The best D-I X-C programs ever, using simple math?

some good observations. We are not developing an award that is going to be on ESPN. Instead, we are looking at the question of the best XC programs and trying to rank them. There is not one choice, but a lot of dimensions and time periods etc.

The model that has been developed here is a great starting point, including the hard work of getting the data together. people have made a number of valid points, and there are ways to address some of the issues to that certain perspectives get more 'weight'. None of them is 'right', although some things can be weaker than others. Also note that it is useful to watch how stable the assessment is as different factors are adjusted in importance. The generalization that I introduced yesterday is a step in that direction but only one way to go. I allows different time periods, and different weighting of qualifying, performance etc.

It was brought up that regions differ in quality. One way to incorporate this element is to measure how well a region does that year and adjust the importance of qualifying by the success of the region. However, the mechanism for getting points etc. can also result in a strong region getting more 'at large' teams in, so someone that understands the details of that system probably needs to discuss how they see that factor.

The index used to rank does not have to be linear or even piecewise linear, although it is usually a bit easier to understand such a mechanism.

I have a bit of background in index number theory and practice but that is not a primary area, just one that I am aware of some of the issues.

Actually, "Some Runner Guy," I'd argue the model does what it says it can do -- rank the best *programs*. It's not trying to measure if a 70s Oregon team could beat a 90s Arkansas team -- although that would be fun (but probably fruitless) to try and do...

It's looking at a sh!tload of historical data and trying to see which program has been consistently the best over time.

This started from a "few beers" conversation where a buddy (non-runner) used Oregon as an analogy for the best running school in history. Keep in mind, for a whole host of reasons (Pre, Nike, etc.) most non-runners immediately think of Oregon when they think college track. That got me thinking, "Sure Oregon is very good, but are they really the best ever? Time to start crunching data..."

So the model's intent is to measure the success of the program, and because we believe it is "harder" to win today, the more recent wins should count for more... whether they should count AS MUCH more as we have weighted them is open for debate.

I would say this overweights appearance, as evidenced by the fact that it's almost linear between appearances and overall rank.

Also, why is it surprising that BC doesn't rank highly? I've never understood the perception around NE that BC is a strong men's XC program.

Can you explain:

(1) How you arrived at 347 for a competitiveness factor for 2010? In your explanation you say it's "based on the total number of D-I schools and some minor intangibles related to competitiveness". Can you fill us in on the math there?

(2) How you decided to you use .9925 as the ratio by which the competitiveness factor decreases each year? I'm not asking why it decreases, you already explained your logic in FAQ #2. But why .9925? Is it arbitrary or is there a reason behind it?

looking at this wrote:

Can you explain:

(1) How you arrived at 347 for a competitiveness factor for 2010? In your explanation you say it's "based on the total number of D-I schools and some minor intangibles related to competitiveness". Can you fill us in on the math there?

(2) How you decided to you use .9925 as the ratio by which the competitiveness factor decreases each year? I'm not asking why it decreases, you already explained your logic in FAQ #2. But why .9925? Is it arbitrary or is there a reason behind it?

It does not make a lot of difference. You need to get the spreadsheet and work with it for a while before making random complaints. If you are not willing to do that then you really need to take a secondary place in the conversation. If you are not capable of working with the data then how are you really gone to understand the set of issues involved in terms of how they impact the net ranking results.

Again, the model is there for people to work with and to make their own assumptions and derive results. I think that they will eventually make a more generalized model available, whether it is the first cut I took or that of someone else with good skills, interest and knowledge of the sport. It would be interesting to see what malmo would do with this. He clearly has the analytical tools and mind, as well as the background etc. I suspect he is not as enamored of models and rankings as some of us; that might not be a bad thing.

"Looking At This" those are fair questions because those metrics -- in my mind -- represent the root of any controversy.

I'm hopping in a meeting, but I'll try and explain in more detail later today. Responding in the little LR chat boxes is kinda bothersome, so I may write out some explanations based on everyone's comments and just post it to the site.

I'll keep you posted either way.

And 26mi235 is right; it kind of doesn't matter because any weighting measurement is arbitrary (mathematically speaking) but logically speaking we wanted to apply weights that made sense.

~RM

In looking at the rankings, I agree with one of your previous posts stating that getting into the meet was weighted to greatly. I believe that getting into the meet is important, but if you're consistantly finish in the 20's than all that could indicate is that you're in a weak region. This is basically why a team like Georgetown is ranked so high. They are always in the meet, but are rarely a relivant team at the meet.

Maybe a better way of scoring it would be to only start giving out points for a top 15 finish. This way, the region you're in no longer matters. If you're finishing in the top 15, chances are you'd make it out of any region.

Great study/analysis by the folks at Faster than Forty!

As a Wisconsin Badger Alumni ('02) I have to say you got it spot on!

Great questions also asked as far as methodology.

I encourage everyone to like Faster Than Forty on Facebook to keep up with all their great articles and reports.

Of course I did look at the spreadsheet and follow the calculations from cell A to cell B, and that's where my questions came from. How else would I know that .9925 is the factor by which the competitivness factor decreases each year? That's not written anywhere except in the excel formula. I'm not sure where your knee-jerk reaction came from?

RMiller wrote:

Hi Rick. I'm the primary author of the rankings. The factors you cite are interesting, but a little hard to quantify. (Hell, getting the data just to rank the schools based on performance over the last 50 years was hard enough.)

All credibility went right out the window when you base your scoring off of this gibberish:

Q. The competition multiplier goes up by a few points each year. Why? Doesn’t that devalue older performances?

A. Yes it does. But we think that’s justified. Collegiate cross country gets progressively more competitive as more schools join Division I and as the fluidity of information exchange becomes more prominent. In the 1970s, perhaps only a handful of coaches understood the science of training elite distance athletes – but as more information becomes freely available via the Internet, the gaps in coaching knowledge narrow. “New School” is tougher than “Old School” in this case.

Hmmmm... So, Dumkopf, you think the level of competition has NOT changed over the 50 years of data we worked from...? I'd be curious to hear this rebuttal.

You're not one of those guys who is all, "My school's hockey team was ranked 15th this year!" -- when there are only 59 NCAA D-I hockey teams to begin with, are you?

Same rule applies to X-C... There were fewer teams 20-, 30-, 50- years ago than today. Any weighting needs to reflect that.

~RM

I have a suspicion that 26mi235 is Elliot Krause. The poster's understanding of mathemical concepts and insight on the Wisconsin teams seems to point to this.

BC is not a strong mens XC program unfortunately. Lack of scholarships and funding for the program has led to them not being in contention for nationals for quite some time. In recent years bad coaching has led to the program being on the chopping block

Change the damn assumption to one that you think is more realistic and report your results (including indicating what your change is in the parameters.

I change the temporal element so that the weight is even across years and there is no change in the top five teams in any of the last 50/40/30/20/10-year rankings, and you get a switch of Georgetown and Tennessee because TN were in the rankings in the early years and not later. I have done a little bit of interacting this sensitivity analysis with the ones I was implementing wherein the weight was relatively larger for higher placings (for instance, 1 for entry two for 30th, 3 for 29th, etc.). In this case as well, there is not a large difference.

As you weight old years more heavily schools like Tennessee will do better and OSU worse, of course, but the 'decay of 0.0075 per year is pretty modest. As you weight high places more heavily, schools like Stanford that have not scored some many years as other schools but have had top placings rate higher. However, Wisconsin has both a large number of appearances and a large number of high finishes ever since the 1970s, and especially the 1980s and gets #1 ahead of Arkansas for most reasonable choices of weighting values. In part this is due to not finishing lower than 19th since 1971.

This made me laugh more than anything I have read on LetsRun in at least a week or two. Though understandable in terms of what most people think of math, you really do not know what math is all about. I think Krause has said he hasn't used numbers in a year or two. MATH, of the type Krause does, is not about something simple like spreadsheets and calculations, it is an entirely different world of differential geometry, abstract algebra, and more esoteric topics. Go read A Beautiful Mind (Nasar), made into the movie of the same name about the mathematician John Nash (game theory => Economics Nobel prize about 1996, essentially for his 37-page thesis in 1950).

[Besides, it is known by some that I officiate track meets, so that takes me out of possibly being Krause (and I am neither that smart or that fast).]

RMiller wrote:

Hmmmm... So, Dumkopf, you think the level of competition has NOT changed over the 50 years of data we worked from...? I'd be curious to hear this rebuttal....Same rule applies to X-C... There were fewer teams 20-, 30-, 50- years ago than today. Any weighting needs to reflect that.

~RM

Not saying I agree either way, but to play devil's advocate for a second, I could come up with a few arguments.

(1) More doesn't always equal better. It often does, but it doesn't have to. For example, from the Wikipedia "List of Division I institutions" page, I see at the bottom that the following schools have begun the transition to Division 1: Bryant University, University of Nebraska-Omaha, University of North Dakota, Seattle University, University of South Dakota, Southern Illinois University Edwardsville. Most schools have an XC team, so let's assume they will too. So in the near future (let's say by 2013) there will be 6 more DI teams.

This could mean a few things - either it's 6 more teams Wisconsin has to beat every year to stay the best, so we could argue that 2013 Wisconsin really earns those extra points and deserves them more than 2011 Wisconsin. Or maybe we could argue 6 more DI teams actually dilutes the talent pool. The top runner on each of these teams could have gone to Oregon and fought to be their #7 but instead there are more DI scholarships out there now so he ends up at U South Dakota, meaning Oregon is not as strong as it could have been. So the era bonus doesn't make much sense from this perspective.

Or there is a third possibility, mostly likely, that these 6 teams will have no national impact, so the only effect they are having is that the increased number of DI teams pads the competitiveness factor, thereby creating a larger gap between present teams and past teams, when really at the top level the increase in teams has meant nothing.

So more teams could mean more competition...or it could mean a dilution of the talent pool...or it could just mean that more crappy XC teams are in DI now and it has no impact at the top, so why are we giving today's top teams bonus points that past teams didn't receive?

(2) There is also this - it can be argued that running at the NCAA (and American) levels hasn't markedly improved in 40 years. I know we can't compare XC times, but look at our track depth (NCAA all-time lists from T&FN):

13:08.4 Henry Rono’ (Washington St) 04/08/78

13:15.77 Bill McChesney (Oregon) 05/16/81

13:16.98 Alistair Cragg’ (Arkansas) 04/30/04

13:18.12i Galen Rupp (Oregon) 02/13/09

13:18.46 Brent Vaughn (Colorado) 05/04/08

13:19.22 Rudy Chapa (Oregon) 04/07/79

13:19.79i Sam Chelanga’ (Liberty) 02/13/09

13:20.2 ——Rono’ 05/20/78

13:20.43 Josh McDougal (Liberty) 04/13/07

13:20.4i Suleiman Nyambui’ (UTEP) 02/06/81

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

13:20.63 Sydney Maree’ (Villanova) 06/02/79

U.S. Additions:

13:21.0+ Gerry Lindgren (Washington St) 05/14/66

13:22.32 Ryan Hall (Stanford) 06/11/05

13:22.54 Ian Dobson (Stanford) 06/11/05

13:22.6i Alberto Salazar (Oregon) 02/06/81

13:24.56 Jorge Torres (Colorado) 05/02/03

27:08.39 Sam Chelanga’ (Liberty) 05/01/10

27:28.48 ——Chelanga’ 04/24/09

27:28.64 Stephen Sambu’ (Arizona) 05/01/11

27:29.40 Leoard Korir’ (Iona) 05/01/11

27:33.48 Galen Rupp (Oregon) 04/29/07

27:36.2 Gabriel Kamau’ (UTEP) 04/24/82

27:38.1 Gidamis Shahanga’ (UTEP) 04/24/82

27:38.50 Dathan Ritzenhein (Colorado) 04/30/04

27:38.6 Zack Barie’ (UTEP) 04/24/82

27:41.05 Ed Eyestone (BYU) 04/27/85

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

7:46.47 Boaz Cheboiywo’ (Eastern Michigan) 05/02/03

U.S. Additions:

27:55.86 Josh Rohatinsky (BYU) 04/29/07

27:59.43 Craig Virgin (Illinois) 06/22/76

27:59.72 Ian Dobson (Stanford) 05/01/05

28:03.72 Sean Quigley (La Salle) 04/04/08

28:04.46 Tim Nelson (Wisconsin) 04/29/07

28:06.12 Alberto Salazar (Oregon) 06/01/79

28:08.12 Brad Hauser (Stanford) 05/07/99

I count 13 out of those 33 names as being more than 25 years ago. That's pretty significant, especially when we are trying to argue that each year that goes by is more competitive than the last.

One final point on this: let's look at that 1982 UTEP team that had 27:36 Kamu, 27:38 Shahanga, and 27:28 Barie (see above). That would have been the 1981 XC season. Not surprisingly they won NCAA's. Based on the stats, they earned 283 points for that victory. However, fast forward 10 years, and because we're assuming that each year between 1981 and 1991 was more competitive than the last, 283 points in 1991 is only good for 20th place at NCAA's. The freaking 1990's - low point of American running - has a bunch of probably 30-31 minute guys worth more than a team of 27:30-28:30 guys simply because of the era.

This is the problem with the baseline assumption that the present is always more competitive than the past, it can lead to ridiculous results such as this.

Honestly I'm not sure how to easily fix this. We can't compare XC times, and it would be too unwieldy to research every guy's track times on every team for the last 50 years. Maybe we should eliminate the competitiveness factor...or maybe it would be modified to look something like -

In 2010 the top 20 collegiate 5k times added up to 270 minutes (or whatever). In 1970 the top 20 times added up to 280. So we use the ratio 270/280 to compare those two years. Or maybe a ratio is the wrong idea, since 13:30 and 14:00 is a fairly significant different at 5k, but judging by the ratio it doesn't seem to be, 13.5/14 is fairly close to 1 (it's 0.96). Of course now we are getting into using track times to compare XC, and again, it requires a lot of research, 50 years worth of top performances at 5k and 10k.

There needs to be some compromise between blindly assuming each year is more competitive than the last (which I think is flat out wrong), and actually researching stats so that whatever present-past competitiveness factor we use reflects reality (which is labor-intensive). It's a tough one. I don't know, how about using world records or American records?

RMiller wrote:

Hi Rick. I'm the primary author of the rankings. The factors you cite are interesting, but a little hard to quantify. (Hell, getting the data just to rank the schools based on performance over the last 50 years was hard enough.)

I dunno whether the overall experience for a runner is better at Wisconsin or Oregon or Fairleigh Dickinson for that matter; the goal here was to quantitatively determine which schools had the most success.

~RM

Faster Than Forty

You wasted your time. This all the math you needed:1+2+3+5+6=17 UTEP 1981 the best NCAA Division 1 program ever.

The only programs that have a "brand" on college cross country are Wisconsin and Oregon. The rest come and go over the years.

If Wisconsin can win it this year that will make 4 different coaches that won titles there. That's a program.

The rest of them, including Stanford, Colorado, and Providence don't have that. Arkansas lost it. The greatest teams at UTEP and Iowa State and Washington State had their day.

Only Oregon and and Wisconsin always have been there year after year. A bad year for those teams is 15th place.

Are grad rates going to vary that much? Longevity of coaching staff? I have seen a bunch of awful coaches who have been there forever. Who cares about facilities, we're runners. We need a shower and a locker.

rick m g wrote:

This ranking seems like a one trick pony. How about incorporating other inputs...graduation rates, quality of facilities, longevity of coaching staff?