Interleague begins today: AL teams look to pad their records!
by JinAZ on May 21, 2010 10:00 AM EDT
in Positional Reviews
Email
PrintIf you like discussions like this, you'd love our newsletter. Sign up now!
Tonight, interleague play enters its 14th year. Here is a graphic I posted last year showing interleague records since play began in 1997. The line is a two-year moving average.
Something seemed to happen around 2005 or 2006 in which the AL gained a distinct advantage over the NL. It's a massive disparity. A .566 W% (the 5-year AL average) translates to a 90-72 record.
There have been a number of arguments that try to explain what has happened here. Below I summarize some of the main ideas I've seen put forth, as well as to provide some information about whether they have some validity:
1. Rule differences give AL teams an unfair advantage in interleague.
The argument here is that AL teams often employ a DH (e.g. Vladimir Guerrero for TEX) whose exclusive job it is to bash the ball. NL teams don't have this, and therefore are at a disadvantage when playing under AL rules. The counter to this argument is that if you don't have to pay a DH, you can spend more elsewhere, and thus should have better players at other positions. Furthermore, if you want to argue this, I think you have to explain why the NL teams did just fine vs. the AL teams from 1997-2004 (see graph above). As far as I'm concerned, this argument is a dead end.
2. The AL dominance is not a league-wide phenomenon, but rather is caused by on a few top AL teams beating the hell out of NL teams.
Here's a list of individual team interleague winning percentages over the past 5 years (2005-2009) for each team. Leagues are color coded.
I am honestly shocked at how clean the split is between the two leagues. This is not an instance where one or two teams are causing the entire effect. NL teams have just been flat out inferior, and AL teams are systematically better.
Here's another look at the same thing:
Ok, so at this point I think we have to accept that there's a talent disparity across the leagues. This has been supported by studies examining what happens when players change to new leagues (e.g. MGL's series). Why is the AL so much better? Below are a few ideas.
3. AL teams have higher payrolls and therefore can secure better talent.
As Jeff recently discussed, the average AL payroll at end of season in 2009 was $10 million higher than the NL ($102.5 M vs. $92.4 M) source: bizofbaseball.com's salary database). This means that they're (probably) spending more on talent, and therefore have better teams. The biggest component of this is the Yankees with their $220 M payroll at end of 2009, and if you eliminate the Yankees there is virtually no difference in average payroll across the leagues ($93 vs. $92 M). I think this is a factor. But it's not just the Yankees kicking butt in interleague (in fact, they've only been the 7th best AL team over the past 5 years), so there must be more to it that this.
4. AL teams are spending more on amateur talent and international free agents, and thus have better talent.
Matt Swartz had a nice piece on this a few weeks ago at BPro. He found that it was actually the non-market player spending (amateurs and free agents) that differed more between the two leagues, rather than free agent spending. The biggest extra spenders were in the AL East and AL Central. We often do see AL teams paying over slot bonuses (the Tigers and Yankees come to mind), and may be helping them acquire a talent advantage. This is the first I'd seen on this, but I think it culd be a big part of the explanation.
5. AL teams have better front offices than NL teams.
I think the first person to really articulate this argument was Matt Klaassen. There's some truth to it. Think of the front offices that seem to be really heads and tails above others: Boston, Tampa Bay, Yankees, Angels, Twins...most of the ones that come to my mind are in the AL. Then think of the really bad ones. Sure, there are the Royals. But there's also the Pirates (pre-Huntington, anyway), Mets, Astros, and Cubs. Rating front offices is a very subjective exercise, and you may not agree with my examples. But I bet if we did a community wide voting effort to rate front offices, the AL teams would, on average, come out ahead.
So what do you think? Are there other reasons for the disparity that I haven't considered?
126 comments
|
2 recs |
Do you like this story?
Comments
What about sentiment?
The AL is supposed to be the stronger league, so NL teams might subconsciously see interleague games as a “minimise losses” event rather than a “fill yr boots” opportunity, and vice versa.
Lack of risk
Teams in the NL don’t have to be as good as a team in the AL to make the playoffs. There is no one that is driving that side of the league like the NYY does in the AL. Because of this, NL teams can build a safer less dynamic lineup. In a dual league system that has limited interaction between both leagues, GM’s build to compete against their own league. This initially is not problematic. The problem is that it has been stagnant for so long that the NL never made the aggressive changes in the front office and scouting/player acquisition that the AL teams did.
JD’s like, "you want some fucking pitching? Here’s all the pitching you can stand. Now choke on it, bitches!"- RCCook
I like this idea, at least as an explanation for why it hasn't corrected
If you know that your competition isn’t going to push any harder than a certain level, you yourself only need to push to that level to get into playoff contention. You can then pocket the extra money/time.
I dunno, maybe. Winning doesn’t always seem to be the main goal of some franchises, so this could be true.
-j
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
Before making any conclusions, I think everybody needs to look at
KCR’s .556 record against the NL from 2005-2009.
Let’s review the Royal’s overall record from the same time period:
Year: Record :: Interleague Record
2005: 56-106 :: 9-9
2006: 62-100 :: 10-8
2007: 69-93 :: 10-8
2008: 75-87 :: 13-5
2009: 65-97 :: 8-10
Let’s be clear—this is one of the absolute worst franchises in baseball over that time perioed (if not one of the worst ever), and yet, we’re seeing them do better than split their interleague series. That’s amazing.
So whether it’s the DH (partly), or the front offices (this makes me nervous), payrolls (hello ALE), or otherwise, it’s a shocking disparity. It’s probably some factor of all of them.
Side note: Has anyone looked at AL park advantages over the NL?
See Data Differently. Beyond The Boxscore. | Follow me @justinbopp
Two Out Rally, the new BASEBALL MMORPG! | Facebook | Twitter
What are the chances that a team of the Royals talent would have a .556 win% over 90 games?
by Dan Turkenkopf on May 21, 2010 11:00 AM EDT up reply actions
Some thoughts
First, I don’t think the DH argument can be taken seriously. Again, you can spend that money elsewhere in the NL and therefore negate those losses, as long as you keep payroll constant (which it is, save for the Yankees). And you have to explain why this was a non-factor from 1997-2004.
Park advantages: I don’t see why it would matter, as you play 50% of your interleague games at home and 50% on the road. And it would have to be pretty systematic in AL vs. NL.
Royals are an interesting case. Here’s the schedule they’ve faced over the last few years:
2009: ARI:1-2,CIN:3-0,HOU:2-1,PIT:1-2,STL:1-5
2008: ARI:2-1,COL:3-0,FLA:2-1,SFG:2-1,STL:4-2
2007: COL:2-1,FLA:2-1,MIL:1-2,PHI:2-1,STL:3-3
2006: CIN:1-2,HOU:2-1,MIL:2-1,PIT:3-0,STL:2-4
2005: ARI:2-1,COL:0-3,HOU:1-2,LAD:3-0,SFG:2-1,STL:1-2
I’d have to go through and do a more systematic analysis to be sure, but it seems like a fairly average sampling of NL teams. There’s a a decent number of crappy teams that they’ve faced (PIT and/or CIN almost every year). But also some good teams: STL is almost always solid, and that’s their designated “natural rival” (though they have a losing record vs. STL, which is reassuring).
One thing to keep in mind (and just saw Dan write this): we’re dealing with a half-season worth of games here, so any one team can deviate pretty well from their true talent level. But when all the AL teams are better than all the NL teams (which is more or less the case here), the entire effect can’t be chance.
-j
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
I think it's the culmination of a lot of things
If there was imbalance in only a couple places, we wouldn’t see this kind of difference.
I think that it’s the combination of larger payrolls, superior front offices, the rules, and more spending on amateurs in the AL that makes the difference. Everything has kind of combined for this perfect storm in the AL, where you have all of these teams that are not only ran well, but are putting some serious money behind it as well.
Then again, I certainly haven’t put any serious research into that.
I like baseball.
I write for Beyond the Box Score and The Hardball Times Fantasy
by Satchel Price on May 21, 2010 11:01 AM EDT reply actions 1 recs
Theory #1
The one thing I’d love to have the numbers on, but don’t have the database skills to check (and I can’t find a site that offers the splits to run it), is what the is the W-L split in NL parks only. Granted this would potentially skewed by whatever home park advantage a team might enjoy, but I think removing the DH from the equation would offer a clearer picture.
Of course, with how useless DH’s have been league-wide, we might get an answer on this anyway.
It also is likely that these things tend to swing back and forth in all sports for whatever reason. The NFL and NBA has also experienced this for no seemingly good reason.
JinAZ linked to MGL's direct study based on linear weights
They were all done based on road games and had DH and pitcher hitting removed from the stat lines. Your answer could be in there.
Marlin Maniac, a Florida Marlins blog
Writer, Beyond the Box Score
Writer, Baseball Propsectus Fantasy Beat
Writer, Heater Magazine
Well, I wouldn’t want to remove the DH in that way necessarily. NL teams are at a clear disadvantage in AL parks, because the don’t have a top bat sitting on their bench. So just seeing how the games play out in NL parks with pitchers batting, etc, might show how the DH impacts the game.
by afrosupreme on May 21, 2010 11:17 AM EDT up reply actions
But NL teams should have better players at other positions.
If payrolls are the same—and they are, minus the Yankees—then they should be spending more on other positions. If the market properly valued players (and admittedly it probably doesn’t), this negates the DH advantage.
-j
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
Doesn't the payroll evidence ($10M less per team?)
negate the “they spend the money elsewhere” argument?
See Data Differently. Beyond The Boxscore. | Follow me @justinbopp
Two Out Rally, the new BASEBALL MMORPG! | Facebook | Twitter
by Justin Bopp on May 21, 2010 12:00 PM EDT up reply actions
Again, if you ignore the Yankees, payrolls are the same for NL vs. AL
~$93 M.
It’s not just the Yankees that are driving the W% disparity.
-j
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
Again
If you’re going to make any argument about the DH, I think you need to explain why it was a non-factor from 1997-2004. I guess you could argue that, in fact, back then the NL was superior and that negated the effect. Whereas now, maybe they’re equal, but the DH gives the AL teams the advantage…
But if that’s true, why do players moving to the AL see their stats take a hit beyond what would be expected by aging, regression, etc?
-j
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
I think you need to explain why it was a non-factor from 1997-2004
Who said it was a non-factor then, though? You could be right that it is, but simply saying the NL was better than the AL so it wasn’t a factor is a really incomplete analysis. It might have been a negative factor that the NL more than overcame. I don’t know, just saying the argument is missing a lot of relevant information.
by Missing Barry on May 21, 2010 11:19 AM EDT up reply actions
Ha, should have read your whole comment, as you DID mention that. My bad.
by Missing Barry on May 21, 2010 11:20 AM EDT up reply actions
Right
I can accept that the DH can be a factor. But there must have been some kind of change in the relative quality of the leagues around 2005 if you’re going to explain the first graph. The DH effect, whatever it might be, is constant from 1997-2009.
-j
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
Yeah, and even if it is a factor….the evidence for it when people have looks doesn’t seem to be easy to find, so I have a hard time believing it’s anything but a small factor. Definitely not a major part of the explanation for the current disparity.
by Missing Barry on May 21, 2010 12:08 PM EDT up reply actions
What if the existence of the DH changes how AL teams
develop players, structure rosters, plan in-game strategy? Maybe the pre-2004 seasons are the outliers, and AL having a ~.525 win% was the true steady-state all along.
by SagehenMacGyver47 on May 21, 2010 2:09 PM EDT up reply actions
Ok, maybe
Any specific ideas how this would affect player development(?), roster structuring (more relievers in AL and fewer bench?), and in-game strategy (isn’t that something that can be adjusted on the fly?)?
-j
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
A few things strike me right off the bat
First is just how bad two of the best NL teams, Philly and LA have been against AL opponents. The two of highest ranking NL teams are typically very good, St, Louis and Colorado, but even so STL is under .500. It is surprising how little the success in inter-league correlates with overall winning % in the NL. This leads me to think that two things are going on. First, while there is a disparity it is exaggerated by the size of the sample. Second, laxtonto is on to something. Building a team in the NL removes the pressure to regularly compete with the best teams in baseball, NYY, BOS and TBR- so you can win with less and the NL teams embrace this.
I am interested in the payroll discrepancy if you remove the top two teams in both leagues, my reason being, the Mets are the highest NL team (comparable in salary to Boston) but unlike many top AL teams they are wasting huge chunks of that money. Just a thought.
- Matt Sullivan
A good friend of mine used to say, "This is a very simple game. You throw the ball, you catch the ball, you hit the ball. Sometimes you win, sometimes you lose, sometimes it rains." Think about that for a while. - Nuke LaLoosh
Only 90 games
As Dan mentioned above, we’re only looking at ~90 games of a sample here. Teams vary dramatically from the first to second half of the season, so looking at what any one team is doing involves some significant error.
At the population level, this should balance out.
-j
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
Good point.
Counterpoint:
720 games against the AL: .404
90 games against the NL: .556
The inverse of the argument is particularly salient: not only are the Royals good against the NL, they’re bad against the AL.
See Data Differently. Beyond The Boxscore. | Follow me @justinbopp
Two Out Rally, the new BASEBALL MMORPG! | Facebook | Twitter
by Justin Bopp on May 21, 2010 11:47 AM EDT up reply actions
(this is the Royals record over the selected years)
See Data Differently. Beyond The Boxscore. | Follow me @justinbopp
Two Out Rally, the new BASEBALL MMORPG! | Facebook | Twitter
by Justin Bopp on May 21, 2010 11:51 AM EDT up reply actions
League adjustment
The power ranking league adjustment typically bumps teams by ~0.025 in winning percentage or so. So, maybe you can set the the true talent level of the Royals at 0.429 based on their AL record over that time. And you’d pull back their vs. NL record to 0.531 with this adjustment. Can you believe that a 0.429 team could play 0.531 baseball over 90 games? It’s not likely, but I can buy that.
-j
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
I'm thinking this isn't just sample size for one team
That represents more than 11% of the games they play. So for every 9 games they play, one of them is against the NL, and they’re winning more than half of them.
See Data Differently. Beyond The Boxscore. | Follow me @justinbopp
Two Out Rally, the new BASEBALL MMORPG! | Facebook | Twitter
by Justin Bopp on May 21, 2010 12:05 PM EDT up reply actions
Not following.
90 games is 90 games. ..??
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
Oh god, I can't believe we haven't thought about this.
Surely you guys recall what happened in 2006, right?
See Data Differently. Beyond The Boxscore. | Follow me @justinbopp
Two Out Rally, the new BASEBALL MMORPG! | Facebook | Twitter
The Joint Drug Prevention and Treatment Program was adopted in the Spring of 2006.
http://en.wikipedia.org/wiki/Major_League_Baseball_drug_policy
See Data Differently. Beyond The Boxscore. | Follow me @justinbopp
Two Out Rally, the new BASEBALL MMORPG! | Facebook | Twitter
by Justin Bopp on May 21, 2010 11:23 AM EDT up reply actions
Did nobody else notice
the height of the steroid era and interleague non-disparity are the exact same years?
1997-2004?
Two potential factors:
1. Perhaps PEDs helped the NL keep pace with the AL through the now-cliche, “nobody considers the pitchers were doing it too” argument.
2. Perhaps PEDs actually did help pitchers more than hitters, and when they were mostly removed, the natural (and obvious) advantage of the DH finally took root.
I dunno, I hate going back to the steroid argument constantly, but it’s the elephant in the room and can’t be ignored any time one references THE EXACT ERA.
See Data Differently. Beyond The Boxscore. | Follow me @justinbopp
Two Out Rally, the new BASEBALL MMORPG! | Facebook | Twitter
by Justin Bopp on May 21, 2010 11:29 AM EDT up reply actions
Ok, trying to work through this.
If steroids favor hitters, and AL teams are the only teams that employ a DH-worthy hitter (which might be true, although the Reds have Jonny Gomes and for a long time had Adam Dunn!), then steroids should disproportionally favor AL teams. This would lead to the opposite effect what we see—AL favored during steroids era, not after.
If steroids favor pitchers, then NL teams should have an advantage, because they dont’ have to employ a DH and therefore can invest more heavily in pitching. I guess? And therefore loss of steroids = loss of NL advantage…?
I dunno, I’m having a really hard time linking the two causally as something that would affect one league more than the other…
-j
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
Two thoughts
Why do we think 1997-2004 is the steroid era?
and
We’ve got an endpoint for interleague play, but that doesn’t mean the league superiority wasn’t shifted prior to 1997.
by Dan Turkenkopf on May 21, 2010 11:46 AM EDT up reply actions
Also, why do we think that steroids explain changes since 2006? We’re still so unsure of what their effects are, and how many people used, and what other factors might be contributing……I’d really hesitate to try to explain any changes just by pointing to steroids.
by Missing Barry on May 21, 2010 12:14 PM EDT up reply actions
Various evidence which may or may not be relevant
http://www.beyondtheboxscore.com/2009/9/21/1045928/total-mlb-home-runs-this-century
http://www.beyondtheboxscore.com/2009/9/11/1025506/total-mlb-complete-games-this
See Data Differently. Beyond The Boxscore. | Follow me @justinbopp
Two Out Rally, the new BASEBALL MMORPG! | Facebook | Twitter
by Justin Bopp on May 21, 2010 11:49 AM EDT up reply actions
Look at those compared to this:
2006 looks to be the funny year that everything changes, doesn’t it? Now, maybe this is correlation != causation (ice cream sales and elderly deaths both go up in the summer), but wow.
I’m going to look through various other counting stats to see if this is worthy of a bigger series.
See Data Differently. Beyond The Boxscore. | Follow me @justinbopp
Two Out Rally, the new BASEBALL MMORPG! | Facebook | Twitter
by Justin Bopp on May 21, 2010 11:56 AM EDT up reply actions
I just can't see the causal connection here.
I read your ideas above, but I’m not buying it. Yet, anyway.
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
Let's explore what else changed after 2003
to make that number spike so significantly. Maybe I’m blaming the ice cream for the old people dying, when I should notice the heat. Or maybe the steroids aren’t the ice cream, but the heat.
See Data Differently. Beyond The Boxscore. | Follow me @justinbopp
Two Out Rally, the new BASEBALL MMORPG! | Facebook | Twitter
by Justin Bopp on May 21, 2010 12:09 PM EDT up reply actions
Yeah, I think
that considering potential steroid use alone ignores a lot of other factors that may have played into the decrease in total offense one way or another. We don’t know how front offices changed their approaches to building their respective teams during this time period, and what effect that would’ve had on the this shift.
Also, what does the dollars per win come out to for each league? I tend to agree with what Satchel said above that amateur spending has played a huge role. If we can isolate that a handful of pre-arb players account for a disproportionate amount of the wins in the AL as compared to the NL, then that leaves the AL with a lot more overhead to spend on other key positions.
Did you see Matt Swartz's article?
Linked in the original post. It’s all about this amateur spending issue.
Also, regarding steroid era: Tango demonstrated in a THT article some years ago (I’ll grab the link if you want) that a small change in ball manufacturing, still well within the published specifications, could cause the change in home run rates that we saw during the late 90’s and early 2000’s. Didn’t prove it DID happen, but there are many other explanations for the increased run environment we saw back then.
-j
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
Steroids
So overrated for so long for much of MLB’s ills (and now known to be explainable), we want to pretend that they weren’t a significant issue at all. Now, I think, the pendulum has swung so far against the argument that they’re being underrated (and ignored).
Freaking amazing.
See Data Differently. Beyond The Boxscore. | Follow me @justinbopp
Two Out Rally, the new BASEBALL MMORPG! | Facebook | Twitter
by Justin Bopp on May 21, 2010 12:20 PM EDT up reply actions
Honestly, I think we should just refrain from calling it either overrated or underrated. We really don’t know the effect steroids have at all (or over what time period we’d see those effects), and without knowing, it’s impossible to even make a judgment whether something is “overrated” or “underrated”.
by Missing Barry on May 21, 2010 3:13 PM EDT up reply actions
I couldn't disagree more.
See Data Differently. Beyond The Boxscore. | Follow me @justinbopp
Two Out Rally, the new BASEBALL MMORPG! | Facebook | Twitter
oops, I hit post too quickly.
I was trying to say, I couldn’t disagree more—we DO know some of the effects of steroids and the evidence is becoming clearer every day.
I’m not sure why we’re afraid of talking about it.
See Data Differently. Beyond The Boxscore. | Follow me @justinbopp
Two Out Rally, the new BASEBALL MMORPG! | Facebook | Twitter
What do we know? I’m not saying don’t talk about it, I’m saying don’t jump to conclusions without evidence to back them up.
by Missing Barry on May 21, 2010 3:41 PM EDT up reply actions
How much evidence do you require?
See Data Differently. Beyond The Boxscore. | Follow me @justinbopp
Two Out Rally, the new BASEBALL MMORPG! | Facebook | Twitter
Oh, I’m willing to read anything you have, I’m not looking at this from a “you must give me this much for me to agree” point of view or anything, just interested to know what specific things you’re talking about. In my own reading up on the issue I’ve been pretty unsatisfied because we know so little to feel we can reach any meaningful conclusions, but if you know some stuff I don’t, I’m all ears!
by Missing Barry on May 21, 2010 4:10 PM EDT up reply actions
We did a two week series on steroids in my baseball class
I think I’d sum it up like this:
1. Yes, steroids clearly do enhance athlete strength and quickness in athletes. There are excellent papers in the scientific literature on this.
2. Yes, increases in strength and bat speed on the order that can be expected with steroids would increase fly ball distance enough to cause meaningful increases in home run rates. Again, good studies in the physics literature on this. You can find these at Alan Nathan’s site.
3. Careful studies of players known to have taken steroids have shown marginal if any effects at the population level. Nate Silver’s from BPro’s book is still the best I’ve seen on this.
4. There are many other alternative explanations for why we had a spike in home run rates. Such spikes are not unprecedented.
Disagreements?
-j
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
I really only take issue with the 4th.
“Such spikes are not unprecedented”
I’d love to verify (or reject) that via a compelling visual. I’ll save it for a GotD next week if you’ll let me.
See Data Differently. Beyond The Boxscore. | Follow me @justinbopp
Two Out Rally, the new BASEBALL MMORPG! | Facebook | Twitter
Go here
http://home.istar.ca/~mbein/baseball.html
Home runs per game.
“Steady” increase, as HR/G is probably a rough indication of overall league quality. But spikes and valleys all over the place. I’d guess that HR rates will continue to climb in future years.
BTW, if you do anything with this, HR/PA is probably better than HR/G.
-j
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
One of my favorite questions...
…is what the hell happened in 1987?
YR: HR/G
1983: 0.78
1984: .77
1985: .86
1986: .91
1987: 1.06 <——
1988: .76
1989: .73
1990: .79
1991: .80
1992: .72
-j
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
And one more
Alternative explanation for the spike in late ’90’s:
http://www.hardballtimes.com/main/article/changes-in-home-run-rates-during-the-retrosheet-years/
Nothing definitive, just points out that there are other explanations.
-j
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
http://steroids-and-baseball.com/
Not sure how much conclusive evidence is in there, but there’s some interesting stuff in it, and it touches on the historical power issue a bit.
by Missing Barry on May 21, 2010 4:12 PM EDT up reply actions
Interesting site
Exhaustive and hard to read, but lots of interesting stuff there. Would love to know who the heck is responsible for it.
-j
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
I believe it’s Eric Walker. He posts over at McC a little bit, worked as a statistical analyst for the Giants at one point (really long ago), and runs a random assortment of baseball sites.
by Missing Barry on May 21, 2010 4:37 PM EDT up reply actions
I don’t have any real disagreements on that, my own take from all the reading I’ve done is that steroids definitely can enhance muscle strength, give you bat speed, etc. Those are positives. (Side note: HGH does not do this, as far as the evidence I’ve seen says). That said, we don’t know who used, we don’t know the timeline for usage, either with players or in the league in general (we know players were using at least as far back into the ’70’s, but how rampant it was, who knows), and I’ve seen suggestions along what you said in #3….so, as of my knowledge right now, it just seems impossible to me to make any meaningful conclusions. We just don’t have the information we need to have any idea how much it helps or doesn’t help performance.
by Missing Barry on May 21, 2010 4:18 PM EDT up reply actions
Right, definitely increases performance
How that translates into baseball skill is hard to say, especially because both pitchers and hitters stand to gain from steroids. And the data on actual performance are, at best, marginal. And, often as not, show no effect.
And yeah, the hGH is apparently all placebo effect once your growth plates fuse. No evidence that I’ve yet seen from a systematic study showing a positive effect, but there are indications that it can actual cause weakness! It does in patients with acromegaly, which is caused by elevated hGH levels in adults.
-j
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
I don't have a BPro subscription, sadly.
I read the first handful of free paragraphs, but can’t get to the rest of it.
I’m not sure I understand his point here:
The average WARP3 of players originally drafted or signed amateurs by National League clubs over the last two years was 39.2, while the average WARP3 of players drafted by American League teams was 43.9. If anything, the turnover for players has actually moved the talent towards the National League instead of in the other direction.
I assume it has something to do with his No Turnover Standings article, but I also can’t read that.
I think the point there is...
That AL teams are drafting better, and the net flux is for those draftees to move from AL to NL more than NL to AL.
-j
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
I guess I don't see 2006 as the inflection point
You can see homers with decreases in 2002 and 2005, with a big jump back up in 2006.
And if the drug agreement were the reason why homers dropped, would that have happened immediately (i.e. in 2006), or is there some delay before the effects of steroids wear off even if they’re undetectable (serious question)?
by Dan Turkenkopf on May 21, 2010 1:07 PM EDT up reply actions
I think that's a vaild observation (and question)
Maybe we actually did see a degradation over time from the increased exposure of PEDs and steroids toward the end of the era, which is exactly what led to the new ‘rule’ in the first place? Perhaps a lot of guys were getting nervous?
See Data Differently. Beyond The Boxscore. | Follow me @justinbopp
Two Out Rally, the new BASEBALL MMORPG! | Facebook | Twitter
isn't one of the immediate benefits of steroids muscle recovery?
that would help pitchers more than batters? And built strength could be sustained by the batter for longer.
SO it would figure that the pitchers would be worse sooner and explain the spike in HRs?
"The ego, the super-ego, and the Ed" - dannycakes
One interesting note is that average fastball velocity has supposedly increased since the ban on steroids. Not sure if that’s a measurement error or a real trend, and why it would be if it was real….just something I found interesting.
by Missing Barry on May 21, 2010 3:10 PM EDT up reply actions
Throwing harder != throwing longer
See Data Differently. Beyond The Boxscore. | Follow me @justinbopp
Two Out Rally, the new BASEBALL MMORPG! | Facebook | Twitter
Well, first, if steroids build strength, maybe they add velocity? I don’t know if they do, but I’ve definitely seen that assumption made. Second, if they speed recovery time – I could see a number of ways that could affect average velocity. A good pitcher who throws harder can pitch more often as a bullpen guy, or possibly deeper into games as a starter (but then again, that might have a negative effect since coming out ot the pen adds velocity so replacing him with a reliever might actually add velocity?), or allow them to keep their arm strength without wearing down as much…..just throwing out possibilities, not actual theories, to show the relationship is plausible.
by Missing Barry on May 21, 2010 3:44 PM EDT up reply actions
how about doing AL/NL pythags for the interconference games to get
a better measure of how even/uneven the results have been?
by SagehenMacGyver47 on May 21, 2010 2:10 PM EDT up reply actions
Or better yet, component-based pythags
I’m all for it. Should be quicker to stabilize than winning percentages.
I won’t get to it today, but if anyone wants to take a crack at it be my guest. It will take some excel wizardry or simple grunt work to pull out of there, but all of the data are on baseball-reference in their head-to-head standings reports.
-j
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
Always gotta put in the subtle subconscious rub at Bonds. I see how it is……
(I was going to make a joke eariler about how Bonds was obviously the only player that used so no steroids only hurt the NL, but decided against it…..)
by Missing Barry on May 21, 2010 12:17 PM EDT up reply actions
The quality of the teams in intra-league play
seems to have little to do with their performance in inter-league play, especially for the Nationals. Strange.
Winner, Beyond the Box Score 32 Predictions Contest, 2009
Nationals = National League
not talking about the Washington team in particular.
Winner, Beyond the Box Score 32 Predictions Contest, 2009
Neat!
And weird!
And my goodness that’s hard to believe. Are the slope coefficients significant?
-j
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
Are the slope coefficients significant?
How do you calculate that?
Winner, Beyond the Box Score 32 Predictions Contest, 2009
I'd have to look it up
I usually press the button on my stat package and it tells me.
Thanks for the data, reported below.
-j
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
My guess from just looking at the scatter plot
Is that for the NL, no, and for the AL, maybe a little bit but not a lot. But I suppose you want a mathematical answer.
Winner, Beyond the Box Score 32 Predictions Contest, 2009
Here is the data
Tm Intra Inter
BAL 0.417 0.500
BOS 0.563 0.689
CHW 0.520 0.600
CLE 0.514 0.478
DET 0.490 0.678
KCR 0.385 0.556
LAA 0.581 0.633
MIN 0.515 0.678
NYY 0.593 0.567
OAK 0.506 0.478
SEA 0.456 0.589
TBR 0.457 0.511
TEX 0.492 0.511
TOR 0.513 0.467
ARI 0.497 0.397
ATL 0.517 0.413
CHC 0.516 0.417
CIN 0.467 0.449
COL 0.483 0.577
FLA 0.497 0.512
HOU 0.508 0.429
LAD 0.536 0.358
MIL 0.510 0.453
NYM 0.536 0.440
PHI 0.572 0.370
PIT 0.412 0.375
SDP 0.505 0.346
SFG 0.478 0.420
STL 0.546 0.493
WSN 0.417 0.478
Winner, Beyond the Box Score 32 Predictions Contest, 2009
Ok, not significantly different from zero for either league.
AL: t = 1.21, P = 0.25
NL: t = -0.64, P = 0.53
I think we’re seeing how much uncertainty there is in 90 games worth of team win data.
That, or we have to accept that interleague baseball is fundamentally different than other baseball, and I don’t buy that.
Only other explanation I can come up with is strength of schedule biases due to the natural rival thing—teams always play ONE other team each season, and that other team might be consistently above- or below- average.
-j
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
I was actually just considering this.
TEX – HOU
NYY – NYM
PHI – PIT
OAK – SFG
MIN – MIL
CLE – CIN
OAK – SFG
LAA – LAD
But:
KC – STL
See Data Differently. Beyond The Boxscore. | Follow me @justinbopp
Two Out Rally, the new BASEBALL MMORPG! | Facebook | Twitter
And here are 95% confidence intervals around the slope estimates
AL: [-0.19, 0.68]
NL: [-0.51, 0.28]
Centering sort of around zero in both cases… Just lots and lots of uncertainty.
-j
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
Might be worth it
To try this using each team and year as inputs. We’d be violating an independence assumption to include teams multiple times, though it is possible to include team in the model (even if it’s a pain to do so). I think we’re sample size limited right now.
-j
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
But here is what it looks like for a similar sample size but different criteria
April winning percentage vs. May-Sept winning percentage
Winner, Beyond the Box Score 32 Predictions Contest, 2009
Try this one
April winning percentage vs. May-Sept winning percentage
Winner, Beyond the Box Score 32 Predictions Contest, 2009
So basically sample size is too small
to draw any conclusions about a particular teams inter-league winning percentage over the period 2005-2009.
Winner, Beyond the Box Score 32 Predictions Contest, 2009
by Mike Fast on May 21, 2010 2:59 PM EDT up reply actions 3 recs
rec'd
Thanks for the work!
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
Wait,
overall record is on the X, right?
See Data Differently. Beyond The Boxscore. | Follow me @justinbopp
Two Out Rally, the new BASEBALL MMORPG! | Facebook | Twitter
by Justin Bopp on May 21, 2010 12:21 PM EDT up reply actions
So
the better the NL team, the worse they were against the AL?
See Data Differently. Beyond The Boxscore. | Follow me @justinbopp
Two Out Rally, the new BASEBALL MMORPG! | Facebook | Twitter
by Justin Bopp on May 21, 2010 12:23 PM EDT up reply actions
The intra-league record, that is
I subtracted out the inter-league record from the overall record.
Winner, Beyond the Box Score 32 Predictions Contest, 2009
Bias in talent assessment and valuation?
Let’s start from a position that the AL already has a greater amount of talent from an absolute perspective. Perhaps this due to the DH, perhaps it’s due to teams trying to keep up with the Yankees, perhaps its smarter GMs. Whatever; the cause isn’t important.
Because baseball is a zero sum game within each league (ignoring interleague for the moment) there will be the same amount of wins of production per player to go around. However, because of the existing talent differences, it takes more talent to produce a win in the AL than in the NL. 1 AL win > 1 NL win.
However, as Tango has shown, FA valuation is based on a single value of a win. If AL wins equate to more talent than NL wins, we have a problem. You get more bang for your buck with AL talent than with NL talent. Now, let’s then assume that players tend to stay in the league they came up in, again, for whatever reason. This systematic bias may rarely reveal itself.
The AL 2 win player signs for $9M a year, goes back to the AL and earns his money. The same happens in the NL. Nobody is the wiser. So even though both leagues are spending roughly the same amount on talent, because NL talent tends to stay in the NL and AL talent tends to stay in the AL, the talent imbalance is perpetuated.
Of course sometimes, that AL 2 win player signs in the NL and vice versa. If my thoery is correct, and assuming AL and NL GMs generally use the same pricing model, FAs previously in the AL who sign with NL teams will tend to outproduce the value of their contracts. And conversely, FAs signed from NL teams signed by AL teams will tend to underproduce their the value of their contracts. Obviously there would be a ridiculous amount of noise here, but I imagine it coudl be tested.
I’d be interested to see average payroll diference charted over the AL win %. If I’m thinking clearly, perhaps there was a spike in AL payroll in the early to mid 2000’s, reflecting a legitimate shift in talent. But due to the dynamics laid out above, the payrolls have since come back level while the talent distribution stayed unequal.
Good points
This is why we use different replacement levels for AL vs. NL. AL replacement is 2.5 wins/season below league average, NL replacement is 2 wins/season below league average. This keeps the cost per win roughly the same for each league—we just discount the NL prodeuction a bit more.
Now, whether the teams do the same…I can’t say. I bet the Mariners do. :)
-j
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
Oh...
I didn’t realize that different baselines were used in the WAR calculation. The more you know…
There should be, anyway
Fangraphs doesn’t do this, at least not last time I checked. They should.
rWAR, which is now at b-ref, does do this.
-j
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
More of the same
http://espn.go.com/mlb/notebook/_/page/bbtn100521/baseball-tonight-clubhouse
“NL is better than the AL” this season. While that would be a helpful data point, I think the old school analysis is going to falter once again.
by SagehenMacGyver47 on May 21, 2010 2:06 PM EDT reply actions
They should read their coworker's blog!
BTW, here are the first three commenters in Neyer's blog.
Good points all:
dkyanks09 (5/21/2010 at 1:49 PM) Report Violation
I think Keith’s talked about this in the past, but NL teams don’t break slot because their direct competitors don’t break slot. In reality, why should the Phillies care that the Twins and the Yankees are spending through the nose on the draft? So long as the Mets and Braves continue to play the game, the NL can keep its cartel pricing. In the AL, a couple teams started cheating on the cartel, and everybody else had to as well to keep up. To the individual teams, the divide between the leagues is nearly immaterial, so long as their relative place within their own league/division remains unchanged.
dayvander (5/21/2010 at 1:54 PM) Report Violation
Then there’s competitive effects. Whatever started the AL’s superiority ($ is a good bet), it then becomes perpetuated as the rest of the AL tries to compete with the top (AL East) teams. Just like the hand-held device market gets more competitive after a strong competitor like Apple leads the way, the same seems to have happened in the AL.
marisb55 (5/21/2010 at 2:22 PM) Report Violation
dkyanks nailed it. Unless someone in the NL (Phils, maybe the Cards) desperately wants to win the WS, you can consider the leagues separate markets beyond the basic level (that every team is playing baseball). So I guess the question is why compare the two? What do you get from that comparison? Why should NL ownership/management care that the AL is better? From strictly a dollars and earnings standpoint, I think that makes the NL smarter, but that could only be determined by comparing (basically) net income totals from year to year. Point is… I think your missing the point from the Front Office’s perspective, Rob. But then again, you’re a baseball fan, not an investor.
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
Why should NL teams go overslot?
Because the rewards are worth it!
If AL teams are winning more because they go over slot, and the they’re getting more wins per dollar than they can get off the free agent market or getting more than they are getting currently, then they should do it. Simply by transferring their budget from something else to going over slot, they could win more games, and they would not be spending any more money.
Does that maybe go back to Roy Neyer’s point that the spending discrepency is really just AL teams having better FO’s?
by Missing Barry on May 21, 2010 3:45 PM EDT up reply actions
I think what you're talking about here
extends beyond statistical analysis and sabermetrics, and heads straight for game theory.
Not that collusion is necessarily in play here, and not that it’s likely that all 16 NL front-offices have come to the same conclusion, but nevertheless…this sounds exactly like the prisoner’s dilemma to me.
Does anyone know which team was the first to go over slot?
Neal Huntington says hi to the first commenter.
as he busts up another slot.
"I generally avoid temptation unless I can't resist it" ~ Mae West
So one player drives the effect for an entire league?
I mean, I know he was other-worldly, but…
SFG Interleague Record 1997-2004:
1997: 10-6
1998: 8-5
1999: 7-8
2000: 8-7
2001: 10-5
2002: 8-10
2003: 10-8
2004: 11-7
Overall: 72-56 (0.562).
That’s a 16 game boost the the NL record over 8 years, or 2 games per year. Not nearly enough to counter the current disparity of ~29 games per year.
-j
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
MoneyBall A's
Thanks for the good article. However, there was one additional connection you should have made. The reason why the American League suddenly got smarter was because the MoneyBall A’s showed the league the importance of objective analysis. After the A’s showed that their good teams weren’t a fluke in the early 2000’s, several American League teams (Boston, the Twins, and later other) started ramping up their sabermeteric operations in order to compete. However, the National League, which didn’t have to compete against the A’s, have lagged behind the American League in their use of objective analysis.
Seems like all our conclusions end up back at the same place – superior front offices in the AL….
by Missing Barry on May 21, 2010 3:45 PM EDT up reply actions
Basic agreement.
Why don’t we actually rank them based on an agreed-upon number of criteria?
See Data Differently. Beyond The Boxscore. | Follow me @justinbopp
Two Out Rally, the new BASEBALL MMORPG! | Facebook | Twitter
How about WAR/$?
While saying up front that I don’t know the nuts-and-bolts of how WAR is calculated, how about Wins Above Replacement per (payroll $ – minimum salary*25)? I say this tentatively, because I am not sure if WAR is separately scaled for each league or not. I suspect that it is.
Like the idea.
As long as we acknowledge that wins can be more valuable for teams on the cusp, and thus it’s justifiable to pay more.
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
Twins?
I thought they were all scouting, all the time.
-j
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
Doh, meant the Blue Jays
Doh! I meant the Blue Jays. Twins, Blue Jays. You know, cold places. I’m not exactly saying the Blue Jays perfected anything either, but Toronto and Boston were the first teams to jump in after the A’s, and then objective analysis spread further. I think the idea of objective analysis spread faster in the American League simply because the American League saw and experienced it in action by playing the A’s in those early 2000’s years. After they saw it in action, I think some general managers and perhaps more importantly owners (new Boston owner John Henry was the one who really pushed it in the Red Sox organization) realized that their operations had to change quickly.
You could be right.
Wouldn’t surprise me if it’s a factor. The Twins kind of buck the trend there, but they also hit the catching jackpot, which must help them a lot.
-j
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
Great article and great topic
only other thing I could add is luck, but that could only be a part of it. How many standard deviations are the current empirical results away from the mean given a 50/50 coin flip win for each interleague game? I think we should measure that.
vr, Xei
Great question
I don’t know. Someone want to tackle that?
-j
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
I think its pretty obvious that there is just a lot more talent in the AL...
Im not entirely sure why this is, and this isn’t really the best proof of it, but look at how dominant the AL has been over the NL over the last however many years in the All-Star game.
Sometimes the best way to convince someone he is wrong is to let him have his way. --- Red O'Donnell
...
A .566 W% (the 5-year AL average) translates to a 90-72 record.
Uh not necessarily. A .566 W% means that the AL has won 90 out of 162 games against the NL and the NL has won 72 against the NL. Using the log5 formula that implies around an 86 win team.
My point
Was just to illustrate how large a 0.566 w% is. Comparing it to a 162-game record makes it more approachable—people know 90 wins is a great season, but don’t know a .530 from a .560 w%. You’re right of course that the losing team should share some “credit” for the winning team’s success.
-j
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
by JinAZ on May 22, 2010 1:48 PM EDT via mobile up reply actions
Emailed comment, thought I'd repost here
Sir:
In answer to your question on the statistical significance of AL dominance in the last half-decade or so, it appears to be a real effect. I calculate a 4.7-sigma pull from average for this period.
From 2005-9, the AL won 714 games vs. 546 for the NL (.567 for the AL). For a set of 1260 games (630 games = .500), there is a standard deviation of 17.7 games (i.e., sqrt(1260)/2), so (714-630)/17.7 = 4.7 sigma.
Four additional thoughts on the “why” of this follow.
* There aren’t enough data points to really test it, but the results in 2001-3 are consistent with the “Moneyball” hypothesis — Oakland was 37-17 in years where the NL was winning a small majority of games.
* The individual seasons are not statistically independent phenomena. A five-season stretch is well within the career of a cohort of players, so there may not be any inherent AL advantage, just a fluctuation of better players over a stretch of time. This “ought” to cancel out on a league-wide basis, but maybe it’s just a fluke of talent distribution. If it is such a fluke, then it ought to go away in the next couple of years. I would bet against this one.
* Almost every explanation fails to account for the fact that better NL teams have not won more interleague games than worse teams.
* Some information that might shed light on the “front office” and “payroll” hypotheses. I previously found that the regular-season wins of teams had a total variation of 11.9 games/team/season. The expected variation due to chance if all teams were equal would be 6.4 games/team/season (sqrt(162)/2). Looking at how this varied with payroll led me to break down the variation as 28.4% chance, 28.3% payroll, and 43.2% “other.” The “other” category represents everything that isn’t chance or payroll — front offices, injuries, Manny Ramirez deciding that he wants to be traded, etc. So seeing a 2-sigma effect in interleague play would be rather consistent with payroll/front office effects. A 5-sigma effect is harder to explain — unless the difference in front offices within each league is much smaller than the interleague difference. The existence of the Royals and Indians leads me to doubt that the interleague gap is that wide.
I hope you find this useful.
—Jeremy
I write at:
Beyond the Boxscore | Red Reporter | Basement-Dwellers.com | Twitter: @jinazreds
by JinAZ on May 22, 2010 1:52 PM EDT via mobile reply actions
Almost every explanation fails to account for the fact that better NL teams have not won more interleague games than worse teams.
This we believe is simply due to the small sample size, right?
Better teams have not won more April games, either. (Btw, I didn’t examine any other months. The reason I chose April was that I knew that it wouldn’t contain interleague games. The only other month that I know that is true for is September, and I thought September records might be affected by whether a team was in the pennant race or not.)
Winner, Beyond the Box Score 32 Predictions Contest, 2009
Comments For This Post Are Closed
Secondary Sidebar
User Tools
FanPosts
Community blog posts and discussion.
Recent FanPosts
Follow us on Twitter!
BTB: @BtBScore
Justin: @justinbopp
Satchel: @SatchelPrice
J-Doug: @RationalPastime
Josh: @garik16
Julian: @giantsnirvana
Bill: @BillPetti
Lucas: @DBITLefty
Jeff: @jeffwzimmerman
Lewis: @LewsOnFirst
