Graz Sports Talk

At lunch a couple of weeks ago, a couple of guys from the office and I were talking about sports in the US as opposed to Europe, specifically the structure of leagues and issues of (forgive the cliche) parity.  The US leagues have all these mechanisms—absent in Europe— to push teams to the same level in order to make the games interesting: whether it’s the NFL’s schedule adjustment to give tougher schedules to teams that finish well, various leagues’ drafts, and salary caps.  Of course baseball doesn’t really have the same situation regarding the latter two: the draft isn’t as impactful, it seems, since it takes a while to bring up prospects, they get traded often, and there’s a huge number of picks.  Meanwhile, the luxury tax acts as a penalty and not a constraint (and not much of a penalty in the later Steinbrenner years, if spending is a clue).

But there’s a lot of fluff games in the US,  intuitively, at least.  As of this writing, the Cubs are in the basement of the NL Central and still have a series against the Phillies, who might be able to get to .500;one against the Marlins, who are sub-.400; and two against the Brewers who are a whole 1.5 ahead of the Cubs.   In the NFL’s week 15 last season, did anyone outside of Michigan and Arizona care about the 4-10 Lions hosting the 5-9 Cardinals?

On the other hand, in European soccer leagues, you have leagues where the title is,before the season begins, clearly  to be decided by at most 3 or 4  teams.  We “know” La Liga’s going to go to one of two teams.  In the last two years, how many weeks in did it take to decide that it would be Dortmund or Bayern for the Bundesliga?  Of course over the course of several years, teams fade from being locks of the top few spots, but it’s rare, it seems, to have a truly wide open title race between more than the elite handful.  Compare that to the NFL, where a cursory Google of betting lines suggests that two teams—the Broncos and 49ers— have a 10% or better odds of winning the Super Bowl.  In La Liga two teams essentially have 40%+ chances and the rest essentially none.

But this is of course neglecting that aspect of European soccer that, along with enjoying watching the games themselves, made me a fan (lapsed that I am without TV these days): Champions League/UEFA spots and (especially) relegation.  Ah, I remember watching my first season of EPL with Joseph Cross in the apartment senior year at Richmond and marveling at the drama of those bottom rung games.

These thoughts got me to thinking about what makes me find a league interesting (outside of being a fan of a team).  I’d like a metric which allows for comparison within a sport’s leagues  (for soccer) and, if possible, across sports.  I’m not necessarily even caring about “competitiveness.”  I like dynasties in general, as long as they’re not too extreme.  So “parity” and the like don’t matter for me. League spread is tricky to use: even if you take a fictional baseball season where a team matched the  worst baseball record since the World Series started (.248) and a team matched the .721 by ’54 Indians, you’d have a league which looks tighter than most modern football seasons.  Last season in the NFL you had 5 teams at .250 or under and 5 at.719 or higher.   I remember the 2007-2008 EPL where you had Derby at 11 points (around 9.5% of the possible total) and Man City with 76% of a perfect total.  Not that you couldn’t plot all of the distributions, but things would get screwy.  And I’m looking for a low-bias metric if possible.

Maybe league spread could work.  But this article on Slate from a while back got me thinking about other means of comparison.  There, the notion of  “meaningful” games is floated.  But how do you describe “meaningful”? In that article, it’s simply how many games a team plays before they are eliminated from the playoffs or have clinched (they add in competing for home-field or a bye in the NFL).

This seems a bit too generous.  Technically, my Cubs weren’t eliminated from a division title a few days ago, but being 20.5 games back with 23 games left means that no one has believed them to still be in the hunt for a long long time (April? maybe?).  So, maybe “meaningful” isn’t the right term.  Unless we’re talking a handful of games back, I’m not that foolhardy in my hope (Cubs fan notwithstanding).  

I don’t know the right term for the notion I have; for now I’ll run with “interesting” with respect to a neutral fan.  And no, team pride/bragging rights/history doesn’t count for my neutral fan.  I want something more stringent but along similar lines as “meaningful”.  But it will also be a bit fuzzy. I’m also throwing out everything involving the playoffs in US leagues and any non-league matches in football.  Just regular season games/matches.  This includes things like the relegation/promotion playoffs.  I next have to define what makes a game “interesting.”  I’d say at least one team has to be in the hunt for a spot.  But this is still fuzzy.  So to make things (more) precise: a game shall be labeled “interesting” if at least one team’s current situation means they have a reasonable chance of being, on the final game/round of the season, within one game of either:

  • for U.S. leagues, making the playoffs–or better;
  • for European soccer leagues, the cutoff due to table position for a European cup competition (or better) or the highest relegation spot.

This second criterion is asymmetrical for a couple of reasons.  As a neutral fan, I don’t count a team locked in the basement of the table as being relevant; so I want to throw out teams like the aforementioned 07-08 Derby.  And the top of the table has several races going on at the cup spot and better: there’s UEFA, Champions League, and the title.  Also I don’t want to discount the appeal of watching the better teams competing.  For U.S. leagues, it’s a bit trickier since this considers a 14-1 team in the NFL (which will surely be resting all its starters) interesting.  But I want to be fair to both sides of the sports Atlantic and keeping things binary means I don’t have to worry about sorting through the relevant importance of UEFA/Intertoto versus UCL spots.  And I don’t have rank the importance of a team trying to get into the playoffs versus one trying to get home-field advantage or a better seed.

The criterion will be a bit crude  with respect to asynchronous situations due to byes, postponements, etc.  Simply, I will ask how many teams with a given record (in NFL, NBA, MLB) or points in a round (Euro soccer leagues) ended up satisfying the above criterion out of the total with that record/point total. With enough seasons (more on that below), this gives me an idea on the chance a team has of being relevant in the end.  Here’s the main difference with the Slate piece; we will factor in things like how no team 20 games back (I assume) with 30 games to play made the baseball playoffs or how few 0-6 NFL teams make the playoffs despite being not eliminated.  The nice thing with this is that a team can stop being interesting during a season then become interesting again with a surge or (in soccer) a plummet in performance.  So a team we give up on early in the season rattles off a good performance and we as fans care about them again. We write off an 0-4 team in the NFL and then, after a big wins, care about the 7-6 team.

I have to next introduce some bias by selecting a minimal chance/percentage to consider.  I admit, I’m not an expert or even more than roughly familiar with user-bias issues, but this is the one number I can tweak.  I am not sure how it will affect the comparisons between leagues, but I’ll update this when i get a chance to compare different percentages.  For now, I’ll go with a (possibly) mild criteria of 40%.  That is, you need a 40% or better chance to end within one of the above cutoffs at the end of the season to be interesting at the point in question.

Since this has already gone long, I’ll drop some numbers on you.  I got a table of data from the nice blog Clearly and Simply for both EPL and German Bundesliga seasons for 15 seasons (1996/97-2010/11) and 14 (1996/97-2009/10), respectively.  It’d be nice to get more seasons, but I’m not sure where to get easy to work with data sets.  If you have an idea, please let me know.  Thanks to some help from Joseph Cross, I got to the point where I could muddle through my rudimentary Excel knowledge (I have never really used Excel before) to calculate the EPL numbers and Bundesliga as well.  The ones for the MLB are coming shortly.

Season
1996-97
1997-98
1998-99
1999-00
2000-01
2001-02
2002-03
2003-04
2004-05
2005-06
2006-07
2007-08
2008-09
2009-10
2010-11
Games of Interest329303310325314275296311318314318337318330288
% of Games86,578947%79,736842%81,578947%85,526316%82,631579%72,368421%77,894737%81,842105%83,684211%82,631579%83,684211%88,684211%83,684211%86,842105%75,789474%
Average % per season82,210526%

Here you see the portion of games each season with at least one team whose record going into the match hits the criterion above; for week 1 I assume every team is in the hunt artificially.  Below, I’ve included a chart showing the portion of games each week over all seasons satisfying the criteria.

Week to week over all seasons.

EPL week to week over all seasons.

I ran the same tests on 1.Bundesliga matches:

Season
1996-97
1997-98
1998-99
1999-00
2000-01
2001-02
2002-03
2003-04
2004-05
2005-06
2006-07
2007-08
2008-09
2009-10
Games of Interest253251270270274255258248278244255270271262
% of Games82,679739%82,026144%88,235294%88,235294%89,542484%83,333333%84,313725%81,045752%90,849673%79,738562%83,333333%88,235294%88,562092%85,620915%
Average % per season85,410831%
Bundesliga week to week over all seasons

Bundesliga week to week over all seasons (ignore the weird numbers at the bottom; first number shown is the round)

There’s a caveat in these numbers as the number of European cup spots each league gets changes from season to season.  So hopefully it washes out.  But it’s something to keep in mind if I do leagues with traditionally fewer spots.  Off the bat, you see that even if on average 1.Bundesliga has more matches with one of these interesting teams, as the season progress, the league has started to settle more.  Qualitatively, I noticed in the data that the final table tends to not have too many teams near the border when compared with the English league.

Joseph mentioned he thought maybe this was too generous a metric and wanted to know if I required a match to have both teams in a race. So here you go:

Season
1996-97
1997-98
1998-99
1999-00
2000-01
2001-02
2002-03
2003-04
2004-05
2005-06
2006-07
2007-08
2008-09
2009-10
Games of Interest122149151139160128136118155116137161155154
% of Games39,869281%48,692810%49,346405%45,424837%52,287582%41,830065%44,444444%38,562092%50,653595%37,908497%44,771242%52,614379%50,653595%50,326797%
Average % per season46,241830%

(For the Bundesliga)

Season
1996-97
1997-98
1998-99
1999-00
2000-01
2001-02
2002-03
2003-04
2004-05
2005-06
2006-07
2007-08
2008-09
2009-10
2010-11
Games of Interest157145115161132107122128150152150181135150111
% of Games41,315789%38,157895%30,263158%42,368421%34,736842%28,157895%32,105263%33,684211%39,473684%40,000000%39,473684%47,631579%35,526316%39,473684%29,210526%
Average % per season36,771930%

(For the EPL)

Bundesliga for both Teams being of interest in a game

Bundesliga for both Teams being of interest in a game

EPL for both teams being of interest in a game

EPL for both teams being of interest in a game

I’m curious how this might play out on other leagues.  Like I said, the next step is to run the same on MLB and maybe NFL.  But for now, I need a break from Excel ‘fun’ projects.