A couple of weeks ago, I posted SwimSwam’s 2018 updated Swimulator program rankings. With the help of some insightful user feedback, we have added a couple of revisions to our formula. The revisions to the formula include:
- The rankings now encompass data from 2010 to 2018
- Recent seasons are more heavily weighted than older seasons
- A bug was corrected where some year’s times from nationals weren’t reflected in team strength
First off, we added three more years worth of data so that the new program rankings encompass the past nine NCAA seasons. We avoided going any further back as that would run into the tech suit era. While going through the data, we saw that team strength was a far more stable metric than either team attrition and team improvement. Many of the teams that improved their swimmers steadily for a long period of time regressed more towards the mean when looking at more data. We believe that a longer period of time should help counteract this. Nebraska’s women’s team, topping our last rankings, was a notable drop due to this change .
When taking more data, teams that have improved recently, but where not quite as fast or had higher attrition rates in the early 2010s, fell in the rankings after incorporating the new data. NC State comes to mind here. To counteract this, I put in a discount so that more recent seasons were weighted more heavily than older seasons. The discount metric used – an exponential discount of 9/10 per season – is admittedly arbitrary. However, we thought it struck a good balance. It ended up weighting the 2010 season about 1/3 as heavily as 2018 and the 2013 season 1/2 as much. Below are the updated top ten rankings for NCAA DI programs for women and men.
Top 10 NCAA DI Women’s Programs
| Rank | Team | Combined Score | Strength Rank | Team Strength | Attrition Rank | Attrition Rate | Improvement Rank | Improvement % |
|---|---|---|---|---|---|---|---|---|
| 1 | Georgia | 3.8 | 1 | 680.379 | 26 | -0.131 | 96 | 0.189 |
| 2 | Stanford | 3.79 | 3 | 676.389 | 3 | -0.1 | 129 | 0.067 |
| 3 | Texas A&M | 3.58 | 4 | 646.919 | 78 | -0.173 | 41 | 0.365 |
| 4 | UMBC (W) | 3.44 | 74 | 322.0 | 1 | -0.074 | 3 | 0.878 |
| 5 | California | 3.38 | 2 | 676.716 | 33 | -0.135 | 139 | 0.037 |
| 6 | NC State | 3.13 | 22 | 482.302 | 30 | -0.133 | 24 | 0.505 |
| 7 | Virginia | 3.08 | 6 | 608.827 | 32 | -0.134 | 119 | 0.104 |
| 8 | UNC | 3.06 | 12 | 558.234 | 27 | -0.132 | 79 | 0.239 |
| 9 | LSU | 3.05 | 24 | 461.24 | 9 | -0.104 | 33 | 0.411 |
| 10 | Missouri | 3.02 | 21 | 499.27 | 19 | -0.122 | 44 | 0.357 |
Top 10 NCAA DI Men’s Programs
| Rank | Team | Combined Score | Strength Rank | Team Strength | Attrition Rank | Attrition Rate | Improvement Rank | Improvement % |
|---|---|---|---|---|---|---|---|---|
| 1 | California | 4.1 | 1 | 658.244 | 17 | -0.136 | 40 | 0.381 |
| 2 | Texas | 3.48 | 2 | 649.664 | 24 | -0.146 | 86 | 0.17 |
| 3 | Louisville | 3.42 | 11 | 519.026 | 10 | -0.127 | 24 | 0.479 |
| 4 | Michigan | 3.26 | 4 | 605.343 | 44 | -0.168 | 62 | 0.293 |
| 5 | NC State | 3.13 | 13 | 515.319 | 38 | -0.165 | 18 | 0.512 |
| 6 | Florida | 3.12 | 3 | 620.971 | 70 | -0.195 | 63 | 0.291 |
| 7 | Air Force (M) | 2.9 | 32 | 401.727 | 46 | -0.172 | 3 | 0.8 |
| 8 | Missouri | 2.77 | 18 | 472.197 | 58 | -0.185 | 9 | 0.568 |
| 9 | Georgia | 2.75 | 9 | 549.955 | 9 | -0.125 | 115 | 0.075 |
| 10 | Ohio St | 2.67 | 10 | 525.342 | 87 | -0.224 | 19 | 0.511 |
The updated rankings adhere a bit more closely to traditional powerhouses. Georgia, Stanford, Texas A&M, and California all make the top five women’s programs and Texas, Cal, Michigan, and Florida all make the top six men’s programs. UBMC’s women’s team did manage to make the top 10 with only the 74th fastest team due to their impressive attrition and improvement scores. Strength is still weighted as heavily as the other two factors, but it ended up being by far the most important distinguishing factor. Between the other two factors, it looks like attrition rates were more of a distinguishing factor than improvement.
While the program rankings aren’t available on a year-by-year basis currently, individual team’s strengths and improvement data is on the Swimulator page. We plan on leaving the rankings until we have new data for next year!
1) Not all improvements are equal. When you are already elite, it is that much more difficult to drop time.
2) What about going Pro? Does Ledecky and Manuel going pro and not swimming 4 consecutive years contribute to attrition and therefore a negative?
1. I thought so too, but it turns out that in college, slower swimmers and faster swimmer improve at about the same rate: https://staging.swimswam.com/how-much-faster-do-swimmers-get-in-college/
2. They would. Though technically Manuel was already counted against Stanford’s attrition for her 2016 redshirt season while she trained for the olympics. Probably not fair to count against the teams, but I don’t think its typical for more than 1-2 swimmers go pro out of college in a year.
Love the information. Looking at the full site. Would love to see the raw analysis data for my team. Any way that is available?
Thanks! The raw data’s not available right now, but I’ll look into it.
I assume there was no deviated discrete enumerator application to your formula. The statistical verticality of the inverse proportionality factor or .0224 should have been integrated. Please revise by either using the Dogmatical Reverse or Empirical Radiconial Anaylsis System.
I am confused. How is the overall score created? I thought schedule strength, attrition and improvement were equally weighted. If I weight your rankings equally it should go as follows: Here are my back of the napkin results:
1. NC St 18.54
2. LSU 21.78
3. UMBC 27.72
3. Misso 27.72
5. UNC 38.94
6. Geo 40.59
7. T A&M 43.23
8. Stan 44.55
9. Vir 51.81
10. Cal 57.42
The problem with this top 10 is I am not going past top 10 for order. Very interesting stuff. Let me know where I am going wrong
The overall score is created by adding together the z-scores – number of standard deviations from the mean – in each category. So they are equally weighted, but by using this metric, first in a category by a lot matters more than being first by a little bit.
Ok, thanks for the response. I understand z scores, use them in our own data analysis. Any way to display them?
Plus, I dont think “strength score” is the same as strength of schedule… you are thinking football and basketball and other sports with too many arbitrary rankings… strength I would assume is some kind of number formulated by NCAA and conference performance.
Just curious, if a person graduates with eligibility left does that factor in to the rate of attrition?
Any swimmer that swims at least one season, but doesn’t swim four years consecutively, would factor into the attrition rate. I think there are ways you can technically compete in at least one meet and not use up a year of eligibility, but basically yes.
So, for example, if a team appropriately red-shirts an athlete sometime In his/her career for any of a number of valid reasons, and yet he/she still completes 4 years of eligibility, your model treats that negatively towards a team ranking as “attrition”? Or if an athlete graduates in 3 years instead of 4, you consider that as “attrition” and a negative?
How do you deal with transfer students, who come with less than 4 years of remaining eligibility? Do they also get recorded negatively as “attrition” because they do not swim four years consecutively?
It is not at all clear to me that the notion of attrition as not completing 4 consecutive years of swimming is in any… Read more »
I agree…maybe use the numbers provided by the NCAA that actually takes into account graduating in 5 years instead of 4 as that has been the national trend among all students, not just athletes.
Attrition only gets counted towards swimmers who compete for an NCAA team one year, still have NCAA swimming eligibility left, and don’t compete for a team the next year.
Look, there are plenty of valid reasons for swimmers to compete one year and choose not to do so the next year, heath, grades, family stuff ect. But on the whole, it a program has a lot of people choosing not to compete on that team again the next year, its not good for the program as a whole. And as a prospective swimmer that would be a red flag for me. Maybe a school like BYU where a large number of students leave for missions in the middle of their… Read more »
So, by your judgement it was bad for our program when we redshirted two of our men’s team swimmers their 4th year while we were not eligible for the NCAA Championships, so they could help us go to the NCAA Championships in their 5th year as half of our medley relays and one individual qualifier? How is that a “red flag”?
And it is bad for our program to allow have homesick athlete to return to his home country and free up his scholarship money for someone who will actually do a better job for us?
Or it is bad for our program to allow a swimmer to redshirt after getting injured during the summer at home so… Read more »
I guess you keep adjusting until you get the results you want….
Only if diving shows up.
Of the top teams, is Texas really the one that needs to start hoping their diving performs?
No sour grapes just telling it like it is. Texas swimming was not up to Texas standards and had to rely on diving to win it. Texas fans don’t want to acknowledge that. Their little horns got bruised.
Yeah, it’s tough lifting four national championship
trophies in the air without bruising your Horns!
Congrats to Cal on winning the Swimlator Award!
Must be a nice trophy with lots of recognition attached!
UMBC women and Air Force men, huh? Okay…