 10 points above the aggregate for the entire interleague history - but 3 of the last 4 seasons have been: .544, .546 and .532. Since the anhilation of 2006, the trend has been back toward parity.
When the NL won 54% of the games in 2003, was the AL a minor league THAT season? Because that's basically the same thing the AL has done the past two seasons.
Recently its been closer to the NL playing around .450 W% against the AL since '05. The early decade was very even though.
The seperation in talent happened about 5 years ago. And yes, there is a slight trend down. '11 will show us whether this a continuing trend or if it flattens out at around a .550W% for the AL again.
2005-2009 Study, completed January of 2009 when someone posted here regarding a fangraphs (or was it THT?) report on the disparity between leagues:
NL W%: .449
NL PythagenPat W%: .416 (there was some apparent bad luck for AL teams...not sure why it would be quite that extreme...but the RS/RA data was checked when I made this study against the report on league contrasts posted here.
NL PythagenMatt W%: .408 (not .378...the .378 was 2006...and I'm not quite as prepared to call that an outlier...I called attention to that in my notes and it stuck out in my mind as theyear the NL truly became a minor league...if only for a time).
I went back and added 2000-2004:
NL W%: .474
NL PythagenPat: .459
NL PythagenMatt: .458
That's still AWEFUL...but not quite as bad as I was incorrectly remembering.
GAH...I have about five billion Excel files with little mini experiments over here...it took me an hour to find the one on the interleague play disparity.
In the same report, BTW, I noted that the Fiato/Souder Matrix asserted that the average NL team was a full run per game per side worse than the average AL team in 2006 and 2008 and eight tenths of a run worse in 2007. If you don't recall (and I wouldn't blame you for not recalling...it's been a while since I've talked about the F/S Matrix)...this is a tool that takes all of the RS/RA data for each game in each park against each starting pitcher in each team vs. team match-up with each home plate umpire and attempts to attribute all of the runs to some linear combination (by solving the matrix) of team offensive and defensive skill, the skill of the starting pitchers, and the skewing impacts of strength of schedule, home plate umpires and ballparks. From this concept I have derived everything from linear (non-ratio) park factors that work notably better than the compeition to umpire factors to comprehensive strengths of schedule. WHen one league is a full 1 or 1.2 R/G worse than the other league in a matrix solution that is increasingly confident with increasing games played (and IL games are a small sample compared to the larger league schedule!) then something is very...very wrong with league parity.
A .450 W% would mean the NL is playing around 73-89 ball against the AL.
It basically means the average NL team is around 8 Wins worse than the average AL team.
It means that a 90 Win NL club is basically around the same thing as a .500 AL club.
Thats pretty significant, but what would an average AAA team do? .260? Worse?
...but yes...worse than the NL. In the AL, the average AAA tream would probably be a .320-.350 ballclub (replacement level or a shade worse)....that is after all the idea behind replacement level. That your typical first call-ups from AAA perform at that level and most farm teams have several "first call up" types and a good cluster of guys close to that level behind that and there are some teams (like this year's Rainiers) who have a lot more talent than that.
When last a serious lok was taken, the AAA EqA was .256 to the AL's .270...that's about .415 offensive W% skill...but the pitching is also aroud the same level, so if you double that delta to .500, you get a rough estimate of the league W% in the AL (aroud .330).
So perhaps I've got AAA too high sitting at 90. It should perhaps go:
AL - 100
NL - 93
NPB - 90
AAA - 87
VWL - 82
CUB - 78
AA - 75
I think thats a lot closer to the realistic gap, but I still feel that AAA should be lower. RL players are very good AAA players. The average AAA player isn't a RL player.
Between the five of us, we're triangulating the truth here ...
Hey, even Einstein was capable of sloppy arithmetic now and then :- )
No debit points for the 0-1 count we worked ourselves into, where would you guys index these leagues vs. the AL's 100 now?
Matt has:
AL - 100
NL - 93
NPB - 90
AAA - 87
VWL - 82
CUB - 78
AA - 75
True that the actual W/L is pretty good for the NL, but also true that the bases/runs/whatever used by Matty are showing a pretty fair gap...
Intuitively speaking, I'd have guessed something like Matt's table above, with the NL at 94-95, NPB 92, AAA 87-88 as he has it... could be wrong...
My problem wasn't sloppy arithmetic...it was sloppy record-keeping of my many little daytime experiments with MLB data...LOL Seriously...I should really index my experiments if I'm going to reference them in quick posts. Otherwise I screw up from time to time and get my own results quoted wrongly.
*sigh*