I calculated league quality as well, back in the day...the blue line in that graph made me laugh, as I definitely remember when that article came out thinking how absurd their results were. :)
Although I ascribe to the curve-fitting approach which assumes that the best players in 1930 or even 1880 were probably pretty comparable to the best players today in terms of talent...and that if those same players were born into our system, and had all of its advantages in training and conditioning, they would be stars just the same...which means I intrinsically recognize that the raw number relating league quality should not be assuming to apply linearly to all players...I also recognize that people are interested in what league quality looks like.
My method was to take James' insight as to what a bad league looks like to its logical conclusion. All of those things that James noticed happen more in poor league than in good ones...they all express themselves in one simple way. They increase the variability of run scoring game by game, and, moreover, they increase the kurtosis (a statistical term for the compactness of a distribution).
Basically, I noticed that in bad leagues, there are more shutouts and more 15 run showings.
So I computed the skewness and kurtosis of the RS distribution for all game-sides in each league and guess what...unlike either of the other systems in that above graph...mine "saw" expansion seasons, the steroid era, WW2 (those other methods see WW2 as a mere blip, mine sees it as a catastrophic reduction in quality...historians will tell you the latter is the case, not the former), and the periods where there was a third league, and sees those things well.
Mine also looks EXPONENTIAL/ASYMPTOTIC rather than linear. Which makes a lot more sense to me. Like most sports, the quality of performance will have some upper limit as we reach the boundaries of what humans can do. And like most sports, athletes getting paid to be good at that sport will very quickly approach those limits. So my system sees league quality of:
1871: 48%
1881: 68%
1898: 80%
1920 AL: 90% / NL 87%
1935 AL: 92% / NL 84%
1944 AL: 79% / NL 79%
1960 AL: 94% / NL 92%
1961 AL 90% / NL 92%
1962 AL 91% / NL 88% (as expansion impacts each league, the results are visible...especially this is true in 1993 and 1998)
1985 AL 96% / NL 97%
1993 AL 97% / NL 94%
1998 AL 91% / NL 90%
2001 AL 99% / NL 95%
2014 AL 100% / NL 98%
BTW, mlb.com is showcasing the next-gen tracking statistics that will be available to us in the coming years statcast videos are fascinating to watch because they give you a direct numerical summary of just how RIDICULOUSLY good MLB players now are. They show, for example, that Jerrod Dyson runs at 22 mph top speed, gets his first step 0.15 seconds after the pitcher begins his pitching motion (a slide step), and reaches top speed in 7 strides...and is still thrown out by Caleb Joseph, whose pop-time is a blistering 0.67 seconds! WOW. Think about that.
.
You know one way to tell that high school baseball is lower-quality than the National League? I mean, besides just watching the games? By using James' indicators that compare league quality.
Many years ago, James put a framework into place to triangulate this problem. He drew a loose triangle with his "points system," but the triangle did contain the answer inside it. Compared to the American League, high school baseball has:
- More fielding errors
- Better-hitting pitchers
- Fewer double plays
- More .400 hitters
- etc ... on to 16 James indicators
For example, take the "better-hitting pitchers" in high school (and in the National League in 1906). That implies that an Ordinary Joe can do better against the top flight - so obviously the top flight isn't as elite yet.
So you can add these indicators up, for any given league. You can use this logic to compare, say, Japanese baseball to the National League. Loosely.
........
Baseball Prospectus and the Hardball Times tried to move in the sides of James' triangle, to make the triangle smaller. If they did it, they did it in the most appropriate way - with more sophisticated mathematics. Here's a HBT article on it.
In the below graph - directly visible here - the blue line is BP's calculation and the red one is HBT's. These mathemeticians beliee that they can measure the amount of league differential (better than James had). It's a noble effort.
Another way to say this is that according to Baseball Prospectus (the blue line), the American League of 1927 was only half as good as today's. Babe Ruth's wOBA would be cut in half today -- not only would Ruth, Gehrig, Grove, etc. not be good players; they'd be terrible players, not able to make any ML roster.
I, and Bill James, have always believed this doesn't pass the chuckle test. If your system tells you that the best CF of a decade -- Joe DiMaggio, Willie Mays, Ken Griffey Jr. -- was grossly different than the CF of another decade? Find another system.
Still, league quality has changed a lot -- and Honus Wagner's play would have adapted to the new environment he grew up in. James (and I) believe that Honus Wagner would be a star today, had he be born in 1985.
........
Notice that HBT's calculation (the red line) is grossly, grossly different from the blue line. Yet saber dudes who draw the blue lines -- easily subject to refutation in the future -- are far too confident in their positions, usually stating them as "correct" belief systems.
It's a lesson I'll probably never stop harping on: Respect for the Complexity of the Problem.
........
HBT does a vivid job of summarizing their own math:
In fact, (our) method (the red line) still shows league quality getting better over time; it just shows a much less pronounced rise. By 1925, the average player was 80% as good as today’s average player; by 1973, he was 90% as good. Conversely, the first method says that the quality of competition was not 80% as good as today’s until 1982, and he was not 90% as good until 1994.
.........
Bill James, in today's Hey Bill, puts all this onto the TV monitor for us:
.
....
Us oldtimers, in our 50's and 60's, are dazzled by the speed and efficiency of the play today.
And it's one reason that I believe Russell Wilson is an evolution in NFL football: the game is played lighter-and-quicker in his world.
....
Baseball GM's have to adapt to the adaptation. If long-sequence baseball stops working, you need to put a thumb on the scale against players who benefit from long sequences. For example, basepath speed tends not to impact a modern MLB game the way it can in a lower-quality league. Billy Beane, father of the Matt Stairs template, seems vaguely aware of this.
One Seattle Mariner that I believe benefits from Russell Wilson "evolution" syndrome: Hisashi Iwakuma. I think that he flits around ungainly enemy power in a way that big dumb howitzer arms don't. I ain't saying that Iwakuma is the best pitcher -- just that he's a lot better than he "should" be, because he's a type of next-generation player.
I've suspected that about Japanese pitchers generally since when ... oh, a long time. Not that they're the best, but that they're underrated. Their physical ability tends to be a bit lower, but they're next-gen.
Any NPB pitchers comin' out?
Robinson Cano, like Ichiro, adapted his game to his new surroundings with ease. He simply walked into Safeco and decided he'd hit for a little more OBP and a little less SLG. In a lot of ways, Robby strikes me as a next-gen ballplayer.
The Mariners' bullpen, in 2014, was probably next-generation. They should pass a rule against so many pitching changes, but as long as they don't, more switches is probably better. Consider the fact that Lloyd McClendon is more onto this point than other managers are.
Cheers,
Dr D
Comments
Matt - I realize you might prefer not to answer this question and totally understand if so, but I am curious - based on your time in a cutting edge MLB franchise, how would you compare your own analytical abilities to those who are employed in similar roles in MLB? I think your stuff seems brilliant, complex but synthesized in a simple way which is quite difficult to do...but I don't know the state of the art :) Anyway it's good to have you on the site contributing at least.
I am not more than an average scientist by today's high standard for scientific achievement and skill. Bill James is better at setting up an empirical study than I am. The difference between a successful analyst in MLB and someone like - not to put too fine a point on it - Tony Blengino (who is NOT successful, and I can see why) is his/her ability to ingest other ideas and see it when those other ideas are better than their own. Any scientist faces the possibility of tunnel vision. Well...any PERSON does, but scientists, because their knowledge is so specialized and because they work so hard to get what knowledge they possess, are much more likely than average to start falling in love with their own works.
I was like that in 2003/2004 as I was developing Pythagorean Comparative Analysis - which was my attempt at an uberstat metric. I was dead set convinced that James had some good ideas and Palmer had some good ideas and Tom Tango had some good ideas...but I had better ones. I was certain I was seeing a way to combine a lot of good ideas into a more cohesive logical structure that would produce results that were simply more correct. You know what cured me of that certainty? A debate I had with a hard-headed "scout-ist" (someone who thinks the scouting view is superior to any numerical one) over whether Tom Glavine was a hall of famer. I had my defensive independent metrics with some adjustments for pitcher fielding and batted ball trajectory and I thought that conclusively proved that Tom Glavine should not even be considered for the hall. Dr. D would truly have hated me in 2003/2004. :) In fact, I think I used to discuss things on the fanhome forums and Dr. D was there and frequently ready with BABVA phrases and comments about how I needed to get smart enough to realize I was not that smart. :) But this debate went on about Glavine for an astonishing 3 months and over 1000 posts back and forth over at the forum baseball-fever.com. And during the debate, I was challenged to prove numerically that Glavine did NOT positively influence the batted ball results consistently as I claimed. So I devised the prototype for Defense Neutral Run Average - which used play by play data to determine the quantifiable impact (year to year) of the pitcher on batted balls in play - it was essentially DIPS plus a with or without you framework for understanding balls in play. And...holy hannah...Tom Glavine really does get weak contact...and what's more...his high BB rate was LEVERAGED...he was walking people more when a walk changed the run expectancy of an inning less (!). Picking and choosing who he wanted to get out.
I eventually became a loud defender of Glavine's HOF case against other sabertistas as Doc calls them in that same thread.
When I see good evidence that I'm just plain wrong about something now, I try to incorporate that evidence and invent new tools. I still fight like the dickens defending my ideas but I do it now not because I'm 100% certain that I'm right, but because I want to see the opposition's best evidence that I'm wrong.
Even with that light bulb coming on back in 2004/2005, I don't have the temperament to be a great MLB analyst for a living...I'm an emotional sort of guy and you really can't be emotional in baseball. Blengino does not, I don't think, have a good temperament either. As a result, his analysis is stale and not particularly inventive. If I had more time these days, I'd be doing more of the analysis stuff, because I miss it. But, alas, all I can really do is throw out ideas and hope someone carries them forward. Darn you, life, you're getting in the way!
Thanks. I would consider temperament and intellectual open mindedness to be integral components of "ability"...but I totally get your distinction of the different facets, and it makes perfect sense to me. I like your point about Bill James being great at setting up the experiment. That type of creativity is rare and a distinguishing skill.
Looks like an easy URL or three :- ) Referring even more to your first comment/post/article than the second.
Simply outstanding. :: golfclap ::
Their main mistake was to assuming that, on net, changes in player performance year to year reflect changes in league difficulty. Their curve is too steep because (prepare to laugh at the obvious mistake)...they forgot that players age.
Ooooopsie. :)