Monday, November 19, 2007

Graphs Gone Wild

It may look like a mess, but the 2007 Dodgers' "Performance/Salary Map" presented below is teeming with brilliant insight. So please bear with me (click on graph to enlarge):

Here are the basics of what's going on above:

  • The on-field performance* and annual salary** of each player on the 2007 Dodgers*** is plotted along the horizontal and vertical axes, respectively.
  • Blue is for hitters, red for pitchers.
  • The players' 2004 performance/salary is marked by the small dot.
  • The players' 2007 performance/salary is represented by the larger circle.
  • If a player was not in the majors in 2004, he has only the circle, not the dot
  • The size of the circle corresponds to the age of the player.
  • The players' 3-Year Trendline (2004 to 2007) is marked by the arrow connecting the dot and the circle.

*Measured by OPS for hitters and ERA for pitchers.
**From USA Today.
***Only players with at least 400 AB's or 80 IP's in 2007 are included, with Saito (64.1 IP) the only exception.

So what does this tell us? Well let's first look at the four primary quadrants, which players fall into each, and what it all means:

  • Upper-right quadrant (Stars) - This represents players who are paid well and perform accordingly. Jeff Kent and Brad Penny are the only players who fall into this category.
  • Lower-right quadrant (Bargains) - These guys have a relatively low salary yet put up strong numbers. Not surprisingly, all the players here in their 1st or 2nd year (Saito, Martin, Broxton, Ethier) and thus can still be gotten cheaply.
  • Lower-left quadrant (Role Players) - This quadrant is where 2nd-tier players with 2nd-tier salaries land. And sure enough, the two players in this category (Hendrickson and Tomko) personify 2nd tier.
  • Upper-left quadrant (Busts) - These guys signed handsome contracts yet aren't getting it done on the field. Sadly, this is also the most crowded quadrant on the Dodgers' Map, with Nomar, Wolf, Lowe, Furcal, and Pierre elbowing for position.

Even more can be culled from noting the direction in which the players' 3-Year Trendlines are pointing. Again, there are roughly four categories:

  • Up and to the Right (Approaching Prime) - Presumably players approaching or in their prime who show on-field improvement and receive a corresponding improvement in salary. Lowe and Penny are the two Dodgers in this category.
  • Down and to the Left (Past Prime) - Theoretically, a trendline in this direction would represent a player past their prime - i.e., one who is showing a decline in both output and salary. Gonzo and Nomar are the two Dodgers in this category, so I guess we can get rid of the word 'theoretically' above.
  • Down and to the Right (Players Getting Screwed by Their Ballclubs) - These players are getting a raw deal salary-wise from their team, as they are being paid less over the years despite putting up better numbers. Shockingly, there were no Dodgers in this group.
  • Up and to the Left (Players Screwing Their Ballclubs) - Players who have managed to get larger contracts from their current ballclub despite performance decreases. Alas, this is where the plurality of the '07 Dodgers fall, with Pierre, Furcal, Hendrickson, Wolf, and Tomko. Yikes.

So I guess it doesn't bode well that both least desirable quadrant and the least-desirable Trendline category have the most Dodgers. But I guess since I'm too lazy to create Performance/Salary Maps for other clubs for comparison purposes, I can't make any hard conclusions. But to put things in some perspective, let's see where A-Rod's would appear on this Map:

(click on graph to enlarge)

So A-Rod's performance and salary are literally "off-the-chart". That, my friends, is the brilliant insight I promised at the top.


Orel said...

One look at that graph and I thought, Hey cool, EK posted today. Always enjoy some hot graph-on-graph action to start my week. Excellent job!

Why do you suppose Furcal has ascended into "Bust" territory? Of course he was playing injured last season, but he had a solid 2006.

Eric Karros said...

Alas, the data isn't aware of who's playing injured nor how the guy did last year. So Furcal's .688 OPS in '07 coupled with a huge salary (over $13m) puts him way in bust territory.

StolenMonkey86 said...

Sorry, but you guys have to be kidding me. You can't call a 3.60 ERA average at the same time as you call a .790 OPS average. 3.6 was about an average ERA in the American League in 1988.

I wouldn't call Lowe a bust - a league average pitcher (4.5 ERA in the new park factor 100 Dodger Stadium) with the run support he got would have a W-L of like 15-49 after 3 years in LA. He's about the only mistake on there, though.

Eric Karros said...

Good point S-Monk and Orel...maybe I should state some caveats regarding this analysis to make sure it's not over-interpreted:

1) Scale was determined by simply taking the highest and lowest value of qualifying players for each metric and using them to define the end values. This means that the origin is located not based on the median or average OPS/ERA/Salary of the group of players (as would be more mathematically sound), but rather on the approximate average of only the extreme high and low values. The primary goal was to fit it all on one page and look pretty. This also means I didn't try to equate a particular ERA with a particular OPS.

2) Where the dots/circles are plotted is a communication of raw data based on the players' ERA/OPS and salary, nothing more. It's oblivious to guys playing hurt, poor run support, stadium configuration, a great outfield arm, the fact that a guy only has to pitch to left-handed hitters, or any other factors that ERA/OPS do not consider.

3) The four quadrants were assigned based on what the salary/performance combo directionally means (i.e., a high-salary & low-performance directionally means a bust). But because the scale and origin were imperfectly determined as described in 1) and the OPS/ERA stats have the limitations described in 2), a guy falling into the "Busts" or "Stars" quadrant doesn't necessarily mean he is that.

jose said...

I would suggest using Win Shares instead of OPS and ERA, since they account for defense, as well as for the fact that some players play more games than others and are thus more valuable.

Cubnut said...

I thought this was great fun. Not perfect--is any baseball metric?--but totally original and quite provocative.

Good job.

Steve Sax said...

EK, I love it. Great work--you've sparked a lot of thought about business frameworks can be overlaid onto baseball, Porter's five forces???