Ultiworld Ultiworld DG

Paving A Better Road To DIII Nationals

This article was written by Hill Balliet and Bridget Pranzatelli, players for Claremont and the TDs of DIII Warmup.

Among top DIII teams, there is a widespread belief that the bid allocation to Nationals fails to adequately represent the division’s talent. That is not too far from the truth.

For teams who fall just short of earning a strength bid to DIII Nationals, it can be easy to look at the lack of connectivity between regions and decide that the USA Ultimate algorithm could not possibly have been fully accurate. Since DIII teams rarely travel to out-of-region tournaments, the sample size of games played across regions in a given season is significantly smaller than that of Division I, where the top teams in the country will play each other multiple times each season. The resulting lack of connectivity between regions in DIII often skews the final rankings.

Not only does internal legitimacy suffer from skepticism in the bid allocation process, but external legitimacy, or how seriously DIII is taken by the ultimate community at large, suffers due to a lack of interest and media coverage. Increasing the frequency with which DIII teams travel to play out of region games would combat both of these issues by increasing connectivity thereby improving the bid allocation process and generating more newsworthy content in the division.

The graph below illustrates the connectivity issues faced by DIII, and hints at the value that teams traveling to out of region tournaments could bring to the division.

[caption id=“attachment_49713” align=“aligncenter” width=“1068”]

Figure 1: Chart of all 2016 DIII Men’s teams and the D-I teams they played.[/caption]

In Figure 1, the red dots represent teams that went to Nationals, and the lines between dots mean that the two dots on either end played each other at some point in the season before bids were allocated. ((Connecticut College is not pictured because they did not have any sanctioned games listed on http://play.usaultimate.org that occurred before bids were allocated.))

The physical distance between dots in the image roughly corresponds to how many games you must traverse to get from one to another, and the shortest path across edges to get from one team to another represents the series of games that most significantly affect the two teams’ relative rankings.

The graph below excludes teams that are not relevant to the shortest paths between Nationals-attending teams and shows more clearly how these teams were connected when the bid allocations were announced.

[caption id=“attachment_49715” align=“aligncenter” width=“1150”]

Figure 2: Graph of shortest paths between each of the men’s teams that attended DIII Nationals.[/caption]

Corresponding graphs for the women’s division are in the Appendix.

The two Nationals teams that are the farthest apart on the graph are Bryant and Colorado College, who are separated by six games as shown with the red dotted line in the figures. The distance between two teams is important because the farther apart they are, the more likely that their relative rankings are wrong. For instance, if each game in the path that separates them has a 20% chance that the less talented team will upset the more talented team, then there is a 74% chance that the link between Bryant and Knox, which determines their rankings compared to each other, contains an upset ((The number of upsets is measured by a binomial random variable on 6 trials with probability 0.2. Therefore the probability that zero games were upsets was (6 choose 0)*(1-0.2)^6 or 0.26, so the probability at least one upset occurred was 74%. Similar calculations were used to compute the probability of an upset linking John Brown and Georgia College.)).

While the distance between Bryant and Knox on the graph is an outlier because Bryant only played four games before the series, large degrees of separation are quite common. Even among teams that are relatively well connected to the rest of the graph, like John Brown and Georgia College, there can be upwards of four games separating them as shown by the solid red line. The probability that an upset links John Brown and Georgia College is 59%, which means that their rankings are more likely to be wrong than not. If top teams in each region are all this likely to be linked by upsets, the bid allocations will almost certainly not be representative of the true distribution of talent in the country.

Increasing the connectivity will increase transparency and accuracy in the rankings, thereby improving the bid allocation process and increasing the validity and level of competition at Nationals. By better ensuring that the top sixteen teams are represented at Nationals, rather than sixteen of the top thirty teams, DIII teams can make their championship tournament and, by extension, their division more legitimate.

Teams can also build legitimacy in the eyes of the broader ultimate community by traveling to play out-of-region games because top teams playing each other more often will generate more high quality media content in the form of tournament previews, recaps, and film (where applicable). If top teams compete more frequently throughout the season building up to Nationals (as in D-I), they build a storyline that people in and out of the division can follow. This makes everyone more interested and invested in the outcomes; if nobody knows what to expect from teams, there is no such thing as an upset or a rivalry, and games are not as interesting to follow. The increase in newsworthy content makes the sport more exciting and legitimate for both DIII players and the wider ultimate community.

Unfortunately, traveling to play out-of-region games is not a simple task for many teams in DIII since most receive little support or subsidy from their administrations. Because of this, traveling long distances can seem financially prohibitive for many. There are, however, measures that teams can take to mitigate the costs of traveling to tournaments. Group prices on travel and lodging can provide much needed relief, and fundraising can take a huge dent out of the cost of traveling to tournaments. Five recently published an article about fundraising that not only gives suggestions for how to fundraise, but also discusses how to leverage alumni networks to make those fundraisers more effective.

If you are a player on a DIII team, and you can relate to the frustration of feeling that your tier of ultimate is discounted by the wider community, take it upon yourself to address the problems that lead to that view. It is well worth your time to invest in seeking out top teams and contributing to legitimate competition within the league.

If you want to get a jump on that goal, even while much of the country is still covered in snow, check out DIII Warmup in southern California. The tournament features lined fields, observers, and Nationals-caliber teams in both the Men’s and Women’s division, all in the warm SoCal sun. Contact [email protected] to get more information and to make sure you are kept up to date on bid deadlines.

Appendix: Women's Division Charts

[caption id="attachment_49717" align="aligncenter" width="1174"] Figure 1A: Graph of all DIII women’s teams plus the D-I teams they played[/caption]

Connectivity in the Women’s Division is even worse than in the Men’s Division. Where it took 45 teams to connect the top sixteen in the men’s division, it took 75 to connect the those teams in the Women’s Division.

[caption id=“attachment_49718” align=“aligncenter” width=“1166”]

Figure 2A: Graph of shortest paths between each of the women’s teams that attended DIII Nationals.[/caption]

Above, the shortest paths between Puget Sound and Amherst as well as Elon and Valparaiso have been highlighted. Both pairs are separated by six games, so as with Bryant and Knox in the Men’s Division, if each game in the path that separates them has a 20% chance that the less talented team will upset the more talented team, then there is a 74% chance that the link contains an upset.

Originally published at: https://ultiworld.com/2016/12/01/paving-better-road-diii-nationals/

Great article Hill and Bridget. I love the data driven approach to looking at why the USAU algorithm is weak for DIII. In the past, people have brought up the idea of adding a human component do (DIII) nationals selection. Do you think this analysis suggests that the algorithm alone may not be sufficient, especially in DIII?

As a participant and follower of D-III ultimate over the past 6 years I agree with the sentiments of this article. The suggested way to solving this problem is for D-III teams to travel and play more outside of their region, but this can be even more difficult for a D-III program than I think the article gives credit.

Money --> Support for D-III clubs and organizations tends to receive proportionally smaller funds than their D-I counterparts. D-I programs also generally have a larger pool of alumni, family, and individuals from which to pull funds. D-1 programs can even pull money form their secondary and tertiary teams to help fund runs for their most competitive team.

Travel --> Sometimes it can be difficult to even afford the trips for in-region competition. My college team would routinely drive for 10 hours just to compete in sectionals. Regionals could add another 2 hours on top of that. There was one spring season where our team had only one sanctioned tournament out of the entire schedule that required less than 9 hours of driving. What is worse is that when we would attend these far flung tournaments we would often fin ourselves playing the same in-region opponents who were also committing a serious investment in the hope to play teams from outside the region. It continues to be one of my greatest points of frustration for my college team. So even though teams may travel long distances, the tournament seeding or setup can still stymie the overlap the D-III system is looking for. I assume programs in smaller regions or located at the edge of regions will not have quite the same difficulties.

Legitimacy --> The authors indicate that travelling and team overlap leads to more high quality media content followed by greater public interest and eventually greater legitimacy as the story lines follow. While I agree with this I also think that D-III schools still have a tougher time following this path to legitimacy than their D-! counterparts. First, this is in part because D-III schools often have fewer resources to make these large trips. Second, there are few premier tournaments focused on D-III competition. Sometimes D-III schools are invited to premier tournaments along with D-I teams (such as Missouri Loves Company) but this doesn’t do too much either, often resulting in shallow D-III teams outclassed by large, deep D-I programs (undercutting the legitimacy of talent pool in D-III) and reduced media coverage because the focus is on the D-I programs at the event with the strength and large talent pools that can carry them through the whole weekend. Third, as Charlie has mentioned in many of the podcasts, college ultimate is often easier to sell because there is a built in, passionate group of followers for each team composed of alumni and individuals that support the particular school. D-I schools have massive alumni fanbases compared to D-III counterparts. These fan bases translate to more clicks, likes, and interest in particular teams. D-III programs have smaller fanbases which can’t generate nearly as much interest.

The authors also suggest that D-III players invest time into seeking out positions on top teams and contriuting to legitimate competition within the ultimate community. From my first hand experience in tryouts and discussions with individuals who have tried out, top tier teams often ask during the application/tryouts at what levels of competition have people played. D-I programs are heavily favored over D-III, and while you have the chance to show your skills during a tryout I have little doubt that if two players of equal ability were presented but one had D-I experience and the other had D-III experience the captains would opt for the D-I individual.

I am very happy that this article was written, but I don’t think the community always appreciates D-III or recognizes the additional struggles they have to contend with.

I totally get this issue, and I appreciate the data in support of an argument for more established DIII infrastructure.

I’d just like to say that I think in the interest of gender equity, it’s concerning that the article lists the women’s division as an “appendix” which seems to be as an afterthought. As you explain, the women’s situation with this issue is even more severe, and I’m sure that structures of sexism have something to do with that. This piece highlights a legitimate problem that affects both the women’s and men’s divisions and I think it was a missed opportunity to bring both of those situations into the same conversation instead of centering the men’s play and separating the women to an appendix.

Thanks for this work, I appreciate it. I’d like to hear your thoughts.

That’s a great question. I don’t know what the best way to determine bids is as each way is going to have its inherent biases and flaws. For instance, human bid allocation or power rankings can better filter out anomalous games, but it often tends to be biased towards established programs and the logic behind rankings is not publicly available. I am convinced, however, that generating better data will make accurate bid allocation easier regardless of how it is done.

These are all real issues that many DIII teams face, and the authors are right to suggest that lack of connectivity undermines the algorithm. But I think we also have to be honest and acknowledge that DIII teams do not help themselves or the division gain legitimacy. I played in the men’s division and know more about it than I know about women’s DIII, so that’s what I want to talk about.

Two years ago, quarterfinalist Georgia College and semifinalist Bryant did not play 10 sanctioned games during the regular season. This past season, overall #1 seed Bryant did not play 10 sanctioned games either. Quarterfinalist Knox did not play a nationals-qualifier during the regular season, and eventual winner Georgia College played 1 nationals-qualifier during the regular season.

None of these teams has to travel 10 hours to get to sectionals, or even 5 hours to make a top-tier regular season tournament. Alabama-Huntsville and Union out of the southeast attended top tournaments like T-Town Throwdown and DIII Easterns; Georgia College did not. Wheaton and Valpo out of the Great Lakes attended DIII Midwesterns; Knox did not.

At a certain point, we have to acknowledge that choices matters. Money and institutional support are serious concerns that DI and DIII teams face. Travel is a concern all teams face coming out of the plains and mountain west.

But the teams we are talking about come out of the Southeast, Great Lakes, and New England. Travel is not a barrier - or at least, other teams from those regions seem to face it. Connectivity is about who you play, and who you play depends on money and which tournaments you choose to submit bids to. And when some of the division’s top teams - teams who would get bids to top DIII tournaments if they submitted bids to attend them - choose to not attend quality tournaments or play quality opponents, it should come as no surprise when those outside of the division do not take DIII ultimate so seriously.

We must push for greater institutional support and recognize that college ultimate can be prohibitively expensive for many college students and work to alleviate those costs where possible. But if we seek legitimacy and USAU rankings that actually capture the talent in the division, we must also push teams who can attend quality tournaments to attend them, and we must recognize that teams’ choices are critically important to the division’s success.

Just here to be honest, there really isn’t a DIII team that would compete at any DI regionals tournament (excluding metro east). The better DIII teams (top 10 at nationals, for arguments sake) would probably make their regionals DI tournament, but none (maybe with an exception depending on the team/year) would seriously compete. bracket play at pretty much any DI regionals is higher quality than DIII nationals. It’s why most people don’t care that much about DIII, same as any other college sport.

Hey GangBuster,

I played at a DI school in several different sports so I agree with you for the most part however to say that DIII teams can’t compete with DI schools consistently couldn’t be farther from the truth. I now coach at Davidson, a DIII school where I also had the absolute pleasure of coaching Kyle Taylor there, where I am 100% sure that we could compete with some of the best DI teams in the country. Just last year, we knocked off a top 25 team in pool play and lost to a team that made college Natty’s by one in a single tournament. To say that DIII teams couldn’t get past forequarters of DI regionals is ludicrous. It may be tough in the new SE region with perennial National semis teams like UNC and UNC Wilmington. But I promise you this, give us one more year and we will consistently beat your DI teams.

References: Michigan vs Appalachian State,
Duke vs Lehigh, UNC vs Weber State, Iowa State vs Hampton University, North Dakota State football vs the world