I don't think they are as much of a lock as everyone thinks. They literally have only one good win on their resume, albeit against us. Their other two chances to prove themselves against ranked teams ended up in losses. Now add to that a loss against a VERY bubble team in St Mary's..if the committee looked at it objectively they should be a 2 seed. Or maybe the LAST one seed at best.
If you look back to 2015 season, the zags were our #2 seed even with a 35-2 record! Their losses came to a no.3 zona squad and byu in their last regular season game.
Granted, this years squad comes with alot more hype, but the overall resume does not look good for them. Looking at it objectively, they really should be a 2 seed now.
The Gonzaga loss to Saint Mary's is staggering. They beat the Gaels by 14 on the road about 10 days ago. It's a staggering fall. That puts Gonzaga at 4-3 in Quadrant 1 games. Duke has a seriously stronger resume at this moment with wins over Virginia x 2, Kentucky, and Texas Tech. I really hate the idea of Duke being a 1 Seed out west with Gonzaga as the 2 Seed.
The situation is still the same for Duke: Win on Friday night and guarantee a 1 Seed.
What I think this does is open the door for Duke to finish the regular season as the AP #1 team and overall #1 team in the tournament. In fact, I think the winner of the ACC Tournament is going to get that designation unless a crazy upset happens.
Given how Gonzaga seems to actually not be good against top teams (aside from the early season duke win, the gaels are their best wins...), i would be completely fine being matched up with them. I also find the "duke doesn't play well out west" mantra over-used. We have poor showings in lots of locations...and there is the hidden variable of...duke teams that are sent out west are on average, not as good as duke teams that get to stay east.
Enough east coast teams have succeeded out of the west region for me to have no reason that with a good team, we could too.
April 1
I think the #2 team in both NET and Kenpom is probably safely a #1 seed. Most likely the worst they can finish after the rest of the tournaments finish up is #3 in both, where they still most likely get a 1 seed (and falling to #3 isn't a given, they might well still be #2 on Sunday).
Well, if Duke heads out West, something bad happened in the ACC tourney most likely, and I don't see why we would feel confident then.
Also, Gonzaga's sample against good teams is as small as the Duke's sample out West, so I'm not sure I would ignore one in favor of the other. You're a big kenpom advocate and I don't see why you would ditch him after one loss. For me, I wouldn't like being matched up with the kenpom #2 team out West in their home region.
I don't think the committee is going to use NET (or kenpom) that bluntly, but I do think they WILL use the quadrants pretty bluntly, and Gonzaga is only 4-3 in Q1 now.
The Zags are probably in trouble if both Kentucky and MSU win their conference tournaments. Then the 1-seeds will be those two, UVA, and one of Duke/UNC, imo.
I think the Zags get a 1-seed because most likely, UK and MSU won't both win their conference tourneys. That's different from calling Gonzaga safe as a 1-seed, though.
Why? It would perhaps undermine the credibility of those interpreting the data, not the data.
Interesting K-pom has St Marys at #29 ahead of: Marquette #30, NC State #32 and Syracuse #39.
T-Rank has St. Marys at #33 ahead of: Villanova #34, Marquette #36 and NC State at #37.
They seem to be better than I thought going into the Zags game.
GoDuke!
If the committee ends up using quadrants bluntly, Gonzaga could be in a bit of trouble. As you point out, Gonzaga "only" 4-3 vs Q1 (and 6-0 vs Q2).
For context, most people scoff at the idea of giving Houston a 1 seed. However, if Houston ends up winning the AAC tournament, they are likely to finish 7-2 vs Q1 and 10-0 vs Q2. Sure, Gonzaga's win vs Duke counts for something, but should 4-3 vs Q1 and 6-0 vs Q2 really be considered "safe 1 seed" when 7-2 vs Q1 and 10-0 vs Q2 isn't generally in the conversation for a 1 seed?*
Heck, almost nobody is talking about Texas Tech as a potential 1 seed. However, if Texas Tech wins the B12 tourney, they are likely to finish 9-5 vs Q1 and 10-0 vs Q2. Is 9-5 better than 4-3? I don't know, but we may find out what the committee thinks.
*Just to be clear, I don't think Houston deserves a 1 seed. Houston's Q1 record is very misleading because they haven't played any top 25 NET teams (Gonzaga has played 3 top 7 NET teams, all away from home). I am mostly just pointing out the hazards of using quadrants too bluntly, just in case anyone from the committee is reading DBR
While KP doesn't list confidence intervals on individual teams, such an interval would shrink the more data we have between a given team and similarly ranked teams. KP readily accepts this, which is why the weighting of a game is diminished the further teams are apart in efficiency. In the case of most teams, there are enough games against similar teams that those other games can be properly "demoted" (?) in importance. In the case of gonzaga, we have only 3 data points, all of which came from months ago (KP also deweights games with time).
So what we're left with is a gonzaga team for which a HUGE amount of their rating is comprised of data that is already accepted to be significantly less reliable....which would make any error bars on their ranking enormous.
April 1
One of the challenges of comparing potential 1 seeds is that the level of competition they face varies greatly. We all know that Duke played a more difficult schedule than Gonzaga and Houston, but how can you quantify this? Is 30-3 vs Gonzaga's schedule more of an accomplishment than 25-6 vs Duke's? Within the analytics community there is growing interest in using "Wins Above Bubble" as a metric to evaluate a team's win/loss resume with respect to its schedule. In short, Wins Above Bubble is the difference in the actual wins for a team vs the number of wins an average bubble team would be expected to have when facing the team's schedule.
For example, based on KemPom ratings, the average bubble team (AdjEM = 15) would be expected to win 20.1 games against NC State's schedule to date. NC State has won 22 games, therefore, NC State is currently sitting at 1.9 Wins Above Bubble. On the other hand, the average bubble team would be expected to win 16.6 games against Indiana's schedule. IU has won 17 games, so IU currently has 0.4 Wins Above Bubble. So NC State is currently ahead of IU in terms of Wins Above Bubble.
What does this have to do with 1 seeds? Well, you can rank potential 1 seeds by Wins Above Bubble. A full list of all 353 NCAA teams by wins above bubble is available here.
However, the average strength of a bubble team probably isn't the best perspective to judge potential one seeds. I won't go into detail, but using Wins Above Bubble to rank potential 1 seeds may give too much credit for winning games vs teams against teams ranked 100-200. Winning a Q3 game is a modest accomplishment for a bubble team, but should be considered a "gimme" for a 1 seed.
One alternative is to look at the schedule of potential 1 seeds from the perspective of the "average" top 5 team instead of the average bubble team. Applying this metric to the 12 teams currently listed as 1-3 seeds on BracketMartix, you get the following in terms of Wins Above Top 5 (or WA5 for short).
Some interesting notesCode:
Team Expected Wins Expected Losses Actual Wins Actual Losses Wins Above Top 5 WA5 Rank Virginia 24.93 5.07 28 2 3.07 1 Duke 25.05 5.95 26 5 0.95 2 North Carolina 25.33 5.67 26 5 0.67 3 Tennessee 26.76 4.24 27 4 0.24 4 Gonzaga 29.93 3.07 30 3 0.07 5 Houston 28.98 2.02 29 2 0.02 6 Michigan 26.10 4.90 26 5 -0.10 7 Kentucky 26.16 4.84 26 5 -0.16 8 Michigan St 25.71 5.29 25 6 -0.71 9 Texas Tech 26.85 4.15 26 5 -0.85 10 LSU 27.00 4.00 26 5 -1.00 11 Purdue 25.44 5.56 23 8 -2.44 12
- UVA is the clear leader in "Wins Above Top 5". Even if Duke wins the ACC Tournament, UVA is likely to remain the leader in Wins Above Top 5. Of course, in that scenario, I would still expect Duke to get the overall #1 seed because a) they would be either 2-0 or 3-0 vs UVA, and b) I don't think the committee is looking at Wins Above Top 5.
- Gonzaga took a real hit in terms Wins Above Top 5 after their loss to St. Mary's. A 31-2 Gonzaga team would be second in Wins Above Top 5. However, at 30-3 Gonzaga is currently 5th, barely ahead of Houston and will likely fall behind Mich or Mich St if either wins the Big 10 tourney.
- Wins Above Top 5 doesn't attempt to predict future games. Rather, it attempts to reward teams for the strength of a their win/loss resume based on past games. There is a very interesting philosophical discussion to be had about whether the tournament should be seeded based on descriptive "resume metrics" which focus on wins and losses and schedule strength, or "predictive metrics" (like KenPom or T-Rank) which mostly care about how strong a team is without really caring how many games a team has actually won. For example, if KenPom calculated ratings for 1995, Duke might well have been ranked in the top 30-40, but was Duke's 13-18 win/loss resume deserving of a tourney bid.
What's the alternative though? Being the #2 seed out West to face Gonzaga?
We won't be the #2 to UVA or UNC.
It's lose-lose. We are likely getting an away game in that Elite Eight game.
So from that perspective I'd rather get the 1 seed and get a little bit more probability to win.
On February 9th Gonzaga destroyed St. Mary's 94-46, more than doubling the Gaels' point total. It truly is remarkable that St. Mary's gave up 90 points to Gonzaga in February and just over a month later they they held the bulldogs to 47 points. Gonzaga's margin of victory in the first meeting exceeded their point total in the 3rd meeting.
-48... -14... +13... if they meet a 4th time maybe st. Mary's will win by 33 ;-).
You make a good point about 1 seeds generally getting an easier path than 2 seeds. However, I am not sure that this would necessarily be the case for Duke in 2019. The most significant benefit of being a 1 vs 2 seed is probably the fact that a 2 seed has a chance of being placed in the same region as dominate 1 seed. For example, a 1 seed in 2013 would have given Duke a better shot at the Final 4 by avoiding Louisville's region. If you believe KenPom and T-Rank, UVA is probably the only "dominant 1 seed" this year, and Duke can't be the 2 seed in UVA's bracket.
Duke is probably going to end up in the same region as a Gonzaga, Michigan, Michigan St, UK, or Tennessee. These teams are all in KenPom’s top 8, so it probably doesn't matter a whole lot if Duke is the 1 seed in the region and the other team is the 2 seed, or vice versa. If Duke gets matched with one of these teams, the only meaningful difference in the path to the Final 4 is probably in the Sweet 16, where the 2 seed likely faces a 3 or 6 seed and the 1 seed likely faces a 4 or 5 seed.
Historically, there hasn't been much difference between facing a 3/6 seed vs a 4/5 seed. If you look at the average KenPom ratings of past 3 - 6 seeds, the point spread you would expect vs a healthy Duke team (AdjEM ~34).
Seed Expected Point Spread vs Duke Probability of Duke Victory 3 ~8.5 pts 78% 4 ~9 pts 79% 5 ~10 pts 82% 6 ~11 pts 85%
That being said, no team is going to face an "average bracket" in 2019. The real difference in Duke's path as a 1 vs 2 seed will mostly depend on who the committee actually selects as the 3-5 seeds in Duke's bracket. In a bracket with a strong 3 seed (like Michigan), it is almost certainly better to be a 1 seed. However, in a region with a relatively relatively weak 3 seed (like LSU or Houston) and/or relatively strong 4/5 seed (like Texas Tech, Virginia Tech, Purdue or Wisconsin), it probably makes very little difference if the 1 and 2 seeds are swapped.
As a Duke fan, the biggest reason to pull for Duke getting a 1 seed is that it probably means that Duke beat UNC tonight!
Just FYI as we talk about seeding: Ticketmaster is having a "no fee" sale today only for unsold NCAA tickets (which means 90% off ). The sites within driving distance of Raleigh that could be relevant are Columbia, Jacksonville, and Louisville. We really did a number on KY if they didn't sell out Louisville when those tickets went on sale last November.
One of the undercurrents to tonight's game is where we play the first 2 rounds and who else is there. The #1 overall seed is coming from the ACC, Duke or UVA, and I don't see UVA volunteering to take Columbus. As I've noted before, I could see Duke volunteering to take Hartford, but I think Coach K is really going to want the F/Sun slot in Columbia regardless of the other team.
The way I see it
Duke wins vs UNC and wins vs UVA or FSU then Duke is number 1 overall
Duke wins vs UNC and loses to UVA or FSU Duke 1 seed w UVA overall 1
Duke loses to UNC then Duke is a 2 seed