Yep, the title pretty much says it all. A new season means a new chance to accomplish our goal: Picking the perfect bracket. To do this, we must do a few things different than last year: Learn from our mistakes, expand our horizons, and prepare....prepare.....prepare.
LEARNING FROM OUR MISTAKES
The 2015-2016 was an interesting season, to say the least. For the most part, our analytics got us into the ballpark. I'll highlight a few of the hits, but the point of this exercise is not to pat myself on the back. I'm not trying to get into the ballpark, I'm trying to get front-row seats behind the plate.
- The 2016 Seed Curve Analysis in the Final Edition of the Quality Curve (QC), which conducted a seed-by-seed analysis of the bracket was spot-on when it comes to ball-park accuracy.
- I stated "6-seeds begin a plummet in the QC" and only 1 advanced.
- I stated "7-seeds are way below quality" and the drastically over-seeded 7's both lost.
- I stated "8-seeds are also way below quality" and the only 8 that won played against an 9 that was an 8-seed in disguise, which simply amounts to a coin-flip game.
- So forth for 9's, 10s and 11s if you want to check it out on the link above.
- In the same article in a later section, I said the 2016 curve best resembled 2010, 2012, and 2005 with some non-technical meshing of curves. If you modeled your bracket predictions on the upset totals, the E8 AM and the F4 layout of those 3 years, not only would you have been in the ball park, your seating would have been lower-level with a good possibility of aisle seats.
- The mind of the committee matters! I probably should have put that in all caps as well. I think the approach (and I use that word very loosely) of the committee played a major role in what eventually played out. If I had to guess the approach based on the look of the bracket, I would say the committee, in order:
- Picked their overall #1-seed first (KU, and I agree),
- Picked all of the 1-seeds (UNC, UVA, & ORE, definitely could make the cases for NOVA and MIST, but mostly agree),
- Picked the pod-seeds (2 thru 4 seeds),
- Filled in the rest of the at-large berths by the following criteria:
- First of all, it looks as if the committee showed an over-weighting to conference affiliation rather than team metrics when it came to seeding at-large teams. In terms of 2016 conference RPI rankings, the order went Big-12, PAC-12, ACC, Big-East, Big-10, and SEC. When you are seeding teams based on the quality of their conference rather than the quality of the team itself, you are making huge distortions to the bracket (overseeding, underseeding, or even seeding teams that do not belong in the tournament). If you look at the strength of the seedings of individual teams (especially after the 1-4 seeds), you can clearly see this pattern. Here's the seeding:
- Big-12: 1, 2, 3, 4, 5, 6, 8
- Pac-12: 1, 3, 4, 6, 7, 8, 8
- ACC : 1, 1, 3, 4, 6, 10, 10
- BEC : 2, 2, 6, 9, 9
- B10 : 2, 5, 5, 5, 7, 7, 11
- SEC : 2, 4, 11
- Second of all, even if my guess at their approach is wrong, it still doesn't explain the anomalies in seeding that did appear, most notably Stephen F. Austin and Middle Tennessee State. STFA was given a 14-seed in 2016 even though the same team with the same core of players received a 12-seed in the 2015 tournament. I find it hard to believe that the same team (minus their 2nd leading scorer) in a relatively weaker year (based on QC analysis between the two years) goes down two-seeds. Needless to say, this improved team (with one year's tournament experience under their belt) in a relatively weaker year pulls off the 3v14 upset. I'd chalk this up to bad seeding by the committee. The same goes for MTSU. If the NCAA seeded by conference RPI, they probably should have looked at other conference metrics like enrollment, public/private, and financial prowess of the conference and its member universities as well. If you think about it, teams (or universities) with larger enrollments, public financial backing, and larger cash flows in general should have more opportunities to get better talent/skilled players than those with less/private/less status. Based on this observation alone, I would never give a 15-seed to a Conference USA team, unless they happen to be 10-20 and win their conference tournament as a 7+ seed. MTSU did not meet any of the disqualifiers, and in my opinion, should have been no lower than a 12-seed (but not seeded higher than STFA). Why the committee decided to give them a 15-seed below teams (and/or conferences with these disqualifiers) is something that I will never understand. I know there are other teams that were poorly seeded by the committee, feel free to list them and why in the comment section.
- In summary, figuring out the mind of the committee may be a better factor in predicting match-ups (or even upsets) than the actual Xs and Os or efficiency ratings. I remember watching the interview on the CBS Selection Show with the Committee Chairman and the answers he gave seemed like BS to me. He was first asked about ORE getting a 1-seed versus MIST, and he said conference season and tournament performance (and I was okay with that answer), but then he went to the committee's annual default talking point and said, "overall resume" (and that pisses me off). We want to know what resume-criteria you are looking at and what weightings or priorities you are giving to each criteria. The phrase "overall resume looked better" doesn't tell us anything. The same default talking point was given for the next question: Why Michigan or Tulsa over Monmouth?" Monmouth had the Non-Conference SOS and Road W-L that the committee said they were looking for, and Tulsa certainly did not. So why did this committee go back on its word? My point is simple: Pay attention to these kind of inconsistencies, they may just pick a game or two, especially those games in the coin-toss category.
- Injuries matter! Around tournament time last year, I provided a Resources Page with a link to a good source for injuries. I regret two things: 1) It being at the bottom of the page and 2) It only being one source. While injuries for the round of 32 and later rounds cannot be predicted with any accuracy, the round of 64 games can change due to a sudden injury. Case in point: 4v13 CAL vs HAW. News sources began to report late Wednesday night that CAL's Tyrone Wallace suffered a broken hand in their practice. Then, minutes before the game started, PG Jabari Bird was declared out with back spasms (This is incredibly difficult for bracket pickers as this occurred Friday, not Thursday). While it is impossible to prove a hypothetical game in which a "healthy" CAL defeats HAW, it was obvious during the game that HAW was not afraid of CAL from the perimeter with two CAL starters on the bench in street clothes. They concentrated their efforts on deterring dribble penetration, and this worked wonders as CAL's best player Jaylen Brown picked up two charging fouls in the game trying to force offense against a paint-packed defense. Ultimately, these two injuries resulted in a 11-point upset of 4-seed CAL. Incidents like this make me wonder if there is a WAR-stat (Wins Above Replacement) for college basketball. You have to logically assume that the minutes-per-game usually consumed by Wallace and Bird were not nearly as efficient or productive in the hands of CAL's 9th and 10th man in the rotation. While the cry "Next Man Up" may inspire the new player to rise to the challenge, he is 9th or 10th in the rotation for a reason: If he could do on a consistent basis what the 3rd or 4th man in the rotation does, then he wouldn't be 9th or 10th in the rotation.
- The mind of 18-22 year-olds matters! If I had any stats that could measure this aspect of the game, I'd have my own primetime show on a major sports network. Nonetheless, we have to remember that we are dealing with 18-22 year old males, and this includes anything that could possibly motivate them. In the last couple of years, we've seen a myriad of stories from the players themselves as to why they thought they were successful. Last year, SYR players felt they were disrespected by a 10-seed and the sports media saying they should have been on the bubble or seeded lower. I remember the sports media trashing SYR's seeding, stating they should not have been seeded higher than GONZ, UNI or WICH (all of whom were Mid-Major 11-seeds). SYR showed what they thought of these analysts and ran straight to the Final Four. In 2014, the eventual National Champion UCONN defied all metrics with their motivation: their denial of participation in the previous year's dance. Shabazz Napier said it loud and clear after their final victory, "This is what you get for keeping us out of the tournament!" Even this year's National Champion had the motivation of getting the monkey off their backs. In 2014 and 2015, NOVA failed to advance beyond the first weekend of games even though they carried a 2- and a 1-seed in those years, respectively. In 2016, they were the last team standing (with some help in the final game, of course). Shane Falco said it best, "Pain heals. Chicks dig scars. Glory lasts forever!" Whatever the motivating factor, it seems to work really well for testosterone-flooded adolescent males, so if possible, try to take this into account in 2017.
EXPAND OUR HORIZONS
To pick that perfect bracket, we use data-based predictions following high-percentage rules. These are the methods to our madness, but we can't be content with what we currently have. We need more!
- One of the first things that I am doing is seeking out other advanced data sets. We currently incorporate the KenPom ratings into our system, but I want to see what the other sets are telling us. Are they showing the same patterns? Are they focusing on a different detail that KenPom could be leaving out? That's why we are expanding our horizons. We have to see what else is out there and whether or not it can help us.
- For the current season, KenPom changed the formula for his ratings, and I'm not really sure if what worked in the past for our bracket picks will continue to work this year. This is a very good reason to seek out other sources. (If you want to see what he changed, here is a LINK to the article in which he explains the changes.)
- The data sources that I am following daily are below (and if you know of any others, link them in the comments below):
- KenPom
- Sagarin Ratings
- Sports Reference (Ratings, Advanced Stats, Opponent's Advanced Stats)
- I am taking a daily approach to the numbers. I am copying each day's data, which calculates the ratings for all games completed on the previous day. (If you want to do this for yourself you can, and if I happen to miss a day, I would love to have a backup plan). However, I do hope this approach will give valuable insight into one of the most talked about yet least quantified aspects to March Tournament teams (but I am going to leave you hanging on what that aspect is).
- I've never wanted this project to be simply me. I know what I think, say and believe, but for 20 years, the perfect bracket has still eluded me. I would like another perspective, or two, or three. If you would like to get your ideas, your methods, your data/ratings/rankings out there, get in contact with me. Let me be clear: I do not make any money from this site. I want this to be a community project in the mold of BracketScience. You are always welcome to challenge my ideas or methods in the comments section, but having your own perspective in article-form would seem to be more prestigious. I am picky though:
- It must be free of typographical and grammatical errors.
- If it is an opinion piece, those opinions must be substantiated with physical proof.
- If it is a ratings piece, you can conceal your methodologies, but I will conduct my own calculations to see if I can find patterns (to validate that it simply isn't your own personal Top 25 or Top 50).
- It must be oriented to the game of college basketball and not any other person, place or thing.
- If I have forgotten something, I reserve the right to include that stipulation.
PREPARE.....PREPARE.....PREPARE
The last piece of the puzzle is my plans for the whole year. I plan to publish an article every other Wednesday, and then in March, I'll try to do a more articles leading up to the Selection Show on Sunday Mar 11 and the first official tournament games on Thursday Mar 16. The dates are set below, but the corresponding article is flexible.
- Nov 23: Welcome Article
- Dec 7: Investigating the Aggregation Model
- Dec 21: Warming Up The Crystal Ball
- Jan 4: January Quality Curve Analysis
- Feb 1: February Quality Curve Analysis
- Feb 15: The New Meta
- Feb 22: Most Talked About, Least Quantified Metric (Yeah, I left you hanging in suspense again.)
- Mar 1: March Quality Curve Analysis (By this point in the season, we should know if the Quality Curve using the new KenPom Ratings methodology is reliable or not.)
- Mar 8: ?????
- Mar 12: 2017 Final Quality Curve Analysis
- Mar 15: Final Last-Minute Article
That should wrap up this introduction. Let's make 2017 the year we get perfect!
No comments:
Post a Comment