NCAA Tournament: Predicting the Bracket, Track Your Team

Updated October 2017

Starting with the 2017 season, this page and my RPI and Bracketology for D1 Women's Soccer Blog will provide some tools that will help track individual teams' NCAA Tournament at large selection and seeding prospects over the course of the season:

  • Before the season begins, and then weekly once it begins, I'll publish an Excel Website Factor Workbook as an attachment at the bottom of this page that you can use to gain detailed information about your team's NCAA Tournament prospects.  There's an explanation of how to use the Workbook under the heading "How Do I Follow A Team's Prospects," below.
  • Before the season begins, once I have the schedules of all teams, at the Blog I'll publish simulated end-of-season ratings and other data for the entire season.  The simulated end-of-season ratings and data are based on starting pre-season ratings I assign to the teams, as applied to the teams' schedules and as adjusted for game locations.  I assign the pre-season ratings based on teams' historic ARPI averages: For teams with coaches that have been in place for 9 or more years as of the 2016 season, I base the ratings on the teams' ARPI average over the last 8 years; with coaches in place for 4 to 8 years, I base the ratings on the averages over the last 6 years; and with coaches in place for 3 or fewer years, I base them on the averages over the last 3 years.  These average periods, using coach longevity, are the result of a study I did at the conclusion of the 2016 season.
  • As soon as I have the simulated end-of-season ratings and other data, and also before the season begins, at the Blog I'll also publish simulated NCAA Tournament automatic qualifiers (conference champions), at large selections, and seeds.  I base this simulation on the Women's Soccer Committee's history of decisions over the last 10 years.  It says that if the Committee's decisions follow the same patterns they have over the last 10 years, here's what the Committee's decisions would be this year if the Committee were dealing with the simulated end-of-season ratings and other data.
  • Once the season begins, each week I'll update the simulated end-of-season ratings and other data by substituting, for pre-season simulated game results, the actual game results for games already played.  In addition, starting with the fourth week of the season, I'll substitute teams' then current actual ARPI ratings for their assigned pre-season ratings, since the study I did indicates that's the point at which actual ARPI ratings are more likely to match with future game results than are the assigned pre-season ratings.  This means that over the course of the season, as I increase the ratio of actual results to simulated results and as I shift from assigned pre-season ratings to actual ARPI ratings, the simulation will come closer and closer to what the actual end-of-season ratings and other data will be.
  • Also each week, I'll update the NCAA Tournament bracket simulation of automatic qualifiers, at large selections, and seeds.  As with the simulated ratings and other data, over the course of the season, as I increase the ratio of actual results to simulated results and shift to simulating future results based on actual ARPI ratings, the NCAA Tournament bracket simulation will come closer and closer to what the Women's Soccer Committee's decisions will be if the Committee's decisions this year follow the patterns of the Committee's decisions over the last 10 years.
HOW DO I FOLLOW A TEAM'S PROSPECTS?

Here's a series of steps that you can use weekly to help you follow a particular team's prospects.  Following the written steps, there's an example to help you see how it works.  These all will involve using an Excel Website Factor Workbook for which I will provide a link at the bottom of this page, starting with the workbook titled 2017 Website Factor Workbook Preseason, currently at the bottom of this page.  Once the season starts, after each weekend's games, I'll add an updated Workbook that substitutes the actual game results from the prior week for that week's simulated results.

1.  Your Team's Data.  Each week, for all teams, the Teams Individual Factors Data spreadsheet will show 14 simulated individual factor data items for each team, based on my simulated end-of-season ratings and other data.  The data items are based on the factors that the NCAA requires the Committee to use for its at large selections.  For a detailed description of the 14 data items and how they relate to the NCAA factors, go to the "NCAA Tournament: Predicting the Bracket, At Large Selections" page.  (You can use the Navigation bar to the left to go to that page.)  Your first step in evaluating how a particular team is doing is to take a note of its 14 data items.  There's a limitation to the number of data items for teams outside the Top 60 in the ARPI rankings, since for some of the data items I only do computations for the Top 60 teams..

2.  The Committee's Decision Patterns.  At the bottom of this page, I've published a table showing patterns that all of the Women's Soccer Committee decisions have met, over the last 10 years, when making at large selections and doing the seeding for the NCAA Tournament.  The table also is on the Committee Factor Patterns page of each Website Factor Workbook.  There are patterns for each of the 14 data items, and there also are patterns for each of those items paired with each other item, for a total of 92 individual or paired items.  The one exception is the data item of the number of Head to Head games against Top 60 teams (Factor 92 in the table), which I do not pair with other data items.  Here are some notes about the table:
  • In the "Factor" column, after individual data items 1 through 13, you'll see those items matched together in pairs.  For each pair, the Factor column gives a formula for computing a value for that "paired data item."  To get the paired data item value, you go to the particular team's individual data items, identify the value for each individual item contributing to the pair, and plug those values into the formula.  To the left of the "Factor" column, you'll see a "Code" column.  The Code column simply numbers the Factors from 1 through 92.
  • In the "<= or >=" column, you'll see one of two entries for each Factor -- either ">=Yes; <=No" or "<=Yes; >=No."  Next to that column, you'll see a series of paired "Yes" and "No" columns, such as the first two which are "1 Yes" and "1 No," which refer to #1 seeds.  Here's how the columns work together, using a team's ARPI data item and the question of how the team fits the Committee's #1 seed pattern as an example: 
    • For the ARPI, the "<= or >=" column gives the instruction that if a team's ARPI value is >= (greater than or equal) to the value in a "Yes" column then the team will get a "yes" decision from the Committee; and that if the team's ARPI value is <= (less than or equal) to the value in a "no" column then the team will get a "no" decision from the Committee.
    • Applying this to the #1 seeds, if a team's ARPI is >= to 0.7002, the pattern says the team will get a "yes" #1 seed decision from the Committee.  If the team's ARPI is <= 0.6430, it will get a "no" #1 seed decision from the Committee.  And, if a team's ARPI is between those two values, it might or might not get a #1 seed.
3.  Applying the Committee's Decision Patterns to a Team's Data.  Your final step in evaluating how a particular team is doing is to match up the team's data against the Committee's decision patterns, to see what decisions the Committee has made historically for a team with those data.  For the 14 individual data items, this will be easy.  For the 78 "paired" data items, it will be a little more difficult since you'll need to apply the formula in the Factor column to the team's data pair to come up with the paired data item value to use across the rows of the chart for that factor.

4.  Short Cut for Applying the Committee's Decision Patterns to a Team's Data.  The Website Factor Workbook has a third spreadsheet titled "Teams' Factor Results."  This spreadsheet covers teams in the Top 150 ARPI ranks and shows (1) the total number of "yes" decisions each team gets for seeds and at large selection and the total number of "no" decisions it gets and (2) the individual data items for which the team gets a "yes" or a "no."

Here's how to use the Teams' Factor Results worksheet:
  • Use the pertinent week's Website Factor Workbook at the bottom of this page.  The first step, for your convenience, is to create your own copy of the workbook.  You'll be able to do this if you have Excel.
    • Go to the bottom of this page. You'll see the workbook's title. Opposite the title, on the right hand side of this page, you'll see a download arrow.  Use the arrow to download the workbook.
    • Open the workbook.
    • Near the top of the worksheet, in the yellow Protected View box, click on Enable Editing.
    • Use your Save As function to save the workbook in a file and with a name of your choosing.
  • If you don't have Excel, go to the bottom of this page and click on the workbook's title.  This may open the workbook, but not in a very useful format.  Otherwise, on your computer screen, you should be able to use the Open With Google Sheets command to open the workbook as a Google document if you have Google Documents capability.  You then will be able to save the workbook as one of your Google documents.
  • On the Teams' Factor Results spreadsheet, you'll see the following information:
    • The teams will be arranged in the order of their ARPI rankings.  The first four columns will be, in order:
      • (A) the simulated Committee decisions on seeds and at large selections [1 = #1 seed; 2 = #2 seed; 3 = #3 seed; 4 = #4 seed; 5 = unseeded Automatic Qualifier; 6 = unseeded at large selection; 7 = bubble team but no at large selection].  You'll notice that for teams with ARPI rankings poorer than 60, the Automatic Qualifiers are identified with a 5, but there is no entry for any other team.  This is because history says that the Committee does not even consider teams ranked more poorly than 60 for at large selections.;
      • (B) the Automatic Qualifiers;
      • (C) the teams' ARPI Ranks; and
      • (D) the teams.
    • Next, you'll see five pairs of columns, starting with "1 Seed Total" and "No 1 Seed Total" as the first pair, followed by pairs for #2, #3, and #4 seeds, and ending with the pair "At Large In Total" and "At Large Out Total."  The number in each column is the total number of patterns the team meets for that column.  For example, the number for a team in the "1 Seed Total" column is the number of patterns the team meets that say, "yes," a team meeting this pattern always has received a #1 seed over the last 10 years.
    • Next, you'll see a long series of five pairs of color coded columns.  The first five pairs are "1.1 Yes," "1.1 No," "1.2 Yes," "1.2 No," ... through "1.5 Yes," "1.5 No."  Referring to the table at the bottom of this page, the "1.1 Yes" cell for a team shows how the team fared as to the Factor 1 (ARPI) "yes" pattern for a #1 seed; the "1.1 No" cell shows how the team fared as to the Factor 1 "no" pattern for a #1 seed, the "1.2 Yes" cell shows how the team fared as to the Factor 1 "yes" pattern for a #2 seed, and so on.  The entry in a cell will be either a "1," a "0," or a blank.  A "1" means the team met the standard applicable to that cell.  For example, a "1" in the team's cell for "1.1 Yes" means that the team meets that factor's pattern for getting a #1 seed.  A "0" means the team does not meet that factor's pattern for a #1 seed.  A blank cell for a factor means that the factor does not have a pattern for that particular decision or that my system does not provide a factor value for a team at that team's ranking level.  For example, looking at the table at the bottom of this page, the ANCRPI factor (Factor 4) does not have a pattern for "yes" a team gets a #1 seed, so teams always will have blank cells in the "4.1 Yes" column.
    • To use the spreadsheet, find the particular team you're interested in.  You'll easily be able to see the number of "yes" and "no" patterns it meets for each of the seed positions and for at large selection.  Then, by scrolling to the right and looking for the cells with a "1," you'll be able to see each pattern for which the team meets a "yes" pattern and each for which it meets a "no" pattern.  (If your team is fairly far down on the worksheet and you're using it in Excel, you can use the Excel "hide" function to hide the teams above the one you're looking at so that your team's row is immediately below the column headings, to easily match the "1" entries with the column headings.  This will give you a picture of where the team stands in relation to the factor patterns that appear to be the ones that will decide its NCAA Tournament seed and at large selection outcome.

Factor List for 2017



Ĉ
Chris Thomas,
Oct 16, 2017, 3:38 PM
Ĉ
Chris Thomas,
Oct 2, 2017, 4:18 PM
Ĉ
Chris Thomas,
Oct 23, 2017, 11:07 PM
Ĉ
Chris Thomas,
Oct 30, 2017, 4:52 PM
Ĉ
Chris Thomas,
Oct 9, 2017, 3:53 PM
Ĉ
Chris Thomas,
Nov 6, 2017, 10:26 AM
Ĉ
Chris Thomas,
Aug 23, 2017, 11:09 AM
Ĉ
Chris Thomas,
Aug 28, 2017, 3:38 PM
Ĉ
Chris Thomas,
Sep 11, 2017, 4:12 PM
Ĉ
Chris Thomas,
Sep 18, 2017, 1:50 PM
Ĉ
Chris Thomas,
Sep 25, 2017, 2:45 PM
Ĉ
Chris Thomas,
Sep 4, 2017, 11:21 AM
Ĉ
Chris Thomas,
Aug 23, 2017, 11:06 AM
Comments