Results 1 to 14 of 14

Thread: pre-season composite ranking

  1. #1
    Join Date
    Nov 2017
    Posts
    321
    Contribute to D2
    Rep Power
    35414

    Default pre-season composite ranking

    I want to share something I have put together for a few years now and hope some of you will enjoy. This is a composite ranking of the four systems (that I know of) that rank D2 schools each week. To be in the composite a team must be in the top 25 of at least one of the rankings. So currently, 36 teams meet that basic criteria and I'd like to think those are the favorites to at least make the playoffs.

    I know Massey and Bennett aren't accurate at this time as they rely on game data for their rankings, which we have none. Then again, at this time the human polls have nothing to rely on either. So all in all everything is a crap shoot to start preseason. Stat systems (Massey and Bennett) do have one (huge?) flaw in comparing the MIAA due to the silo schedule, there is nothing to grab outside of the MIAA to judge against, but we all know that. So those systems tend to really appreciate MIAA schools, as you'll probably see as the season goes on. If this is something others like, I'd post weekly updates as the new rankings come out.

    On the right you'll see a couple other indicators, one is the average rank of the four polls...if they don't show up in the D2 poll, which doesn't go beyond 25, then they are moved below into 3 ranking tier, and if they aren't mentioned in AFCA as receiving votes but show up in the Massey or Bennett they are moved to the 2 ranking tier. The second indicator is rank discrepancy* which is simply how far the highest and lowest ranking for the team is apart. As you can see, the rankings generally agree TAMU-C is ranked accurately and generally agree is the best team in the nation to start. VA State on the other hand is liked enough by D2 to be in the top 25 and absolutely hated by Massey. (of note, VA State plays two FCS schools to start, so they may go 0-2 and we still not know much about how good they are until after week 3). I also color code by region as that becomes very important later in the year and it's always good to know what schools are where.

    I also have a visual map of all the ranked teams by location if you want to get a sense of the landscape. This map is also a lot of fun when we get deeper in the season to get an idea where playoff teams may have to travel.
    https://www.google.com/maps/@38.3692...p5Pm8vG2krtGXs

    Enjoy! and comment



    *spelled wrong in the document

  2. #2
    Join Date
    Oct 2004
    Posts
    19,332
    Rep Power
    640208

    Default Re: pre-season composite ranking

    Nice

  3. #3
    Join Date
    Nov 2008
    Location
    Seattle, WA
    Posts
    183
    Rep Power
    15123

    Default Re: pre-season composite ranking

    Now we need to add them to the pick'em to see how ranking and ratings compare to actual game performance.

  4. #4
    Join Date
    Jul 2008
    Location
    St. Peters, Missouri
    Posts
    4,472
    Rep Power
    182850

    Default Re: pre-season composite ranking

    In what ranking is Lindenwood in the top 25?

  5. #5
    Join Date
    Nov 2008
    Location
    Seattle, WA
    Posts
    183
    Rep Power
    15123

    Default Re: pre-season composite ranking

    Quote Originally Posted by bigsig50 View Post
    In what ranking is Lindenwood in the top 25?
    Bennett d2 rankings on hero sports lists Lindenwood as 25.

    However, data is from 11/9 so it's not super up to date. But this is covered in main post

  6. #6
    Join Date
    Nov 2017
    Posts
    321
    Rep Power
    35414

    Default Re: pre-season composite ranking

    Quote Originally Posted by da_miner View Post
    Bennett d2 rankings on hero sports lists Lindenwood as 25.

    However, data is from 11/9 so it's not super up to date. But this is covered in main post
    Bennett did...something...in the last few days. So while it says 11/9, I know the teams have shifted from where they ended the season ranked. So, it's not a pre-season per se, they've at least wiped out data to prepare for the new season. I like Bennett as the season goes on as they incorporate margin of victory into their system.

  7. #7
    Join Date
    Jul 2008
    Location
    St. Peters, Missouri
    Posts
    4,472
    Rep Power
    182850

    Default Re: pre-season composite ranking

    Quote Originally Posted by da_miner View Post
    Bennett d2 rankings on hero sports lists Lindenwood as 25.

    However, data is from 11/9 so it's not super up to date. But this is covered in main post
    Dang they love the MIAA. Washburn at number 11?

  8. #8
    Join Date
    Nov 2017
    Posts
    321
    Rep Power
    35414

    Default Re: pre-season composite ranking

    Quote Originally Posted by bigsig50 View Post
    Dang they love the MIAA. Washburn at number 11?
    and why UCO and Pitt make the top 5, who knows, but we all know that's absurd. My guess is it uses some sort of conference rank, having the MIAA #1 overall which makes the SOS for the MIAA teams at this time very very high with no games. Bennett during the season tends to drop MIAA teams as that SOS becomes .500 for every team. Somehow, Massey compensates for this flaw.

  9. #9
    Join Date
    Nov 2008
    Location
    Seattle, WA
    Posts
    183
    Rep Power
    15123

    Default Re: pre-season composite ranking

    Taken from the herosports page describing their algorithm:

    How the BennettRank works

    TEAMS RANKED
    All teams are ranked twice daily after the third week of the season. For example, 438 D3 Women's Soccer teams in 2017.

    Here's how it all works.

    RANKING CRITERIA

    Adjusted Margin of Victory
    Every sports has their own adjustment factor. The overall concept is that there is value in margin of victory. Minimum and maximum thresholds are given to give adequate measure to a win, but minimize the potential impact of a blowout. As an example, for soccer the minimum value for a win is 1.4 goals and it is capped at 4 goals.

    Week of the season
    After Week 4, the weight of each week's games are incremented by 5 percent over the previous week. In this manner, more recent games have more weight than earlier games. After 14 weeks, this weeks games will have 50 percent more weight in the formula than week 4 games.

    Home field advantage
    Yes, there is one for most sports, and it is measurable. As an example soccer has a .4 goal advantage for the home team and we adjust the Adjusted Margin of Victory depending upon whether the team is home or away.

    RANKING METHODOLOGY
    Two factors are used to compute a preliminary ranking. It is the sum of the average Adjusted Margin of Victory of a team, and the average opponent ranks. It is an iterative process that takes between 10 and 100 iterations. More simply, it is a function of your team's average margin of victory and the opponent's average ranks. Based on a team's raw scores, which are scaled between 1.000 and 0.000, the teams are ranked.

    TIMING
    BennettRanks usually take about six games within the division to make intuitive sense. Therefore, we do not calculate any BennettRanks until week four of the season. Updated rankings are performed daily at 10:30 p.m. and 7:30 a.m. Since football has only one game per week, we use an alternative methodology until week seven, when most teams have played six games within the division.

    EXCLUDED GAMES
    All games played outside of the division are NOT included in the ranking, although they will show up on the HERO Sports website.

    RESULTS
    BennettRank consistently outperforms the RPI and Coaches Polls in predicting post-season tournament results. See numerous articles on this subject right here.

    BennettRank's methodical analysis and incorporation of strength of schedule concepts into it's formula are the main factors in such results. We observe, especially for D2 and D3, that Coaches Polls tend to overweight raw wins and underweighs strength of schedule. BennettRank mends that.
    Since last season they ranked MIAA's SoS really high, it means the MIAA schools will be ranked higher.

    As part of the preseason composite, BennettRank is pretty useless since it hasn't been run yet, and that's old data/nonsensical data.

  10. #10
    Join Date
    Nov 2017
    Posts
    321
    Rep Power
    35414

    Default Re: pre-season composite ranking

    Quote Originally Posted by da_miner View Post
    Taken from the herosports page describing their algorithm:



    Since last season they ranked MIAA's SoS really high, it means the MIAA schools will be ranked higher.

    As part of the preseason composite, BennettRank is pretty useless since it hasn't been run yet, and that's old data/nonsensical data.
    Agreed, but it wouldn't change much if I removed Bennett at this point. The biggest impact would be West Florida would jump up to 4th, but they are in the top 10 currently so I'm not going to boo hoo them. And Pittsburg would drop to 24th, which is probably a bit more accurate on them than being 15th. All the other teams would stay the same or move up or down just a spot or two.

    I've liked doing the composite for myself for the last few years as it will expose flaws/quirks of each of the individual rankings. I particularly like seeing teams that the rankings really disagree, like CSU-Pueblo, why would humans like them better than computers? and teams like Assumption that all four basically agree on even though that team is in the middle of the pack.

  11. #11
    Join Date
    Nov 2008
    Location
    Seattle, WA
    Posts
    183
    Rep Power
    15123

    Default Re: pre-season composite ranking

    Let's try taking some of the previous season on the massey website and run it through 538's NFL prediction algorithm and see how the rankings look afterwards.

  12. #12

    Default Re: pre-season composite ranking

    I don't see the standings so......................... FAKE NEWS!.....BTW You all would be very very poor without me.

  13. #13
    Join Date
    Oct 2004
    Posts
    19,332
    Rep Power
    640208

    Default Re: pre-season composite ranking

    I had never heard of Bennett

  14. #14
    Join Date
    Oct 2015
    Posts
    2,961
    Blog Entries
    59
    Rep Power
    484779

    Default Re: pre-season composite ranking

    Quote Originally Posted by Predatory Primates View Post
    I had never heard of Bennett
    Bennett=Hero Sports. Maybe they’ve improved, but they were the ones that had 9 NSIC schools (including a couple 4-7 teams) ranked in their initial top 25 in 2014. Top 3 schools were all NSIC, while defending champ NWMSU was ranked 4th even though they had almost everyone back. Obviously they weren’t far off with NW with the way the season ended, but to this day, I have no idea where all the love of the NSIC (read: data points) came from before the season started.

    Mankato ended up playing for the Natty that year, so hopefully (for MIAA fans) it will mean good things this year.

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •