Roster Continuity and the Iowa Hawkeyes
It seems that every off-season you repeatedly hear some variant of "this team should be good because they return everyone from last year's squad." Since Iowa will be one of "those teams" this year, I'm curious if there is a strong relationship between roster continuity and year-to-year improvement, and if so, how big of an impact does it have?

What I'm Doing
To approach this topic, I selected teams that, coming into the 2004-05 season, returned every player who scored at least 5 ppg the previous year. For example, Iowa's 2003-04 squad included three seniors who averaged at least 8 ppg, so this study does not include their program. The 2003-04 Syracuse team lost only the 3 ppg of Jeremey McNeil, so the Orangemen will be examined. Due to my own limited resources, I'm only looking at teams from the following conferences - Big Ten, Big XII, ACC, SEC, Pac-10, C-USA, and Big East.

[Update, 6:05 pm - I ended up with a pretty small sample, so I'll probably try something similar to this in the near future, but with easier requirements to meet. I'll let you know if I come up with any different conclusions.]

I will then see how much (if at all) each team improved according to a stat I call Efficiency Margin (EM), which is simply the difference between a team's offensive and defensive ratings. I prefer to use only conference games when calculating EM, and that's what I did here. (Learn about efficiency ratings in the HH Stats Primer.)

Dirty Work
When they started the 2004-05 season, the following nine major conference teams still had all players who scored 5+ ppg for them the previous year.

North Carolina
Wake Forest
Syracuse
Villanova
Illinois
Michigan State
Oregon State
Washington
Florida

This first table shows each team's Efficiency Margin (again, that's points scored - minus points allowed, per 100 possessions) for 2003-04 and 2004-05, as well as the net gain from the first year to the second.

Efficiency Margin Gain

Team	EM Yr 1	EM Yr 2	EM Gain
North Carolina	2.9	23.5	20.6
Wake Forest	2.5	16.5	14.0
Syracuse	2.4	7.8	5.4
Villanova	-0.8	11.5	12.3
Illinois	14.0	24.6	10.6
Michigan State	10.0	18.5	8.5
Oregon State	-7.7	-4.9	2.8
Washington	5.4	15.0	9.6
Florida	4.8	13.2	8.4
Average	3.7	14.0	10.2

Wow. Although the sample size is small, every team that returned essentially all of its players improved in the second season, and most of those teams were a lot better. Yes, the average is pulled upward quite a bit by North Carolina's improvement, but I would suggest dropping both them and Oregon State (which turns out to be convenient since they're also the sample's extremes). North Carolina welcomed an NBA-lottery-ready Marvin Williams, while Oregon State gave significant minutes to a transfer (former Hawk Nick DeWitz) and two freshmen. Neither of those teams possessed quite the roster continuity we originally set out to explore. Even with those teams removed, the group's average EM improvement is a robust 9.8.

How Did They Improve?
After seeing that all of these teams got better, my reaction was to find out if the gains were due to better defense, offense, or both. Are point guards more aware of where their scorers like the ball? do players get more comfortable with their coach's defensive system? Maybe shooters just improve with the additional year of practice? Well, you get the idea. Here's the breakdown of how each team got better in 2004-05. OE is improvement in points scored per 100 possessions, DE is improvement in points allowed per 100 possessions.

Breakdown of EM Gain

Team	OE Gain	DE Gain
North Carolina	6.8	13.8
Wake Forest	11.1	2.8
Syracuse	9.9	-4.5
Villanova	6.5	5.8
Illinois	6.6	4.0
Michigan State	2.3	6.2
Oregon State	2.3	0.5
Washington	6.0	3.6
Florida	1.4	6.9
Average	5.9	4.3

Well, that's not all that conclusive. Six of the nine teams improved more through offensive gains. It is interesting to note how much better North Carolina got defensively during their championship season. Check out how their D compared to the rest of the ACC in my PPP graph from the off-season. Is that defensive dominance due to the addition of Marvin Williams? The coaching influence of Roy Williams? I don't know the answer to that one, but it's a fun debate.

Does the Conclusion Make Sense?
We came into this post with the notion that teams generally get better when they bring back all their players. Today's data seem to reinforce the idea, though small sample size caveats obviously still apply.

At the individual level, I'm convinced that players generally see decreases in their turnover rate and foul rate as they get older. These improvements, all else equal, would lead to better team offense and defense when you're looking at the same group of players from one year to the next.

Other factors, like familiarity with teammates' tendencies and comfort with coaches' plans/ideas, are much less measurable. All in all though, it seems very reasonable for teams to improve when they don't lose any players.

What Does All This Mean for Iowa?
Iowa obviously fits the same description as the nine above teams, since Jack Brownlee is the only player missing from last year's team (he of the 1 ppg). Their EM of 1.5 (104.0 OE, 102.5 DE) puts them in a similar starting spot as North Carolina, Wake Forest, Syracuse, and Villanova from last year. They definitely don't have the athletes that some of those teams did, but they won't be facing them in their conference, either.

If Iowa gained the group's average spread of 10.2, their 2005-06 EM would shoot to 11.7. Of the 17 teams with a double digit EM last year, 10 advanced to the Sweet Sixteen, with two of the others (Cincinnati to Kentucky, Florida to Villanova) losing to double digit EM teams before they could make the second weekend. That's pretty good company.

Time to Get Confusing
A team with an EM of 11.7 would be expected to win right around 12 of their 16 conference games (assuming Iowa uses the same number of possessions as last year, and using the ol' Pythagorean Theorem to calculate expected winning percentage. I can email you the math if you're really interested [read: as nerdy as me]). That's pretty good, and probably higher than most Hawkeye fans' expectations for this year. The next table shows the "expected" record of a team with certain characteristics.

Gain - improvement over last year's EM of 1.5 (remember the 10.2 avg from the first table?)
EM - the resulting efficiency margin for 2005-06
80 W, 80 L - the expected record of a team that scores 80 ppg, assuming Iowa maintains its 65.6 poss / G avg
65 W, 65 L - the other end of the spectrum

Expected Win Matrix

Gain	EM	80 W	80 L	65 W	65 L
3	4.5	9.5	6.5	9.8	6.8
4	5.5	9.8	6.2	10.2	5.8
5	6.5	10.1	5.9	10.6	5.4
6	7.5	10.5	5.5	11.0	5.0
7	8.5	10.8	5.2	11.4	4.6
8	9.5	11.1	4.9	11.7	4.3
9	10.5	11.4	4.6	12.1	3.9
10	11.5	11.7	4.3	12.4	3.6
11	12.5	11.9	4.1	12.7	3.3
12	13.5	12.2	3.8	13.0	3.0
13	14.5	12.5	3.5	13.3	2.7
14	15.5	12.7	3.3	13.5	2.5

Hope I didn't lose anyone along the way. According to the table, if Iowa gains a meager 3 on last year's EM, they'd score about 4.5 more points per 100 possessions than they allow, and they'd be a 9-10 win team (just in the Big Ten). If they improve as much as the teams from last year, say by increasing their EM to 11.5, they should win 11 or 12 of their games. And if the improvement is really dramatic, like Wake Forest and UNC last year, Iowa would be expected to win 13 or more games.

I'm sure no one is going to hold their breath for that last range of figures. Any Hawk fan who watched last year's Capital One Bowl knows that it's a tad early to be asking for miracles. But the point is clear - teams who return all their players tend to improve quite a bit. Iowa returns all their players, so it's not unreasonable to expect this year's team to be a lot better and in the top tier of the Big Ten.

Hedging My Bets
Just don't blame me at the end of the year for getting your hopes up.
PermaLink |