Boyd's World-> Breadcrumbs Back to Omaha-> Adjusted OPS for 2001

About the author, Boyd Nation

Last week, I began a series where I'm taking a look at the best individual player performances for last year. I began by discussing why OPS (on-base percentage plus slugging average) was a good, reasonably simple measure that you can use to get a feel for hitter performance and gave you the top 25 list for last season. This week, I want to take that a bit further and discuss some of the reasons why that list is not the list of the 25 best hitters from last season, and try to get a little closer to identifying the best performances from last year.

Any time someone begins to look closely at raw numbers in baseball, at whatever level, they quickly begin to realize that the raw numbers, by themselves, don't necessarily help you all that much in comparing two players, especially players from different teams. The next step for that someone, assuming they don't just give up on statistical analysis completely and go back to observational determination (and wouldn't you be bored if I did that?) is to try to come up with an adjustment, or a series of adjustments, that makes the comparisons more fair. Obviously, the adjustments have to based on some measurable quantity, but there are several possibilities.

Here are three of the most common adjustments, along with the reasons why it doesn't make good sense or isn't possible to apply them in the context of creating an adjusted OPS for NCAA play:

• Adjusting for home-field park factor. Obviously, some parks are easier to hit in than others. I've talked before about park factors for NCAA teams, and you can get more detail from that column, but the short version is that it's a lot easier to score runs (and do all the things that contribute to scoring runs, including the things that show up in OPS) in some places than others, due to things like altitude and park size. One common adjustment to raw stats is to multiply by the park factor. Unfortunately, that's not a particularly fair thing to do in this case, since the data's just not there (and can't be there) to get an abstractly fair park factor. We can only compare a park to the other parks that the home team plays in, so we end up with results like San Diego State's stadium looking like an extreme pitchers' park, when it's actually a mostly normal park in a league full of high-altitude hitters' fantasies.

• Adjusting for defensive position. Some defensive positions are harder to play than others. Therefore, there are fewer players who can play shortstop, for example, than can play first base. It follows from that fact that, if two players have equal offensive ability, the one who can play shortstop is more valuable than the one who can only manage first. One frequent adjustment is to compare a player's raw stats to those of a theoretical replacement player at his position, so that shortstops and catchers get more credit than first basemen and right fielders. This would be fair to do for college ball, but I just don't have the positional data to compute what the adjustments would be.

• Adjusting for league average. Putting up on OPS of .900 for a first baseman in 2001 in MLB was about average. Putting up an OPS of .900 in 1984 would have been MVP-caliber play. The balance between hitters and pitchers shifts over time, and to compare players between eras, you really need to know something about the offensive context of the time. Fortunately, I'm not trying to compare players from different eras yet, since the data to compute even Division I-wide offensive averages just isn't there yet.

So, what does that leave us able to do to get a more accurate ranking? Well, if you look back at the raw list, you'll notice that only four of the players -- Griffin, Burke, Johnson, and Baker -- came from what I think of as the five power conferences (SEC, Pac-10, Big 12, ACC, and Big West, in no particular order). Now, I know baseball is a team sport and all that, but it just strikes me as unlikely that only four of the nation's best players play in the nation's best conferences, which contain most of the nation's best teams (nineteen of last year's top twenty-five in ISR, for example). So, it seems like a reasonable thing to do is to adjust for the strength of schedule played.

A reasonable way to do this, then, is to compute what I'm calling adjusted OPS, or AOPS, by multiplying OPS by the team strength of schedule divided by a theoretical average strength of 100 -- AOPS = OPS * SoS / 100. There's no reason that this should be a perfect measure, for a number of reasons -- it's not park-adjusted, there's no reason to think that it's 10% more difficult to put up offensive numbers against a 10% tougher schedule, strength of schedule doesn't differentiate between opposition offensive and pitching strength -- but here are the top 25 from the AOPS list for last year:

Team                     Player                   AOPS    OPS

Tennessee                Chris Burke              1.494  1.352
Kent                     John VanBenschoten       1.479  1.533
Florida State            John-Ford Griffin        1.473  1.356
Nebraska                 Dan Johnson              1.441  1.327
Memphis                  Daniel Uggla             1.356  1.288
McNeese State            Kevin Mitchell           1.319  1.286
Clemson                  Jeff Baker               1.315  1.232
Northern Iowa            Ryan Brunner             1.301  1.263
Cincinnati               Kevin Youkilis           1.301  1.263
Utah                     Chris Shelton            1.299  1.212
Georgia                  Jeff Keppinger           1.292  1.171
Southern                 Michael Woods            1.278  1.479
Alabama                  Aaron Clark              1.278  1.153
Mississippi              Burney Hutchinson        1.277  1.162
Baylor                   Kelly Shoppach           1.274  1.128
Oklahoma                 Greg Dobbs               1.264  1.145
Washington State         Stefan Bailie            1.263  1.100
California               Matt Gecan               1.261  1.113
Pepperdine               Jared Pitney             1.254  1.131
Cal State Fullerton      Mike Rouse               1.252  1.072
Cal State Northridge     Robert Smith             1.245  1.117
Texas Tech               Austin Cranford          1.243  1.134
Auburn                   Gabe Gross               1.239  1.113
Cal State Northridge     J. T. Stotts             1.235  1.108
William and Mary         Brendan Harris           1.232  1.238

The raw OPS column is just present for reference, to show how large the adjustment is for strength of schedule.

I really like the way this list shakes out. It doesn't end up with the player that I expected on top (I expected Griffin), which is good, since I would have mistrusted a measure that looked like I had subconsciously manipulated it. It gives a mixture of big-program players and guys from smaller schools, which matches with what post-college development for past players tells us it should do. It's not perfect, for the reasons stated above, but I think it's the best we're going to do.

So, who's impressive? Well, Burke had an absolutely fantastic year -- when you consider that there's no positional adjustment going on here, so he led the field even without considering that he played shortstop reasonably well, it's even better -- and it's nice to see him on top of the list. Van Beschoten also had a remarkable year -- it's somewhat comparable to Todd Helton the last couple of years, where it's still good after you discount for context -- which makes the Pirates' decision to pitch him just another example of why they're the Pirates. It's nice to see Michael Woods still hanging in there -- Southern played a tough enough non-SWAC schedule to keep him in contention. It's also nice to see the big conference guys like Hutchinson or Gross recognized here, since they tend to get less attention than they should because of the level of competition they're facing.

Next week, pitchers.

Boyd's World-> Breadcrumbs Back to Omaha-> Adjusted OPS for 2001

About the author, Boyd Nation