Further Investigation into the Lack of Quality Middle Infielders

Last weekend, I introduced the Gagne Threshold.  I used a variety of criteria to find good middle infielders over the past 30 seasons.  I'm not going to go through everything again, but if you missed it and want to take a look, you can find it here.  To summarize, I found 63 good players, 33 at second and 30 at short.  I also found that there wasn't a specific team that was more or less adept at finding these players, as the sample is so small that no team really stands out.  In the end, 30 shortstops in 30 years seemed like an impossibly low number.  However, the number on its own isn't all that useful, unless it can be compared to other positions. 

Therefore, I decided to investigate if the middle infield positions have been harder to fill with good players over the past 30 years.  Career numbers aren't as important to me, so I focused on seasonal data.  If a team has five or six good, young middle infielders that they rotate through, they may not produce good career players, but would still get good value at those positions on a relatively regular basis. 

I looked at each offensive position, year-by-year, over the past 30 seasons.  I wanted to know how many players would be considered "good" in each year and how "good" the best players were during those 30 seasons.  I decided to use WAR again.  Terrible toothbrush analogies aside, I do think WAR is the best available measure that we can use to compare players from 2012 to 1983.  It isn't perfect, but it gives us some information, which is better than just throwing our hands up in the air and crying under the bed. 

FanGraphs.com has an excellent explanation of how they calculate WAR.  You can start here, and choose to read as much as you'd like.  I read through the all the pages, and I feel that I understand their methods very well.  FanGraphs outlines a player with a 2+ WAR as being a solid starter or better, which is a number I will be using as the basis of this analysis.

Here are the two factors I investigated:

•         How many players provided a 2+ WAR at each position, in each season?  This was an easy way to look at how many good players each position supplied.
•         What was the cumulative WAR for the top ten players at each position?  This allowed me to compare the actual value given at each position.

Here is my resulting chart:

Year	# of 2+ WAR	2+ WAR/ Season	Top 10 Total WAR	Top 10 Avg WAR	Top 10 Player Avg WAR
C	417	13.90	1131.8	37.73	3.77
1B	564	18.80	1512.1	50.40	5.04
2B	487	16.23	1307.6	43.59	4.36
3B	569	18.97	1472.8	49.09	4.91
SS	442	14.73	1308.6	43.62	4.36
LF	592	19.73	1475.2	49.17	4.92
CF	617	20.57	1541.9	51.40	5.14
RF	624	20.80	1427	47.57	4.76
OF	1625	18.06	4389.8	48.78	4.88

I am terrible at labeling fields, so let me just clarify:

•          # of 2+ WAR = Number of players with 2+ WAR in the last 30 seasons
•          # of 2+ WAR/Season = Average number of players with 2+ WAR, over 30 seasons
•          Top 10 Total WAR = Total WAR from all top ten players, added together from the last 30 seasons
•          Top 10 Avg WAR = Seasonal average for the top ten players (as a whole), by WAR
•          Top 10 Player Avg WAR = Average WAR for a top ten player, over the past 30 seasons

I broke down outfielders by individual position, but also by all outfielders.  I did this because many outfielders played multiple outfield positions.  The OF value is most accurate, but the LF, CF and RF values are still interesting to look at.

The results show that the two middle infield positions are pretty hard to find, compared with all other positions other than catcher.  Catcher is the hardest position to fill, according to these findings.  The number of 2+ WAR second basemen was higher, but the overall WAR value was almost identical.  So, it seems to be easier to find second basemen, but good shortstops provide more value.  In short, the average top ten middle infielder between 1983 and 2012 provided 4.36 WAR.  

Teams get roughly 10-15% less value from the middle infield, compared with corner infield and outfield.  Of course, teams get even less value from catcher.  Having Joe Mauer on the team over the last 9 seasons has been a really big boost for the Twins.  Just finding a good catcher is hard to do, but having a consistently good catcher is extremely hard to pull off.  Of course, that is a separate topic. 

The average top ten middle infielder provides about .5 WAR less than the average top ten player at nearly every other position.  This shows that teams are often playing middle infielders that are not only below average, but below the quality of players at other positions.  Something to keep in mind about WAR is that the stat already gives players credit for playing these difficult positions, so some of their value actually comes from simply playing the position, saying nothing about how well they play it. 

Basically, if two players are completely equal in all ways - batting, slugging, baserunning, defense, etc, the shortstop will have a higher WAR than the first baseman, simply for playing the more difficult position; shortstop more than second base, but still.  So, the actual production of these players might even be worse than would be indicated in my chart.  By giving these middle infielders an automatic bump, their WARs are artificially inflated (on purpose).  In addition, the first basemen and corner outfielders are actually penalized in WAR, ever widening the gap. 

There is logic in the positional adjustments made when calculating WAR.  However, these adjustments show that if the sample of shortstops and first basemen are about .5 WAR apart, their production might actually be much farther off than one might think, just looking at WAR.  Finding a good shortstop or second baseman is clearly hard to find, and that is why these players get a boost in their value. 

The actual production is important though.  We can see that over the past 30 seasons, the middle infield has provided less value than most of the other positions and that good middle infielders are clearly rarer than good players at most of the other positions.  While this may seem like an obvious conclusion, having the research and data to support the conclusion is very important. 

The next step is to investigate the samples of good second basemen and shortstops.  Where does their value come from?  Why are they more rare than players at other positions?  I'm working on analyzing some data and I hope to share it with you very soon.  Stay tuned!

Bonus!  Here is a chart of the top 20 players by WAR, last season.  There were 9.5 outfielders (Zobrist was the .5), 5 third basemen, 2 catchers, 2 second basemen, and 1.5 shortstops (Zobrist again).  There was just one middle infielder in the top 10.  Just thought it was an interesting snapshot:

Name	WAR	Pos
Mike Trout	10	OF
Buster Posey	8	C
Ryan Braun	7.9	OF
Robinson Cano	7.8	2B
David Wright	7.8	3B
Chase Headley	7.5	3B
Andrew McCutchen	7.4	OF
Miguel Cabrera	7.1	3B
Jason Heyward	6.6	OF
Adrian Beltre	6.5	3B
Yadier Molina	6.5	C
Aramis Ramirez	6.5	3B
Michael Bourn	6.4	OF
Aaron Hill	6.2	2B
Martin Prado	5.9	OF
Ben Zobrist	5.9	OF/MI
Alex Gordon	5.9	OF
Austin Jackson	5.5	OF
Ian Desmond	5.4	SS
Torii Hunter	5.3	OF

Finally, here is a link to the spreadsheet I created while doing this research.  It breaks down most of this data annually, and it is pretty interesting to look at:  Raw data

Addendum:

Someone on the Twins' Daily Forum asked about middle infield offense.  Here is what I found:

wRC+ is my metric of choice. wRC+ is slick because all the factors that create runs are put together and then the values are normalized and adjusted for park factors and era. Thus, you can compare Greg Gagne to Brendan Ryan, to see who was the better hitter. The metric is based off of the number 100, where 100 is average and each corresponding number is a percentage above or below. So, Gagne's career wRC+ is 82, meaning he was 18% below league average for his career. 

I decided to look by position, to see if good, offensive shortstops are particularly rare. I decided I had no use for players with fewer than 500 games played. Roughly three seasons worth of games seems like a decent sample size. Here is my resulting chart:

	# of Players	Good Offense	Good %
C	238	46	19.33%
1B	230	137	59.57%
2B	284	52	18.31%
3B	249	84	33.73%
SS	239	24	10.04%
LF	341	150	43.99%
CF	269	76	28.25%
RF	296	143	48.31%
	500 games min

n is the number of players, "good" refers to players with above average offensive production (wRC+ > 100) and "good %" shows what percentage of each sample met that criteria. As you can see, the further down the defensive spectrum you go, the harder it is to find a good offensive player at that position. This is hardly breaking news. The part that surprised me was just how small the sample was at short. There were only 24 players from the past 30 seasons who posted a career wRC+ over 100.