Monday, November 7, 2016

Methodology and Preseason Predictions

It's time for my full season preview, which will hopefully give a good example of the things I'll be talking about throughout the season. I'll begin by explaining a bit of my selection methodology, which will come into play once I start making full brackets, and from there I'll give some thoughts on what to expect nationally, in the Big Ten, and with Michigan basketball in particular.

METHODOLOGY NOTES: I'd like to give a quick overview of how I make my selections, and to use this time to remind you that my picks aren't necessarily who I think should be in the tournament, but rather who I believe the selection committee will choose, and where they will seed each team that makes it in. There are two main components taken into account when deciding if a team will get in and where they will be seeded:

-Rankings Factor: This is the starting point--a weighted average of the rankings systems the selection committee has stated that they take into account when making their selections. While the committee has said they have de-emphasized the RPI because it's stupid and flawed (okay maybe I said that last part), further research shows that the RPI is still the #1 factor used in determining where a team is seeded.

-Resume Factor: This is the much more inexact part of the science, but I'm always working on tinkering with my analysis to come up with a more scientific approach to comparing resumes. These are things like "Top 25 Wins," "W/L vs. RPI top 50," "Bad Losses," etc. and it's very difficult to discern how the committee values each of these things compared to each other. However, every selection they've made in past years sets a bit of a precedent, so my main method here is to compare a team's resume to that of teams from the past few years, to see how the committee likes to seed a team with a resume similar to that of the team in question.

As for evaluating how my bracket ends up doing, an explanation on bracket grading can be found here. Basically, a team correctly seeded is worth 6 points, a team whose seed is off by one is worth 4 points, and a team who is correctly picked to be in the tournament but off by more than one seed is  worth 3 points. Thus, a perfect score would be 408, the all-time record is 351, and 320 is considered a very good score. We'll see how things fare for me in my first season of bracketology.

SEASON PREVIEW: Because of the fact that most of my initial rankings factor is based on RPI, which isn't a very meaningful ranking system for the first weeks of the season and at this time doesn't even exist yet, it would be silly for me to publish a full-on preseason bracket, as it would entail me just wildly guessing where teams will fall. That being said, allow me to instead make just a few wild guesses relevant to your interests.

1 Seeds: Don't expect any hot takes or bold predictions here. My predicted 1 seeds are Duke, Villanova, Kentucky, and Kansas. Every preseason ranking you'll find has Duke as the best team in the country, and I see no reason to contend that. Villanova and Kentucky will both roll through their respective conferences and should find themselves on the 1 line as well. The last one is a bit of a tossup, and I expect Oregon to have a strong chance of snatching a 1 seed the way they did last year, but Kansas will have more chances to prove themselves with games against Indiana, Duke, Kentucky, and a strong Big XII conference.

Big Ten Preview: The Big Ten has sent seven teams to the tournament in each of the last two seasons under its new 14-team configuration, and that doesn't look to change this season. Here are my preseason projections for which Big Ten teams will make the tournament and what seed they will be:

Wisconsin - 2 seed
Michigan State - 3 seed
Indiana - 4 seed
Purdue - 5 seed
Ohio State - 5 seed
Michigan - 6 seed
Maryland - 9 seed

Penn State

Wisconsin is the most experienced team in the league and looks like an early favorite to win the conference. Michigan State, on the other hand, has a talented group of freshmen that figure to improve as the year goes on, but inexperience and a very difficult nonconference schedule will keep them just outside the 1 or 2 seed zone. Regardless, expect to roll your eyes at every expert as they all pick the Spartans to win it all come tournament time. Down at the bubble, Maryland and Iowa both look set up to backslide after impressive seasons last year. My guess is that the experience of Melo Trimble running the show at Maryland will keep them in the tournament, and the utter lack of talent returning at Iowa will put them on the outside looking in this season.

Michigan Preview: I have Michigan as a slightly optimistic 6 seed. I see them as a team that hovers in that 6-8 range most of the season, but it's not hard to see them ending up as a 6. This is a chance to try out my fancy new cumulative wins calculator, which takes win probabilities for individual games provided by Kenpom, runs all the different scenarios while adding up the win totals for each scenario, and graphing them based on the probability of the team getting each number of wins.
This is Michigan's nonconference record distribution, which shows they have about a 30% chance of going 10-3 over their first 13 games, with a 25% chance of going 11-2. They have chances to build their resume with top 50-ish games against Marquette, Pitt or SMU, Virginia Tech, Texas, and UCLA. Going 3-2 for those five games would be a great start to the season. As for the conference schedule, things become a bit tougher to predict:
10-8 is the most likely outcome, but even that is only at 20%. And records as extreme as 8-10 or 12-6 still have probabilities in the 10-15% range. As Michigan fans have known for the past few seasons, things like injuries, unexpected improvement or regression out of specific players, or wacky nights where guys have unusually good or bad shooting performances can have a huge affect on final conference records.

All that being said, I like to look at this graph and see that Michigan has about a 62% chance of getting 10 or more conference wins. Tack on another game or two in the Big Ten Tournament and I see Michigan going something like 22-11 with around 8 top 50 wins and rankings in the low to mid 30s. That is a resume I feel very comfortable placing as a 6 seed based on previous selections by the committee. 

I'll be giving updates on the national picture, the Big Ten picture, and details about where Michigan is at with regards to the NCAA Tournament approximately every week throughout the season. Be sure to stay up to date, as the outlook can change dramatically from game to game. Once conference play starts is when the real fun happens and I start posting my full bracket predictions. Until then, enjoy the first week of this new season. 

No comments:

Post a Comment