Spencer Widlicka

MAEL 504
Data collection and analysis

Data Analysis of DIII mens soccer for 2016 on national basis:

Collection Process:

I started by finding different statistical data comprised on a site called D3soccer.com. I

I also collected year to year data on a smaller scale for the Ohio Athletic Conference
(OAC). The OAC is a conference within the DIII level of mens soccer competition. When
looking at data in the conference level I sought to determine if other data points helped show
team success year to year. Some of the OAC data collected was shots per game, shots on goal
percentage, goals per game, and goals against average.

The final collection process was collecting data from top 25 national polls done by two
sites d3soccer.com and ncaa.com. Not much variation was found between the two end of season
polls but a total of 32 teams were listed in top 25 between the two polls. I also collected data of
participates in the NCAA tournament. I collected this data from ncaa.com and listed in the excel
sheet how far into the tournament each team went. Tufts was ultimately the winner of the
tournament for the 2016 season. For a team to get entrance into the tournament they must win
their conference tournament that year, which is held in every conference between the top
approximately 50 percent of teams in a conference. After the conference winners receive
placement for the tournament select teams are chosen based on record and strength of schedule.
For reference John Carroll received the bid from the OAC in 2016 by winning the conference
tournament but Ohio Northern also still got selected into the tournament based off of record, as
well as being recognized throughout the season as a top 25 national program.

Data Hypothesis and Analysis of data:

Below is a list of different inferences I made on the data followed by an analysis paragraph of
some of the areas to which the data provides insight.

1) Hypothesis: Teams with the highest winning percentages on the season will be widely
recognized as top 25 programs and perform well in the NCAA tournament.

Analysis: Every single team that was ranked in either D3 poll top 25 was in the top 25%
of winning percentage nationally. Brandeis was the lowest winning percentage team to be
ranked in the polls (#5 and #4) with a 65.2% win pct. Why was Brandeis ranked so highly in the
polls with a fairly low overall winning percentage? A few reasons can be drawn; in preseason
polls Brandeis was already ranked #4, so a spotlight was already placed on the team. A second
reason is the performance in the NCAA tournament where Brandeis made it to the semifinals.
The final polls shown in the data collection occur after the NCAA tournament and take this
performance into consideration. Only 4 total teams even qualified for the NCAA tournament
with a win pct. lower than 60% (St. Scholastica 59.5%, New England Col. 52.3%, Western New
England 52.3%, St. John Fisher 39.5%). Of all those teams not a single one made it past the first
round of the tournament. Many of the lower winning percentage teams had poor overall seasons
but did well in the conference tournaments in order to win and get an automatic bid. Depauw
with a 65% win pct. was the lowest winning percentage team to advance past the first round,
losing in second round play.

Conclusion: A teams winning percentage is very important to the overall success of the
team, although there are some small statistical inconsistencies that may take place. Teams need
to be in the top 25% of rankings in win percentage to do very well in the NCAA tournament.

2) Hypothesis: Goals per game (GPG) is less impactful on team success than goals against
average (GAA).

Analysis: Of the total 62 teams that participated in the NCAA tournament 41 were in the
top 25% of teams in GPG average. Also 2 of the 4 top ranked teams in terms of GPG did not
qualify for the tournament (New Rochelle 4.58GPG and Thomas(ME) 3.5GPG). Of the 62 teams
in the tournament 39 were in the top 25% of GAA. Although the GAA data doesnt support that
it is more important than GPG, of the final eight teams in the tournament 7 of 8 teams ranked 60th
or better out of 411 total national teams. All 7 of those programs had under a one goal against
average per game.

Conclusion: There is no conclusive data which shows that GPG is less impactful that
having a low GAA. The data does seem to point to the idea that to be truly successful teams
need to carry a low GAA overall.

3) Hypothesis: Teams must have a positive goal differential in order to make the NCAA
tournament or be ranked in the 25 national polls.

Analysis: All but two teams who were ranked in top25 polls or made the tournament had
a season long positive goal differential (St. John Fisher -8 and Western New England 0). One
exemption to this rule was New Rochelle with a goal differential of 68.

Conclusion: Based upon the data a positive goal differential is necessary in most cases to
be a successful team but the amount by which a team is positive is not a direct indicator of the
overall success of a program.

4) Hypothesis: Teams with higher average per game goal differentials will perform the best
over the course of an entire season. The data in this category will be the biggest indicator of
successful programs. This data selection takes into account that a team successfully scores goals
and defends well, keeping teams from scoring at a good rate. This suggests a balance team that
through an entire team will be successful.
 Analysis: Of the final eight teams in the NCAA tournament seven teams had a average
difference in goals of at least one per game. This means that on average these teams scored a full
goal more than their opponents. This put all of those seven programs in the top 25% of ranking
for teams on goal differential average.
Conclusion: In each year for a team to have a chance at success they need to have a
positive average goal differential which is about 1 or higher. The larger this number the
ultimately more successful a team will typically be. However it does not guarantee success as
the data does not take into account the adverse effect to the data of an extra large loss or victory.
As would be the case for New Rochelle 3.58 and Ohio Northern 2.8; both teams had high ranking
in average goal differential coming from performances in multiple games were they blew
opponents out by very large margins of victory.

Final Conclusions: Tufts was the national champion for the 2016 season. Although Tufts
never was near the very top of any statistical data points studied they were successful due to be
good across all areas. Tufts had an average goal differential of 1.03, a goal differential of 22, and
win percentage of 72.7%. Tufts was one of the successful programs because they were steady
throughout all statistically data points. The example of Tufts shows that teams may not need to
be the best in any single area but simply be good across all areas, this will allow teams to
compete with any programs successfully.
Penn State-Behrend, Thomas(ME), and New Rochelle are teams based on statistical data
should all have been ranked in top 25 polls and made the NCAA tournament, only Penn State-
Behrend made the tournament and lost in round one. The top 25 polls are determined based off
opinion. Although all three of these teams ranked extremely high in almost all statistical
categories they did not receive top 25 poll recognition. In my opinion the top 25 polls are based
most off of name recognition and past polls such as teams which were ranked in the preseason
poll. Top 25 polls should consider more the data points taking into consideration throughout this
analysis along with program prestigious and strength of schedule.

References:
D3soccer.com. (2017). Men's Top 25, Final . Retrieved from D3soccer.com:
http://d3soccer.com/top25/men/2016/final
NCAA. (2017, January). Men's Soccer DIII. Retrieved from NCAA: http://www.ncaa.com/sports/soccer-
men/d3
Ohio Athletic Conference. (2016, December). Men's Soccer. Retrieved from Ohio Athletic Conference:
http://oac.org/sports/msoc/index