Arc of Champions - 2023

[fasbjd]

Back again with this years version of the Arc of Champions.  Brief review, the Arcs are based on analytic data from Bart Torvik's T-rank.  This is a little different from last year, which used Ken Pom, and is basically due to a number of factors both functional and practical.   Most importantly, it led to a graph with stronger correlation to tourney results.  It also has the advantage of ease of separating data on the website while being free to use.  It includes data since the '08 season.  The main difference in the two sites is how they adjust for SOS.

 

Now to the Arcs - the graph is broken down into 6 sectors.  The data below is from the '08-'22 seasons

The 1st Sector is the Arc of Champions.  Just over 6% of teams fall within this sector (average ~4/year).  Of those, 71% of champions since '08 have come from this sector - 47% of finalists, and 38% of Final Four teams.  Their average seeding is 1.7 and they average 3.2 tourney wins / team.  If your team is located here, their chances of winning the title is 17.2%.  If you are a 1 seed in sector 1, your chances of winning the title are 24.3% (versus 19.6% of all 1 seeds).

Sector 2 is a larger sector - over 16% of teams come from this sector.  The average seed is 4.1, including 17 (now 19 including this year) 1 seeds, and houses a majority of the 2 and 3 seeds.  On average, they win 1.9 tourney games / team - on average, finishing in the Sweet Sixteen.  21% of champions come from this sector, along with 36% of the finalists and Final Four teams (including Duke and Villanova from last year).  If your teams is located in this sector, their chances of winning the title is 1.9% (just slightly higher than that of a 2 or 3 seed) and making the Final Four is 7%.

Sector 3 arc is labelled the Final Four Arc because outside this arc, your chances of a Final Four finish is only 2.8%.  Sector 3 houses 21.8% of teams with an average seed of 6.4.  The average just under 1 tourney win/team and teams in this arc have a 0.5% chance of winning the title and (despite the name) only ~4% chance of a Final Four.  Indiana was located in Sector 3 last year as were Elite Eight team Arkansas and Sweet Sixteen Texas Tech.

Sector 4 is where we find the current Indiana team - earlier in the year, we were strongly in the 3, nearly 2, Sector but the defense fell off - in large part correlating with the loss of Xavier Johnson.  Historically, 22.5% of teams are from this sector, with an average seeding of 9.3.  They average 0.7 tourney wins/team.  A champion has never came from this Sector but 2 finalists have, including last year's UNC.  Other notable Sector 4 teams from last year include Sweet Sixteen teams Providence (who were also a 4 seed), Michigan, and Iowa State.

The final two sectors house the rest of the field - 23.1% of the teams and have never produced a Final Four team.  Sector 5 teams have a 3% chance of Elite Eight and ~7% chance of Sweet Sixteen.  Sector 6 has produced only 1 Elite Eight team - last year's darling St. Peters.

 

 

This years Arc is displayed with some color coding - teams in white 1-4 seeds, yellow 5-8 seeds, orange 9-12 seeds, and the rest in pink.  Indiana and our first opponent Kent State are highlighted.

This year's arc shows 2 clear front runners for the title in Houston (though another Sasser injury may play a role here) and Alabama.  As opposed to year's past, there seems to be quite a falling off after those two - 4 seed UConn along with Purdue seem to be the next closest (though both would have to win it with defenses worse than only than the '18 champions Villanova who had a far superior offense) along with 2 seed UCLA.  Although there has been a strong call for Kansas to be the top 1 seed, per these analytics, they appear to be the least likely 1 seed to win it this year.

 

As far as Indiana, clearly the defense has to be better to make a splash in the tournament.  In addition, the offense is better than only the '13 championship Louisville team that had a crazy good defense.   In the short term though, in our block of teams for the road to the Sweet Sixteen, our defense in very comparable to both Kent State and Drake.  Offensively, we are quite a bit better than both (particularly Kent State).  Miami, however, is a different animal.  Defensively, they fall between Penn State and Iowa - amongst the worse in the field.  Offensively, they also fall between Penn State and Iowa - amongst the fields most efficient.  Not teams we particularly want an opponent compared too!  Luckily, they only rank 250th in point distribution from 3 pointers - with a similar 3p shooting % to ours while taking about a third of their shots from long range (ranking 268th where Iowa ranks 182 and Penn State 10).

[5fouls]

Thank you @fasbjd.  Your stuff is among my favorite things on this board.

I'm a little disappointed with where IU falls in the chart.  Based on where other teams fell, I would have guessed us to be one arc closer to the championship arc than what we are.

It's a good reminder for IU fans not to get our hopes up.

[rico]

1 hour ago, 5fouls said:

Thank you @fasbjd.  Your stuff is among my favorite things on this board.

I'm a little disappointed with where IU falls in the chart.  Based on where other teams fell, I would have guessed us to be one arc closer to the championship arc than what we are.

It's a good reminder for IU fans not to get our hopes up.

Torvik just didn't like us this year.

[5fouls]

I took a look at some of our individual efficiency ratings, and I was very surprised how inefficient JHS is on some of the metrics.   And, not surprisingly, our defensive efficiency in teh backcourt was bad all around.  

[5fouls]

This is a real shocker.  Despite playing the 2nd most minutes, JHS has the worst Box +- of any of the rotation guys. and a few of the non-rotation ones.

[fasbjd]

37 minutes ago, rico said:

Torvik just didn't like us this year.

I think the computer numbers overall are down - some close wins and less close losses that dilute out the big wins that led to our seeding.

Torvik's numbers are more critical of the offense where he has us ranked 34th where KenPom has at 27th.  The defensive rankings are comparable (KemPom 43 to Torvik 46) and overall Torvik has us at 33 and KenPom at 30, so not far off.

[NotIThatLives]

Remember when our defensive efficiency rating hit 8th?  That was awesome. 

[13th&Jackson]

11 minutes ago, NotIThatLives said:

Remember when our defensive efficiency rating hit 8th?  That was awesome. 

[kyhoosier29]

2 hours ago, 5fouls said:

This is a real shocker.  Despite playing the 2nd most minutes, JHS has the worst Box +- of any of the rotation guys. and a few of the non-rotation ones.

I’m surprised you’re surprised 🤣. He was wildly inconsistent on offense (shooting and taking care of the ball) this year and was really bad on D many times. Isn’t very good at keeping his man in front of him or tracking his man when he doesn’t have the ball. With that said, he won us a couple games and he has all of the potential in the world, but I think he could really benefit from another year. 

[fasbjd]

Here's the updated version for the round of 32 (click for better resolution):

 

Only one sector 1 or 2 team gone in first round...and I'm alright with that!

Edited March 18 by fasbjd

[Drroogh]

Well Miami is farther out than IU

[IUProfessor]

This is real interesting. Have you ever played around with it to see if limiting it to the last 4 or 6 weeks improves the accuracy? The problem with season-long analytics is that teams like Purdue and UConn haven't been playing at the same level at year's end that they were for much of the season, but the models don't fully take that into account. So this probably overrated their chances.

[Sea Turtle]

29 minutes ago, Drroogh said:

Well Miami is farther out than IU

We will have our hands full. Hopefully we can pull it out.

[fasbjd]

25 minutes ago, IUProfessor said:

This is real interesting. Have you ever played around with it to see if limiting it to the last 4 or 6 weeks improves the accuracy? The problem with season-long analytics is that teams like Purdue and UConn haven't been playing at the same level at year's end that they were for much of the season, but the models don't fully take that into account. So this probably overrated their chances.

Good thought..unfortunately with 1000s of data points, won't be an easy task!

[IUProfessor]

18 minutes ago, fasbjd said:

Good thought..unfortunately with 1000s of data points, won't be an easy task!

Didn't mean to suggest you necessarily should do it, just curious if you'd looked at that at all. My guess is that some weighting of most recent play while still factoring in full season numbers would probably provide the greatest accuracy, but I'm not a stats guy. Torvik does make it easy to do cutoffs of, say, February 1st to end of year, if that helps.

[TheWatShot]

LMAO, Fairleigh Dickinson is such an outlier, it needs an arrow pointing toward it. And Purdue lost to that team. 

[5fouls]

On 3/18/2023 at 3:29 PM, TheWatShot said:

LMAO, Fairleigh Dickinson is such an outlier, it needs an arrow pointing toward it. And Purdue lost to that team. 

The arrow is actually pointing off the chart all-together.  Boiler-Up! 

[5fouls]

Gonzaga, Xavier, Michigan State, and Florida Atlantic are Sweet 16 teams that came from IU's Arc.  Miami was one Arc behind us.  Princeton 2 arcs.   The other 10 Sweet 16 teams were positioned better than us on the arc.  

[fasbjd]

14 hours ago, 5fouls said:

The arrow is actually pointing off the chart all-together.  Boiler-Up! 

Yeah...I looked back and they were the worst defense by adjusted efficiency that has made the tournament since '08 per T-rank...should say at least '08 since that's the first year of data on his site).

Edited March 20 by fasbjd

[fasbjd]

Well...here's how the grid stands going into the sweet 16 (again click on the image for better resolution). Field tightening up quite a bit with 2 outliers...Princeton and, unfortunately, Miami.  As I said in the original post, Miami worried me the most of the 3 teams in our section of the bracket even though their position on the grid would suggest it's a team we should beat.  I didn't like their positioning among Iowa and Penn State (teams we showed vulnerability to).  

Despite all the chaos of the first 2 rounds, 9 of the 16 top 4 seeds remain.  Taken seed-by-seed, the survivors make sense:

1: Houston and Alabama looked to be superior to Purdue and Kansas.

2: UCLA and Texas looked better than Arizona and Marquette

3: Kansas St and to a lesser extent Xavier (whose highest seed faced was a First Four 11 seed) and Gonzaga looked superior to Baylor

4: Tennessee and UConn look to be amongst the elite where Virginia and, unfortunately, IU were not

 

[fasbjd]

 

The Final 8 - both sector 1 teams eliminated (won't shed a tear for either of those coaches!).  Mutiple strong contenders still standing.  Miami / Gonzaga / FAU remain the outliers per the grid - but all playing well, so a new standard may be set.

[fasbjd]

 

The Final Four Arc Grid:  Two contenders that are playing by the rules and two that are trying their best to break the grid and ruin my life's work 😜.

Miami is the first team from Sector 5 to make the Final Four.  FAU is the 5th from Sector 4.  Sector 4 had 2 teams make the Finals including UNC last year.  None have won it all.  

On the other hand, Sector 2 has had now 13 teams in the Final Four, about 8% of the teams.  7 of the prior 11 made the Finals and 3 won it all.  The safe bets go to UConn and SDSU...but nothing has been safe this tourney so lets see how the grid holds up!

Edited March 30 by fasbjd

[fasbjd]

The grid did a little flexing last night.  Feel bad for Dusty but the grid gives no quarter!

As for the finalists, an interesting juxtaposition.  Both are 2nd sector teams but similarities end there.  UConn, at first glance, looks like the better positioned team (they seem to have more company of former champions in their general area)...but if they win, they will do so with the second worst defensive efficiency for a champion in grid history (since '08) with the only team worse than them being the team with the best offensive efficiency in grid history (Villanova '18).  In fact, they would only rank as the 10th best in aOE of the 15 represented champs, just behind the current average for champion's aOE (120.2 vs 119.8).

SDSU, on the other hand, would have the second worst offensive efficiency of any champion in grid history with the only team worse being one of the 2-3 best defensive teams in grid history (Louisville '13).  SDSU would have the 7th best defense to win the title with an aDE right at the average of current champions (90.6).

As such, neither team would be a surprise.  My heart will be pulling for SDSU just to keep UConn behind us in the banner count!

 

Edited April 2 by fasbjd

[Purdue7]

On 3/18/2023 at 3:29 PM, TheWatShot said:

LMAO, Fairleigh Dickinson is such an outlier, it needs an arrow pointing toward it. And Purdue lost to that team. 

Just shows you at the end of the day you gotta be able to make shots 

or stats don’t account for choking