Measuring the NAIA
In the last month, we’ve taken a whirlwind tour of NCAA Division 2 and Division III, estimating the quality of play relative to Division I, the brand of college baseball with which we’re most familiar. The only four-year colleges remaining are those of the NAIA.
Last year, 30 NAIA athletes were selected in the MLB draft, starting with Mets pick Kirk Nieuwenhuis at #100. In 2007, Beau Mills of Lewis-Clark State (we’ll be hearing more about them in a few minutes) was the 13th overall choice in the draft.
Part of the reason the NAIA attracts the talent it does is that it doesn’t have the same stringent academic standards as the NCAA. For this and other reasons, it’s a popular destination for transfer students from Division I. It is likely to become even more popular since D-1 has adopted a new policy that transfers between D-1 programs must sit out a full season before playing for their new school.
Inter-association competition
In my last two articles, I estimated the strength of Divisions II and III using the results of games between schools in different divisions. To come up with an approximate level of quality for the NAIA, we’ll do the same.
Division I teams played 143 contests against NAIA opponents, outscoring the smaller programs 1,331 to 817. That’s good for a .710 Pythagorean winning percentage. Nearly all of the games (89 percent) were played at the home fields of the D-1 teams, so we’ll adjust that downward to .676.
The average strength rating of the D-1 teams in those 143 games was .416. Thus, if we assume the NAIA teams were average relative to their level, the result is that the average NAIA program would be a .255-quality team if transplanted into Division I.
Words can’t convey my skepticism at this first finding—.255 is better than the .231 we came up with for Division II, and D-2 schools have more resources and send more than twice as many players to the pros. That said, let’s put a giant “clashes with common sense” asterisk next to that number and move on.
Finally, a man-sized sample
There’s lots of direct interplay between Division II and NAIA, so we can test my intuition that D-2 must be better.
In fact, in 2008, teams from those levels played 461 games against each other. The D-2 teams won by a total score of 3,290 to 2,807, good for a .572 Pythagorean winning percentage. A home-field adjustment (once again, the NAIA schools spend more time on the road) pushes that down to .562.
Again, we have to assume that the participating NAIA clubs were average relative to their level, but we do know that the average strength rating of the D-2 teams is .468. That tells us that, if an average NAIA team were shifted to Division II, they’d manage a winning percentage of about .408.
Adjust that to Division I, and you have a .172-quality team. Still quite a bit better than Division III, but clearly a level down from Division II.
One more division
I don’t have complete 2008 data for either D-3 or NAIA, but I did track a dozen or so teams at each level through the full 2008 season. That resulted in 37 inter-association games between the best teams in these two categories.
NAIA teams outscored their D-3 counterparts 309 to 208, for a Pythag of .674. (The home games were about evenly split.) Recently, I estimated the D-3 level to play to be roughly equivalent to that of a .125 team in Division I, which suggests that an average NAIA team would slot in at about .228. Still a little high for my taste, but only a bit higher than the midpoint between our previous two estimates of .255 and .172.
Lewis-Clark State
I keep talking about the “average” NAIA team, but frankly, I don’t much care about the average NAIA team. Almost any team with a draft prospect is going to be better than that, and since top players head to a few of the best programs (Lewis-Clark State for Mills, Azusa Pacific for Nieuwenhuis), the value in this exercise is largely based on what it tells us about those perennial contenders.
So, as I did with Mount Olive for D-2 and Trinity College for D-3, let’s take a closer look at what these numbers would say about Lewis-Clark state, the three-time reigning national champion of the NAIA.
Last year, LCSC went 58-7, though if anything, that understates their dominance. Fewer than half of those losses came against NAIA opponents. In their NAIA games (which include postseason results), they outscored the other team 278 to 97, for a Pythagorean winning percentage of .873. I don’t have a good enough feel for the quality of specific NAIA teams and conferences to estimate a strength of schedule for them; we’ll stick with .873 as an approximation of their skill level relative to the rest of the NAIA.
Using that assumption, here’s how LCSC comes out relative to Division 1 based on the estimates we’ve reached so far. I’ve included the resulting rank among 297 D-1 teams and a school that is in the same place in the rankings:
if NAIA average is: LCSC is: D1 rank roughly equal to 0.172 0.588 74 Washington 0.194 0.623 58 Arkansas 0.228 0.670 36 Fresno State 0.255 0.702 23 Long Beach State
These translations cast even more doubt on the .228 and .255 estimates. (Again, those put the NAIA as even with or better than Division II.) The second-lowest level of .194—so chosen because .194 is the average, weighted by games, of the three estimates—is plausible, and as we’ll see in a moment, setting LCSC even with the University of Washington is even more appealing.
Head-to-head
Working with Lewis-Clark State gives us a big advantage over any other NAIA team. Unlike our other division champions, Mount Olive and Trinity, LCSC actually played a number of games against D-1 and D-2 opponents.
Let’s start with D-1. LCSC took on Washington and Gonzaga twice each. They split the pair with Washington, and the D-1 school outscored them 12-10. LCSC swept the season series with Gonzaga, though, winning by a combined score of 16-10.
Thus, Lewis-Clark outscored their D-1 opponents 26-22, for a pythagorean winning percentage of .576. Both Washington and Gonzaga are well above average—mixing their strength ratings, we get .568. That implies a rating for LCSC of .641—not too far from the weighted average estimate of .194 for the average NAIA squad.
We have more data to work with between LCSC and Division 2. The Warriors played 10 such games, including six against Dixie State, approximately the 35th-best D-2 team last year.
In those contests, LCSC outscored their D-2 counterparts 109 to 61, for a pythag of .743. Home-field adjusted, we get .731. The D-2 teams average out to .602 relative to their level, so based on these 10 games, LCSC looks like an .804-quality D-2 team, or a .552-quality D-1 team.
Calculate a weighted average of the D-1 and D-2 head-to-head results, and it looks like LCSC is a .577-quality D-1 team, which implies that the average NAIA school would come in at .166 against average Division I competition.
Not bad
If you’ve followed along this far, you recognize that we’re making a lot of assumptions, and while most of them are probably in the ballpark, some of them must be off. To me, getting results of .166 and .172 from somewhat different approaches to the problem is a pretty good sign. Until I gather more data, I’m inclined to use .172 to translate between the two levels.
It doesn’t surprise me that in each method, the most NAIA-friendly results come from the games they (or just LCSC) played against Division I. As I noted in analyzing other divisions, it stands to reason that the smaller-school teams would see those games as big opportunities, while the D-1 teams would look at the contests as mid-week exhibitions.
By way of summary, here’s a table with the results for D-2, D-3, and NAIA, all relative to Division I quality of play. I’ve included numbers for the best team (Mount Olive, Trinity, and Lewis-Clark), along with an approximate threshold for the final eight to provide an idea of what a “really good, not great” team looks like.
Level Average Final8 Best Division 1 0.500 0.732 0.824 Division 2 0.231 0.497 0.698 Division 3 0.125 0.330 0.500 NAIA 0.172 0.410 0.588
When the next Beau Mills come along and crushes NAIA pitching to the tune of .458/.556/1.033, we’ll have a better idea of what that means.