News & Features

by Dallas Oliver, 9 February 2017

The USTA Collegiate Division recently launched new weekly polls ranking the Top 25 NCAA Division I Men's and Women's Tennis Teams. These polls are meant to bring more attention to the sport - augmenting the highly-regarded Oracle/ITA Rankings and amping up the level of excitement and interest in college tennis.

The first Top 25 poll came out yesterday, with the Florida Women and Virginia Men taking the top spots. For more about the Top 25, check out this Q&A piece we ran with the USTA's Scott Treibly.

Many highly-regarded tennis outfits are participating in the poll, including College Tennis Today, Parenting Aces, Texas College Tennis, Universal Tennis, and USTA Collegiate Division. Tennis Recruiting is also honored to participate as a voting organization, and we are excited to be part of it.

Ranking Approach

When we were invited to participate in the poll, we thought hard about the process we would use to order the college teams. There are many ways to go about it, and the various voting entites are all taking this process very seriously.

We decided to play from our strength. The basis for our voting is the same best-of-breed system we use to rank junior players based on head-to-head match play. College tennis dual matches can be treated similarly to junior matches since college teams play head-to-head matches. This similarity allow us to apply our Bradley-Terry based ranking system to the problem - rating and ranking college tennis teams using dual match results.

A few notes about our approach:

• We treat each team as an entity with wins and losses. We look at tennis teams like basketball or football teams. Our system looks at team records and team match results (e.g., 4-3). We do not consider individual performances on the various singles and doubles lines.


• We use our weighted algorithm that yields the most predictive results. To capture the best teams each week, recent results are weighted more heavily than results from earlier in the season, and teams earn more credit for dominant scores (e.g., 7-0) than close scores (e.g., 4-3). This is the same variant we use to calculate our successful tournament forecasts.


• We use only the current season's results. These team rankings differ from our player rankings in that we only use the current college season rather than an entire year's worth of data. Team rosters change year to year, and we want to give credit for the current team's performance. Some strong teams may not appear as high in our rankings during the early weeks because (1) they play weaker opponents early in the season, (2) they pick up an uncharacteristic loss or (3) they win some close matches which do not earn as much credit. As in any sport, there is more movement in our rankings early in the season, but things settle down as all teams develop more established records.

This is the approach we ran with, and we are excited by what we have seen so far.

SLAM·Tennis College Team Rankings

Although each voter's submission to the USTA poll is confidental, we have decided to make our votes public. Each week, we will post our own rankings of college tennis teams. You can view our rankings by visiting our new website - SLAM·Tennis.

Our first SLAM·Tennis College Team Rankings are now available, and we used these rankings as the basis for our votes in the USTA poll. We plan to calculate these rankings on Monday of each week using the previous week's play - and to make them available on Thursdays to avoid interfering with the Oracle/ITA and USTA releases.

Because all of the infrastructure is in place, we will launch weekly rankings for NCAA Division II, NCAA Division III, and NAIA in March - when there is more data available for those divisions.

Let's check out our initial rankings for 2016-17 - and also take a look at week-by-week rankings from past years to get a better idea about how things work ...

Figure 1 above shows the Table View of the final 2015-16 rankings that display the list for a single week. You can explore the rankings in this view as follows:

• For each team, we show our ranking with the underlying rating value, the number of spots the program moved compared to the previous week, win-loss record, strength of schedule, USTA poll ranking, and Oracle/ITA ranking.


• Rankings are published for both men's and women's programs in NCAA Division I, Division II, Division III, and NAIA. (Note that NCAA D-I rankings are the only 2017 lists available until March.) You can select gender, organization, year, and week using the dropdown boxes above the table.


• We have rankings for each week of the spring season. Weekly rankings - from early February to late May - can be chosen using the drop-down boxes.


• Interesting sorts are available. You can select any column header - Rank, Trend, Team, W-L, SoS, etc. - to sort by that column. For example, selecting the ITA column header will sort the teams by their Oracle/ITA ranking for that week.


• You can toggle to the Chart View (see below) by selecting the Chart box at the top right above the table.

If you switch over to the Chart View, you can see week-to-week differences for teams in the rankings:

Figure 2 above shows week-by-week rankings for the Top 20 college teams in each division. Note that the same dropdown boxes are available above the chart, and you can select teams to display in the chart using the legend below the table.

The chart for the 2016-17 year is not very interesting early in the season, but you can take a look at past years - like 2015-16 - to see how the chart will be updated week by week. You can mouse over any team to take a closer look at its journey throughout the season.

Check out both the Table View and the Chart View of the rankings. Those two views give you a detailed look at our rankings.

Behind the Numbers

We have explored the presentation of the rankings at SLAM·Tennis, but how do we come up with the numbers?

As mentioned above, we use our rating system to produce rating values for each team. You can read a description of our rating system by clicking here - and you can learn about much more in other articles from our Inside the Rankings series.

In a nutshell, our system considers all match results for a given season and produces one common set of ratings for all teams across all organizations. So, for example, our rankings include matches between NAIA teams and NCAA D-II or D-I opponents. After calculating ratings for all college teams, we filter out the programs in each organization - NCAA D-I, D-II, D-III, and NAIA - to produce the team rankings in our tables and charts.

Here are a few observations on the rankings and ranking process ...

• As our data of record, we use match results posted on the ITA's web site during the regular season, and we also use the NCAA and NAIA web sites during post-season play. The ITA web site is the definitive source of college tennis data. This policy is similar to the one we use for junior results where we monitor the USTA and ITF web sites.


• As mentioned above, there is more movement in the rankings early in the season since all team records are relatively short. You can observe this movement clearly in the chart views.


• Similar to our player rankings, the ratio between the rating values of two teams can be used to produce expected win percentages. The rating and strength of schedule metrics are also comparable across divisions.


• Teams are rewarded for playing strong schedules. Consider the Alabama women in the 2015-16 season. The Crimson Tide earned only a 5-18 record, but they did so against tough SEC opponents which produced the No. 1 SoS in the country. For this reason, their 5-18 record was still good enough to earn Alabama the No. 42 ranking at season's end.

We may change our approach in future seasons, but this is the approach we will run with for 2016-17 voting, and we are happy with what we have seen so far when applying this method to the 2014-15 and 2015-16 seasons.

We are excited to bring you these college team rankings. College tennis is a wonderful sport, and we hope these SLAM·Tennis Rankings can contribute to the buzz.

Leave a Comment