Overperforming, Dept.
You be the judge.


The Baltimore Orioles have scored fewer runs than they have allowed - a fair bit fewer, in fact.  They've scored 585 and allowed 606; their run differential is worse than the Seattle Mariners' is.  However, their won-loss record is 76-60 and if the season ended today, the Orioles would be in the playoffs, fighting for the hardware.

At Bill James Online, one of the feature writers (Dave Fleming - no, not that Dave Fleming, M's fans) looks at this situation.  He analyzes all historical teams that are way, wayyyyy better than their run differential and finds something interesting:  it may not have been just luck that caused the clubs to outperform Pythag.  

Our Sabertistas' Creed quoth that when a team wins or loses fewer than its run differential suggests, then the one and only cause of these extra wins (losses) is the roll of the dice.  Which is precisely why it's interesting to see something suggestive in the history.  Of all division-era teams that finished much better than their run differential, look at how strongly they followed up in years to come:



Looking at the eight Division Era teams:
1 Year Later
2 Years Later
3 Years Later
4 Years Later
1st, Lost WS
1st, Lost NLCS
1st, Lost WS
1st, Won WS
1st, lost NLCS
1st, Lost NLDS
1st, Lost ALDS
1st, Lost ALDS
2nd, Lost ALDS
1st, Lost NLCS
1st, Lost ALCS
4th (?)
All of these teams reached the playoffs again.
Well…not all of them. The 2009 Seattle Mariners didn’t reach the playoffs in 2010 or 2011, and they probably won’t in 2012. We’ll see how 2013 plays out before we write them off.
But the other seven teams all reached the playoffs in the four-year window following their ‘lucky’ season.
Granted, this is partially because it’s easier to reach the playoffs now. In the Division Era, four teams reached the playoffs every year. In the Wild Card era, it’s been eight teams (and it’ll go up to ten this year).
It’s interesting that two of these teams are ‘dynasty’ teams, the teams of their decade: the Big Red Machine lost the 1972 World Series, but they won in 1975 and 1976, and were in contention in nine of the ten seasons of the decade.
The other dynasty team is the New York Mets. Despite having only one championship to show for it, the Amazings were a terrific team from 1984 thru 1990, finishing first or second in the NL East every year. In a decade where no team managed two World Series victories, we’ll give ‘em the nod.  
Bottom line: teams that have really ‘lucky’ seasons tend to go from ‘lucky’ to ‘really good’ in subsequent years. Good news for Orioles fans. - Dave Fleming


A glance at the table shows that this collection of teams is packed with ballclubs that were just getting really good - and that, in their first season rising to the top, they won lots of games that weren't yet quite supported by their runs gained and lost.  For example, take the Big Red Machine led by Johnny Bench and Pete Rose.  In 1970 they weren't all that overwhelming yet; their run differential was +775, -681.  But in 1970 they won 102 games.  Little Jeffy Clarke was on the scene for it; that ballclub had swagger.  In particular its catcher had swagger, not to mention 148 RBI.

No principle in baseball is absolute; various general principles weave across one another in a super-complex dance of cause and effect.  But it could be a general tendency that the first year a club starts believing in itself, it wins 92 games when its run differential suggested 82.  Sabermetricians examine the following years to see if a team is still outperforming Pythag, and they find "nope.  Not outperforming Pythag.  They only won 96 the next year and Pythag said 95!"  Perhaps Pythag catches up to the W/L, and not vice versa.

The plucky young 1972 Mets, in the table above, had Tom Seaver.

The plucky young 1984 Mets, in the table above, had Doc Gooden.

The plucky young 2012 Second Half Mariners, who are way outperforming Pythag since trading Ichiro, have Felix Hernandez.

Fleming doesn't claim that his study is airtight, not by a long shot, and I don't say that "rising young teams win some tight games" is to be taken as the gospel truth.  But the teams that were WAYYYYYY over Pythag did tend to win a lot in the next few years, so there might be some insight into the game of hardball there.

The M's, since they got dynamic, have been echo'ing this pattern to some degree.  It's an interesting camera angle on the evolution of these Mariners.




ghost's picture

If you scan those eight teams, only 5 of them actually improved their performance in year 2 over year 1. finishing ranks can be deceiving. And some of those teams took 2 or 3 years to get good again after having their lucky year. 2 o 3 years is an eternity in baseball because roster turnover suggests that half of your team is probably different by year 3.


The Mets and Giants were certainly teams transitioning from bad to good ... and the records switched ahead of the run differential.
But ... NY and LAA were both in the midst of a string of division titles when they had their "lucky" years.
And the comp of Arizona is almost laughable ... here are the '07 and '11 primaries:
CA - C. Snyder - M. Montero
1B - C Jackson - J. Mrinana/X. Nady
2B - O. Hudson - K. Johnson
3B - M. Reynolds - R. Roberts
SS - S. Drew - S. Drew
LF - E. Bynes - G. Parra
CF - C. Young - C. Young
RF - C. Quentin - J. Upton
No overlap at all in the rotation.
The fact that the Giants and Mets continued a trend of improvement is certainly an outcome worth hoping for ... but it's an awfully small sample.


I recall looking at the 2009 M's team and what explained the Pythag in my eyes was the 2 black holes at the end of the rotation feeding many more runs than would have been average to opposing teams in losses. Since the context should help some here remember I'll list every pitcher that had at least 1 start and an ERA over 4.00. It adds up to 78 starts between those, but 4 isn't a horrible ERA for 2009, just not good. League average was 4.45.
Garrett Olson 25 years, 3-5, 5.60 ERA, 31 games, 11 Starts
Jason Vargas 26 years, 3-6, 4.91 ERA, 23 games, 14 starts
Ian Snell 27 5 2 .714 4.20 12 12 starts
Luke French 23 years, 3-3, 6.63 ERA, 8 games, 7 starts
Chris Jakubauskas 30 years, 6-7, 5.32 ERA, 35 games, 8 starts
Doug Fister 25 years, 3-4, 4.13 ERA, 11 games, 10 starts
Brandon Morrow 24 years, 2-4, 4.39 ERA, 26 games, 10 starts
Carlos Silva 30 years, 1-3, 8.60 ERA, 8 games, 6 starts
That's 46 starts by pitchers over 4.5 ERA. The bullpen stalls up similarly in that the back end got horrible (notice many of those pitchers appeared in games they didn't start as well). Use whatever stat you prefer, I know ERA isn't the best. Figure it works fine for this though. I have not looked at any of the other teams, but when you have both rotation and bullpen that start strong in the first couple spots and become horrible at the back end your pythag is not very likely to come out accurate compared to w/l. They good relievers often went in to the games they were winning and the bad ones made games they were losing even worse.


To get better in year 2 than in year 1.  Holding their ground on W's, or even going backward by 5 games, would have confirmed the first hypothesis.  
Y1 involved an overperform of Pythag by 10-15 games; the "true" talent was supposedly much worse than Y1 win totals showed.  You're talking about .500 teams that made the playoffs "by luck," so the fact that there were an unusual number of playoff appearances is what caught Fleming's eye.
Three years might be an eternity, but in 1973 the Reds still had Bench, Rose, etc.  Three years ago the M's had Felix.


I guess I should have typed in all caps.  Fleming's little look at +10 Pythag teams isn't intended for the JAMA,  It's a first suggestion of a provocative idea.  Not that "sample" is an appropriate term in your context, but here a guy comes up with something a *little* outside-the-box and our response?  To jeer.  That'll teach him to stick to creed rehearsal in the future.
Wish you'd leave the "laughable," "ridiculous," etc. stuff out.  Your ideas are strong enough witout that.
Fleming didn't say, "Hey look.  All eight Pythag teams were contenders on the rise."  He asked, is there a tendency that is reflected here more often than not?


That is a super strong idea and I don't remember seeing it focused on before.  Somebody should submit it to James... bet you it holds up.
The #4-5 starters are so readily "fixable" in Y2, in a way that (say) a black hole at C isn't.  French's 6.63, in 2/3 a season, and Olson's 5.60 in 3/4 a season, that would have quite an impact on run differential, and it wouldn't really carry across to Y2.
Of course, reality is a weaving intersection of causes and effects, dozens or 100's or millions of them, and the M's internal implosion in 2010 might have trumped the more subtle considerations.
As it pertains to the 2013 M's, it may be another positive indicator.  Noesi's ERA was 5.77 in 17 starts, half a year ... in Safeco that's tough to do.  :- )  Beavan has shown talent but his ERA is 5.01 also, in 2/3 a season, and that's a 75 ERA+.

ghost's picture

I'm sorry for taking mild offense here, but...his idea was indeed discussed on your blog extensively before. By me. In 2009 while that team was playing. Complete with a new method for calculating pythag - do you recall PythagenMatt? You seemed to like it at the time...as did Tom Tango. Do you recall the intense discussion I had about that team being considerably better than traditional pythag because its' offense was consistent and its' pitching was very VERY inconsistent - while possessing a good bullpen? I even looked at 2000+ teams from the modern era and foud that teams dramatically over-performing seasonal pythag were very likely to have a higher PythagenMatt W% and that the over-achievers played to the PythagenMatt in the following season!
Tom Tango covered it on THT's website!
So yes...that idea has been discussed before.


Was referring specifically to the idea that if your #4-5 starters are hemorrhaging runs - to the tune of 6+ ERA's - that the next year your run differential would surely be in for a large easy improvement, and in that sense the Y1 run differential would be indirectly misleading.
PythagenMatt isn't limited to that idea, is it?  
Lemme re-set here, 'cause am curious:  in a situation like 2009, would French's and Olsen's bloated ERA's be "seen through" by PythagenMatt?

ghost's picture

I definitely discussed the matter during the '09 season and '10 offseason - it was one of the reasons I was erroneously confident that the 2010 Mariners would improve despite the bad seasonal pythag.
The 2009 Mariners were a considerably better team than their run differential for three reasons:
1) The three worst starting pitchers were so bad that the team suffered many severe blowout losses and had a far worse record in blowouts than normal for a team of their basic abilities should.
2) Their bullpen was highly effective - they played better in close games than you would expect based on their seasonal pythag (which was misleading due to item 1, but also due to the reality that once they had a late lead, they generally won the game)
3) The offense was more consistent than normal. A consistent offense is good for winning above pythag because you can't win if you don't score at all - most teams have .500+ W% when they themselves score 4 runs or more. The '09 Mariners had a surprisingly knack for scoring 4 runs considering how weak their offense actually was.
PythagenMatt picked up the fact that that 2009 team was better than pythag and that it was not as good as an 85 win team (it estimated they should have won about 79 games - seasonal pythag said 69). My own analysis of over-performing teams revealed that the keys to overperforming were getting blwon out worse than normal, having an unusually good record in close games, and being relatively consistent on offense (a R/G distribution with more kurtosis - a higher peak near your average ability - than the average team). The '09 Mariners were a special case of all three of those being true. Unfortunately, they fixed the pitching and then screwed up the hitting in 2010.

Add comment

Filtered HTML

  • Web page addresses and e-mail addresses turn into links automatically.
  • Allowed HTML tags: <a> <em> <strong> <cite> <blockquote> <code> <ul> <ol> <li> <dl> <dt> <dd><p><br>
  • Lines and paragraphs break automatically.

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.


  • Allowed HTML tags: <a> <em> <strong> <cite> <blockquote> <code> <ul> <ol> <li> <dl> <dt> <dd>
  • Lines and paragraphs break automatically.
  • Web page addresses and e-mail addresses turn into links automatically.