# Even smaller colleges

A couple of weeks ago, we took an in-depth look at the quality of play in NCAA Division 2. While it’s impossible to produce any precise results, it turned out that the average D-2 team would win a little less than one-quarter of its games against an average D-1 schedule, and the top dozen or so D-2 teams would be reasonably competitive in Division 1.

Based on recent draft results, there isn’t much pro-worthy talent at the lower levels, but there is some. In each of the last two years, MLB teams took about 20 D-3 players. Last year, the top pick from the division was Ryan Kulik, chosen by the Cardinals at No. 245 overall.

Because I haven’t collected as much data on D-3 baseball as I have on Division 2, it won’t be possible to get as close to an exact rate for translation for the level. However, we do have enough data to make some reasonable approximations.

### Division 3 versus D-1

Since I don’t have complete results for all 300 or so D-3 teams, I haven’t calculated strength of schedule and relative team strength ratings for all the schools at that level. However, a simple glance at won-loss records tells you that the talent is more widely dispersed and, by extension, the level of play is probably not as high.

Last year, for instance, national champion Trinity College (Connecticut) went undefeated in its regular season, losing only its second-to-last postseason game. Trinity finished 45-1. It wasn’t alone—a couple of other top-ranked teams put together winning percentages well over .850. This isn’t due to some aberration in pro-ready talent: Trinity did have two players drafted, but they went at No. 829 and No. 1,132, respectively.

Let’s dig into the numbers we do have. In 2008, D-3 teams played only 34 games against D-1 opponents. The D-1 teams outscored the smaller schools 373-184—a .785 Pythagorean winning percentage. If we adjust that for the fact that all but six of the contests were home games for the D-1 teams, that’s a .763 mark.

I don’t have strength ratings for the D-3 teams, but I do know that the 34 D-1 teams had an average strength of .424—quite a bit below average for their level, though not nearly as low as the group of teams that played games against D-2 opponents. For now, we’ll assume that the D-3 teams are average; I don’t have much to go on, but I can say that nationally ranked teams are not highly represented in those 34 games.

Running the numbers, then, it looks like an average Division 3 team would be a .186-quality team in D-1. That’s not too far from my D-2 estimate of .231, and it would rank above eight of the 297 D-1 teams.

### Division 3 versus D-2

Of course, 34 games is a pretty weak dataset. If we’re willing to make approximations based on other approximations, though, we can expand things a bit. I have records of 141 games between Division 2 and Division 3 opponents. We can combine the results of those games with the estimate that the average D-2 team would perform at a .231 level in D-1.

In those 141 games, D-2 teams outscored their oppponents 1,075-887, for a Pythagorean winning percentage of .587. Roughly half of the games were home games for the D-3 schools, so no home-field adjustment is required.

However, the quality of the D-2 teams is pretty low. My strength ratings suggest that the average D-2 school in that 141-game dataset was only .386. In D-1 terms, that’s somewhere between .136 and .173. Since those D-2 schools modestly outscored their D-3 counterparts, we can estimate D-3’s level of competition to be between .100 and .128, relative to Division 1. These numbers would put the average D-3 team among the three or four worst teams in Division 1, which may well be an adequate estimate of their level of play.

Clearly one of these first two estimates is wrong. As noted, we’ve made several approximations—in both cases, we assume that the D-3 teams included are average. We’re also basing the second estimate on other approximations—that of D-2’s quality of competition. My initial reaction is to put more weight on the second number, something between .100 and .128. Not only is the second dataset bigger, but I suspect that games between D-1 and D-3 opponents are not reflective of the abilities of those teams. Some big schools may see them as little more than exhibitions or practice games, while the smaller schools are given an opportunity to test themselves. We’ll revisit this theme in more depth when we get to the NAIA.

### The Trinity test

In assigning a quality level to Division 2, it proved useful to look more closely at Mount Olive, the national champion and best team on paper in the division. While still accepting less precision, we can do the same for Division 3 by spending some time analyzing the performance of its level’s best team—Connecticut’s Trinity College.

As I mentioned above, Trinity had an astounding season. It won its first 44 games and ultimately beat Johns Hopkins for the national title. Trinity won its share of close games, so its Pythagorean winning percentage isn’t quite the equal of its actual percentage of .978, but it’s still pretty impressive at .899, the result of scoring 404 runs and allowing 122.

While it did have quite a few games against tough opponents, as Trinity worked its way to the national title game, the school doesn’t play in a very tough conference. I’m stuck guessing here, but I’m going to knock that .899 down to .875 to account for the level of competition.

So before we look at some numbers, let’s consider what we know about Trinity. Like Mount Olive, the team was extremely, even historically, good compared to its level. Two of the team’s players were drafted, and three more were good enough to earn John Manuel’s stamp of approval as pre-season All-Americans for 2009. It’s an extremely disciplined, well-coached group of ballplayers with a few very talented athletes among them, but it’s tough to imagine them holding their own against anything close to top-tier D-1 competition.

Here’s what Trinity’s .875 strength rating suggests about the team at various estimates of the median level of D-3 quality. I’ve also included where each adjusted strength rating would rank Trinity among the 297 D-1 teams:

If average D-3 is… Trinity is… D-1 rank 0.100 0.438 203 0.125 0.500 153 0.150 0.553 102 0.175 0.598 70 0.200 0.636 51

Right away, I think we can throw away the last two possibilities. The small sample of the D-1 vs. D-3 dataset already cast some doubt on the .186 result. A rating of .598 puts Trinity about even with Clemson, a strong program in a tough D-1 division, though one in a off-year. The next level down is roughly equal to the D-1 rating of Santa Clara and Texas Tech—more plausible, but it still strains credulity to suggest that a D-3 school is as strong as a D-1 program with two third-round draft picks.

Once we get down to the .125 level, we’re looking at numbers in line with the estimate based on the D-2 vs. D-3 games—a larger sample, and one without nearly as much potential for bias. Best of all, a .500 translation starts to make more sense. A few similar D-1 schools are San Jose State, Southern Illinois, Akron and Dartmouth. The last two of those schools also had exactly two late-round picks.

It’s impossible to know without pitting a top D-3 squad against a battery of D-1 competition, but .438 seems low to me. That would put Trinity among the likes of Austin Peay, Penn State, Radford and Northwestern—a mix of strong teams in weak conferences and laggards in better leagues.

### Movin’ on down

So far, we’ve established that the average D-2 team is equivalent to about a .231-level D-1 squad. That puts the very best D-2 teams above .600 in Division 1 terms. Now we can further estimate that the average D-3 school would win about one-eighth of its games against D-1 competition, while the best D-3 teams could keep pace with the middle of the D-1 road.

As far as four-year colleges are concerned, this leaves us only with the NAIA to consider. We’ll take a shot at that in another couple of weeks.