Is Purdue Basketball Actually Cursed?

The Purdue Boilermakers, with coach Matt Painter at the helm, have turned into one of the most consistent programs in the country during the regular season. Purdue has finished inside the top 25 of Kenpom.com's season rankings in every season since 2016 and have participated in every NCAA tournament since 2015. In that same time frame, they have had an average seed of 3.9 for the tournament. However out of the last eight tournaments, they have been eliminated four times in the first round and had only one elite eight in 2019. This has led to many people theorizing about why Purdue keeps failing, specifically to lower level teams. So let's take a deeper look into March Madness, and Purdue basketball, to see if Purdue is really cursed (spoiler: they are not, but it feels that way sometimes).

First, let's take a look at the fickle tournament known as March Madness. To do this, I took college basketball team data from Bart Torvik's website for every season since 2013. I used all this data to analyze different factors and their correlation with deep runs in March (I classified a "deep run" as a team making it to the Elite Eight). I started the analysis by looking at teams adjusted offensive and defensive efficiencies to quantify the quality of offenses and defenses that have made it to the Elite Eight (E8).

Off. Quality	ADJOE	Number of Deep Runs	Average ADJDE
Bad	Below 100	0	0
Average	100 - 110	4	92.4
Good	110 - 120	37	91.8
Great	Above 120	23	93.1

Figure 1. Table defining a new category, offensive quality by ADJOE. Then showing the number of E8 runs and the average ADJDE for all the teams that made an E8 run with the corresponding Off. Quality.

Figure 2. Using Figure 1 data, this shows a comparison of total teams with the corresponding Off. Quality, compared to the number of E8 teams.

This bar graph above shows the how the adjusted offensive efficiency rating (ADJOE) of teams are distributed according to the offensive quality, as described in Figure 1. The black bars represent all the teams in the NCAA, and the red bars show the distribution of ADJOE for teams that made it to the elite eight. The average ADJOE for E8 teams was 117.9. There were four outlier teams: Louisville in 2012 as a 4 seed, South Carolina in 2017 as a 7 seed, Kansas State in 2018 as a 9 seed, and Oregon State in 2021 as a 12 seed.

Quality	ADJDE	Number of Deep Runs	Average ADJOE
Bad	Above 100	0	0
Average	100 - 95	1	118.3
Above Average	95 - 90	14	117.5
Good	90 - 85	31	118.6
Great	Below 85	18	116.9

Figure 3. Table defining a new category, defensive quality by ADJDE. Then showing the number of E8 runs and the average ADJOE for all the teams that made an E8 run with the corresponding Def. Quality.

Figure 4. Bar graph comparing the number of E8 teams with the total number of teams for each defensive quality level.

This bar graph is formatted to be the same as Figure 2, but instead of ADJOE the graph depicts the adjusted defensive efficiencies (ADJDE). The average ADJDE was 92.3 for elite eight teams. As expected, the better defense ratings show a higher percentage of those teams making the elite eight.

From these graphs, the characteristics of a successful March team start to show themselves. In general, teams need to reach the benchmarks of an ADJOE of 115 and above, and an ADJDE that is below 95. With these team quality standards, how does Purdue stack up to these marks? In the last 8 seasons, Purdue's ADJOE has been at an average of 118 and ADJDE of 94.7. But, as a Purdue fan, I can personally attest that this benchmark does not guarantee success in March. Thus, the next step is to see if certain play styles perform better in March.

We begin with pace of play, as that can often give insight to how each team plays. The average adjusted tempo for E8 teams is 67.2 possessions per a game, with the max tempo being 74.6 (Gonzaga in 2021) and the slowest playing at the pace of 59.3 (Wisconsin in 2015). The ideal tempo seems to be between 65 and 69.

Pace	Number of E8 Teams	ADJOE	ADJDE	Average Seed
Slow (Below 65 possessions)	15	119.2	92.2	3.2
Mid (65-69 possessions)	30	115.7	92.5	4.5
Fast (Above 69 possessions)	19	119.5	92.0	2.4

Figure 5. Comparing the quality of E8 teams based on pace.

Figure 6. Bar chart showing the number of E8 teams based on tempo.

Now when viewing the table above (Figure 6), it may look like it is actually better to run either a slow or fast paced offense. On the contrary, the table shows that teams in the slow and fast ranges have a much lower margin of error. With a mid level pace, "worse" quality teams still have a very good chance of making an E8 run and have made more of these deep runs compared to the other two categories. Although, to tidy up and standardize the information, the next table removes lower quality teams (teams below a BARTHAG rating of 0.8) and allows us to more accurately compare the different paces of good teams.

Pace	Good Teams	Good E8 Teams %	Good Teams Upset %	Great Teams	Great Teams E8 %	Great Teams Upset %
Slow	46	0%	43%	43	33%	44%
Mid	149	4%	43%	80	25%	34%
Fast	92	1%	53%	54	27%	33%

Figure 7. Based on the pacing categories from figure 5, this table compares deep runs versus upset percentages for each category. Good teams are classified as between 0.8 BARTHAG and 0.9, while great teams are above 0.9 BARTHAG.

In conclusion, by interpreting both Figure 6 and Figure 7 together, the data shows that teams playing at a mid level pace (65-69 possessions a game) can be less efficient in both ADJOE and ADJDE while still having comparable outcomes compared to slower and faster teams.

Next, let's take a look at key indicators for teams to better understand how these teams fair in the possession battle. In other words, how many field goals attempts and free throw attempts each team is getting compared to their opponent. This narrows down to two main factors: turnover margin and rebound margin. An example of both is as follows, team A has 12 turnovers and 5 offensive rebounds compared to their opponent team B who has 16 turnovers and 15 offensive rebounds, and each side had 100 possessions. In this scenario, team A would have a turnover margin of +4 turnovers, a TO margin percentage of +4%, an offensive rebounding margin of -10, and an OR margin of -10%. Overall, team B would come out ahead in the possession battle by 6 possessions and a margin of +6%. Comparing this to our data, we can see some interesting benchmarks of how large the possession margin should be for teams that want to make a deep run in March Madness.

Figure 8. Shows both TOMargin and RBMargin for each E8 team based on tempo.

Each purple dot represents a team's rebounding margin, while the tan dots represent a team's turnover margin. I added Purdue's 2023 team into this graph as bright greet dots to see how this team compared to the data. On the Y axis of the graph is the margin percentage compared to the adjusted tempo on the X axis. So, to simplify this data, the scatterplot below combines the purple and tan dots to gain understanding of how the possession battle plays out for teams in March.

Figure 9. Combined TOMargin + RBMargin compared to tempo.

This figure shows that the teams who are able to win the possession battle by 5 possessions out of every 100 will have a better chance of success in March. While also showing a slight correlation with higher tempo teams needing to win the possession battle by more compared to slower teams.

Figure 10. Scatterplot of teams 2pt% in relation to tempo

Figure 11. A teams 3pt% compared by tempo

In the above graphs, the red and orange dots represent the shooting numbers for E8 teams, and the dominate color shows all other teams. The trendline in each graph represents the the trend of the data with tempo. Note: increased tempo has a positive correlation with 2 point percentage and no correlation with 3 point percentage. Also, a large percentage of E8 teams are above the trendline no matter the tempo.

In an article by Rhonda Magel and Samuel Unruh, they found that the four statistically significant predictors to be turnovers, assists, field goal percentage, and rebounding (Magel and Unruh, 2013). From the data that I have pulled together, that sentiment holds true. To take it one step further, I would argue that pace has a significant role in a teams ability to deal with a tournament like March Madness, because of how many unique styles of play are in college basketball.

Now, to return to the burning question that sparked all of this analysis... Is Purdue basketball really cursed? Yes, ever since Neil Armstrong set foot on the moon I have been condemned to years of torment. In all seriousness, Purdue basketball has reached many of the benchmarks for team quality, but they play consistently at an average pace of 66.3 and struggle to beat teams in the turnover battle. It is no coincidence that Purdue's best March Madness runs (2018 & 2019) both came when the team had a TO% below 16 compared to 17.8 TO% in the other 6 years.

Magel, Rhonda, and Samuel Unruh. “Determining Factors Influencing the Outcome of College Basketball Games.” SCIRP Open Access, Scientific Research Publishing, 16 Aug. 2013, www.scirp.org/html/2-1240217_35927.htm.