IT’S HEEEEEEEEERE!!!
It’s March and the tournaments have started playing! We held our team meeting at Buffalo Wild Wings eating saucy chicken and watching 35 big screen TVs with various games going on. This is the greatest time of the year!
While chatting, we saw an interesting tweet put out by a fellow CBB aficionado, @EvanMiya about Oral Roberts and thought that his claim deserved a little deeper look.
In the tweet, he said, “Oral Roberts has all of the ingredients to be THE cinderella of March Madness:
- Leads the nation in 10-0 scoring runs per game. (31 total this year)
- Shoots threes (top 30 in 3P% and 3PA rate)
- Top 40 offense in the country
- Max Abmas is a veteran (22 PPG this year)”
Those are some solid points and ORU has a legitimate argument that they could be big this year. Even Ken Pom (@KenPomeroy) kicked in that he believes this ORU team deserves an at large bid but won’t get one because of big conference bias by the Selection Committee. They are talented, they have a solid resume and they probably deserve a chance to prove that.
But the original question was, do they have the ingredients to be a cinderella?
Like all financial commercials, let’s start this out by saying that past performance does not indicate future success. But there is a really strong correlation. So lets dig into this massive mine of data and see if we can prove that those particular cherry-picked criteria for Cinderella bear out.
We started out by eliminating all Major and Mid Major conferences, so that the Cinderella’s could be starting from and competing on the same plane as ORU. Then we adjusted for top offenses and filtered for the 3P% and 3PA rate.
I do really like the assertion of 10-0 runs in a game being significantly important. It reminds of the old Arkansas “40 Minutes of Hell” days where Coach Nolan Richardson believed that basketball is a contest of runs and who ever can have the most for the longest wins. It would be interesting to see how those long runs matter if there are fewer total runs in a game, or if they were able to sustain them against superior competition, or if those runs resulted in an above average rate of wins in those games relative to their other games played, or if those runs were indicative of an equally strong defense.
There are a thousand more questions you could investigate inside that data, and we did for a while, but in the end, we still have not found any evidence was that the 10-0 runs were a strong indicator of Cinderella type teams.
If we go back to the data on the first couple of criteria, we come up with 5 past tournament teams that met those same criteria and got into the tournament.
- 2009 NDSU (14 seed)
- 2012 Iona (14 seed)
- 2018 SDSU (12 seed)
- 2021 Colgate (14 seed)
- 2022 SDSU (13 seed)
So, how’d they do? Were any of them Cinderella’s?
Well, unfortunately, no. Not a single team with these same characteristics made it out of the first round. Iona even lost in the play-in round.
That’s not to say that Oral Roberts couldn’t do it this year. They first need to earn their auto bid and then I’ll tell you what, if they advance past the first round, I’ll go out and buy an ORU hat or jersey for everyone on the team, just to keep us humble.
But I don’t think I’m going to have to do that.
Hey before I go, I’m curious. If you were to want a few questions answered that you may not know the answers to, something that we could dig through our data mine and give you an insight that you may not have had before, would you please pass those along in the comments section? I’m not trying to trick some stupid algorithm, we don’t have one. I’m just curious what questions you have about your team or ways that you think we could understand this tournament prediction game better. Drop it below. Or Don’t. We enjoy reading all the spam emails from SEO “companies” in the inbox.
– Bracket Ninja
13 responses to “Is Oral Roberts a Cinderella?”
Hey guys, do you still track the Q1 wins rule that champions usually have a least 9 Q1 RPI wins? I know we don’t use RPI anymore but it’s an interesting stat. It seems recently only 3-4 teams seem to qualify for it- this year I believe only Kansas and Alabama have it so far. We obviously need more NET data, but what thresholds have we seen thus far in the NET era for the champions?
Hey, Chad. Thank you for the question! Great memory. Back when I identified the 9 Q1 win threshold, I hadn’t realized that included NCAA Tournament games, which it shouldn’t have. I’ve updated my historical numbers, and the threshold is only 4 Q1 wins, but 17 of 19 champs had 6+. A more discriminating metric, now, is a 0.500+ Q1 record.
The great thing about RPI is that it’s a straightforward, transparent formula that is consistent across all years. Not only is NET newer, but it’s based on a machine learning model that the NCAA has tweaked at least once (probably more) already. So I don’t plan on using NET for any of my analysis. Keep an eye out for some more findings I’ll post on Twitter (@BracketNinja).
Is there any correlation to ATS trends going into the tournament and performance? I can’t find a stat like last 10 or 15, but it does seem like some teams with conference game ATS records above 70% outperform their seed. Some underperform however and I wonder if there’s a difference between the complete conference schedule and the trend going into the conference tournament?
Overperformance Examples:
– 2022: St. Peter’s – 78.3%
– 2021: Houston – 70%, Oregon St. – 73.9%, Abilene Christian – 70.6%
– 2019: Michigan St. – 75%
Thank you for the comment, Ross. I haven’t looked at ATS trends, but I’m intrigued and loving the suggestion! Where did you find the historical ATS trends? I’m not sure if I have enough time to check it out before this year’s tournament, but I’d like to try.
I used TeamRankings.com to find them but the way you could slice the data was limited. My theory was by the end of the season Vegas should have enough information to make the most accurate lines of the season. If that’s true, then the teams that are outperforming the lines the most and winning enough to make the tournament should be good teams to pick to outperform their seed.
If there is correlation, watch out for Utah Valley, Furman, Vermont and Texas A&M to at least play better than their seeds in March
Ross, I think you are on to something here. We built a previous version of the model where we had high confidence on certain games and then compared just those games to the spread. We did very well for a while. Perhaps it’s time to dust that idea off a bit and potentially integrate Vegas’ late season input into the model to try to refine it.
UVU didn’t get a seat at the NCAA tourney, but they did just beat UNM nd UC Boulder at away games. Playing host to Cinicinatti in the NIT on Wednesday. Looke like you were right on this one.
Awesome! Thanks! I’ll definitely look into this. I’m always searching for new and different signals/predictors.
There’s something there, but it may have more to do with the matchup. Teams >50% ATS v Teams <= 50% are 10-8 so far. That also called Farleigh Dickinson, Princeton (x2) and Furman which is kind of crazy…
What about Oral Roberts team in 2021, 15 seed that made Sweet 16, a true Cinderella?
Below are their stats, very similar to this year’s team:
(top 30 in 3P% and 3PA rate)
Top 74 pre-tourney offense in the country
Max Abbas averaged over 20ppg
Dan, thank you for the comment–that is a really good point. I can see the heavy similarities and I appreciate the high 3P% and 3PA levels. We didn’t initially consider the 2021 ORU team for two reasons. First, they didn’t quite fit the criteria of the tweet I referenced. I suppose that we could have loosened up the criteria a bit but I was just going off the tweet so because the 2021 team was across the board slightly lower it failed to make our initial list.
The second reason it wasn’t considered is really not a reflection on ORU so much as it was on the world as a whole. As you know we lost the entire 2020 NCAA Tournament to COVID cancellation (I’m still shaking my fist to the heavens about that). And 2021 was, from a data science perspective… problematic. The more data points we have the more confidence we can have that the results will be predictive. We had a few teams that were unfortunately pulled from conference and NCAA tournament play because of exposure to COVID, and many others that didn’t play a full season which threw off the model quite a bit. As an example, we didn’t have enough on UCLA to even fathom that they might win an overtime play-in game against Michigan St and then go on, as an 11 seed, to win their next 4 games and get to the Final 4. That makes no sense at all.
2021 ORU was a really solid team who earned their NCAA berth by winning their conference. According to our model they were properly seeded. However, Ohio State was significantly over rated. In our model they were a 6 seed and Florida was ranked 64th overall. Again, not to take away from ORU because they did a great job and did over perform but the matchups were more favorable for them that the selection committee deemed final.
I will say that our model this year really does like ORU. They are improved, and that Sweet 16 run gave them a veteran experience that is hard to quantify. Most models have them as a legit 12 seed right now. It is well earned in our opinion.
Thank you so much for the well thought out reply.