Tuesday, October 11, 2016

Reports Summary: Games Through October 9

The series of posts below this one are my weekly reports covering games through October 9.  In order, they are:

Weekly RPI Report: Games Through October 9.  This is teams' actual Adjusted RPIs as of this stage of the season.  There are some differences from the NCAA's calculated ratings and rankings, due to a couple of errors in this week's NCAA data base.  Both errors involve games that are listed twice in its data base:  Northwestern 1 v Michigan State 0 and Monmouth 5 v Niagara 0.  I'm sure the NCAA will correct these errors before it releases next week's ratings.

For this report, based on the last 9 seasons, here are the teams that are potential seed and at large selection candidates:

#1 seed:  Teams ranked #19 or better
#2 seed:  Teams ranked #31 or better
#3 seed:  Teams ranked # 45 or better
#4 seed:  Teams ranked #53 or better

Likely assured of at large selection:  Teams ranked #17 or better

Potential, but not assured of, at large selection:  Teams ranked #18 through #86

Teams very unlikely to get at large selection:  Teams ranked #87 or poorer

2016 Season Simulation:  Week 8 Update (Through October 9).  This is a simulation of the entire season, including conference tournaments.  It is a hybrid system, using actual results of games already played and simulated results for games not yet played.  The simulated results are based on ratings assigned to teams at the beginning of the season.  Now that conference play is underway, each week I revisit the conference tournament brackets I've set up, based on actual conference game results to date and simulated results for games not yet played.  Once I've updated the brackets each week, I then simulate conference tournament results in the same way I simulate other results, based on ratings assigned at the beginning of the season.  The two places I use current ratings rather than simulated ratings is to determine seeding order if two teams are tied in conference points scored and to determine who advances if a simulated result is a tie.

2016 NCAA Bracket Simulation: Week 8 Update (Through October 9).  This is a simulation of the NCAA Tournament at large selections and seeds, if the Women's Soccer Committee follows its decision patterns over the last 9 years.  The Automatic Qualifiers are based on my conference tournament simulations.  The code for the left hand column is:

1 = #1 seed
2 = #2 seed
3 = #3 seed
4 = #4 seed
5 = Unseeded Automatic Qualifier
6 = Unseeded at large selection
7 = Next teams in line

2016 Season Simulation: How Teams Have Progressed, Through October 9.  This shows how teams' simulated rankings have changed over the course of the season, as I've updated the simulation with actual results each week.  It shows which teams are performing just about as the simulation thought they would, which teams have performed better, and which teams have performed more poorly.  By looking at how teams' simulated ranks have changed from week to week, one also can get an idea of whether teams appear to have stabilized at their likely "true" rankings or whether they appear still to have further room for movement in the rankings.


  1. Your adjusted RPI rankings seem to reward teams for being in a fairly good or competitive conference, or for playing highly ranked RPI teams, even if they have lost every game against a high rpi team. Va. Tech and Iowa State, based on their results, seem overranked, for example. Va. Tech played North Carolina, Virginia and Duke, and lost all three games by a combined score of 9-0. Their best result is a tie against Clemson, yet they have an ARPI in the 30s. Same with Iowa State, which I think has a losing record in its conference--a conference I think is overrated--and yet has an ARPI in the 30s. Who have they beaten? I saw another writer's NCAA picks and he had both Va. Tech and Iowa State in the tournament. No team with a losing conference record should ever be in the tournament. cheers.

  2. Thanks for your comment.

    Just to be clear, it's the NCAA's adjusted RPI, not mine.

    Regarding a team's record in relation to its strength of schedule, on average the ARPI effectively weights them at roughly 50% each. A team almost always will have its rating hurt by a loss, but if the loss is to a strong team the pain won't be as much as for a loss to a weak team. It might surprise you to know that, based on very detailed studies I've done, the ARPI actually favors teams from weaker conferences. In other words, on average it underrates teams from stronger conferences and overrates teams from weaker conferences.

    When it comes to NCAA Tournament selection time, the Women's Soccer Committee doesn't consider only the ARPI. Among other factors, it also considers head-to-head results, results against common opponents, results against top teams (roughly the top 60), and conference strength and the team's standing within the conference.

    For a team from a mid-major that wants to be a contender, what it needs to do with its scheduling is schedule a good number of strong opponents during the non-conference part of the season. Thus the early part of the team's season would be its strongest opposition and the latter part of the season its weaker opposition. This is the opposite of what a good number of power conference teams do.

    Regarding a losing conference record, your suggestion is one I've never seen before -- and I've seen a lot of suggestions. The NCAA has a policy of not selecting teams with overall losing records for the NCAA tournament, but that's it.

    When you get to asking "Whom has a team beaten?", and I'd add "or tied," if you have another contender you like in mind, you also need to ask whom the contender has beaten or tied. This is where mid-major teams sometimes get in trouble in the NCAA tournament selection process. Using Virginia Tech as an example, at the end of the season, they'll end up having a tie against a team likely to be seeded in the tournament. Suppose you're a mid-major contender for an at large spot. Can you match that result? This is why mid-majors need to have strong non-conference schedules.

  3. Thanks your response. If 50 percent of RPI is strength of schedule, then that most certainly benefits almost every team in the ACC, for example, even if several of the teams are not very good. Va. Tech seems a pretty clear example. They play good ACC teams and lose--badly--but benefit in terms of RPI from their schedule. They have two ACC wins--against the current nos. 11 and 12 teams in the conference. Their biggest achievement is a tie against the no. 5 team, Clemson. If you put a team with a losing conference record in the NCAA, a 9th or 10th place team in the ACC, for example, you are rewarding a mediocre team because of the conference it plays in. That's bogus, IMO. Better to reward teams that have more wins against solid if not stellar opponents--which I suppose comes down to body of work over the course of a season (and I would value late-season record more than early-season record), not just a tie against Clemson.

  4. I think it would be good not to pay too much attention to teams' exact rankings right now. Over the time I've tracked the NCAA Tournament selection process -- the last 9 years -- no conference ever has had a #9 or #10 team get an at large selection.

    It sounds like you may have a mid-major conference, or possibly mid-major team(s), you're interested in. If so, then if you will let me know which conference or team(s), I'll give you a detailed analysis of how they stand in terms of at large selection.

    If you would like a complete education on how the RPI works, how it compares to other possible rating systems, and how the Women's Soccer Committee makes its at large selections for the NCAA Tournament, I suggest you go to this Blog site's companion website, RPI for Division I Women's Soccer: https://sites.google.com/site/rpifordivisioniwomenssoccer/.