RPI FAQ and What-ifs?

September 27, 2019 Working General

Print

BASICS QUESTIONS AND ANSWERS ABOUT THE RPI

• RPI was be implemented in all sports as a publicity tool.
• In Football, the RPI will be incorporated into the competition rules as a means of seeding Rounds 3, 4 and 5
• RPI stands for “Ratings Percentage Index”. In short, it is a way to measure a team’s strength relative to other teams, based largely on the strength of their schedules.

What percentages are being used in the formula?

• Starting in 2019-20, the formula will be as follows for all sports:
  • Each game RPI = GR*((0.35 × WP) + (0.35 × OWP) + (0.30 × OOWP)).
  • The formula, which equally weights winning percentage and opponents winning percentage, has been the basis for nearly all states as they have started the use of the RPI.
  • Balancing the formula in this manner, with a lesser impact on the opponent’s opponents also prevents schools that are located next to other states (be mindful that around forty counties in our state border at least one of seven different states) from being adversely affected by scheduling regular season non-district games against out-of-state opponents.
  • The formula will be reviewed at the end of each school year for possible future revision.
  • Total RPI is summative for the season up to and including the moment of release
  • The RPI will be calculated each night
  • GR=Game Result. 1.0 for a win, .5 for a tie, 0 for a loss
  • WP=Winning Percentage of the Team being reviewed
  • OWP=Opponents Winning Percentage
  • OOWP=Opponents’ Opponents Winning Percentage
  • The WP for each team will be adjusted in football to consider a class factor due to the scheduling nature of the sport.

Why use RPI instead of another type of rating system?

• A major advantage to the RPI is the transparency that comes along with its accuracy.
• The RPI is not a poll, and there is nothing subjective about its calculation as it is solely about calculations based on the data.
• The components of the formula are known, and its results can be easily replicated.
• In addition, there is no incentive for a team to defeat a team by a higher margin, margin doesn’t matter.

What data will be used in the RPI formula?

• All varsity games played in the regular season shall be counted toward the RPI calculation.

Where will be the RPI standings be published?

• The official KHSAA RPI standings will be published on khsaa.org.
• It is worth noting that any RPI data published elsewhere is unofficial and should be viewed with skepticism as it may not be correct.

Why does the RPI change after Saturday football games when only one or two games are played?

• The key is to remember that EVERY game matters and there will not be (and has not been) any changes made in calculation methodology since the adoption of the RPI in January of 2019.
• For each school being review, the WP calculation in football considers nine opponents. So if one of them plays, the results will change.
• For each school whose OWP is being calculated, each of their opponents is changing, so now that is potentially 81 (nine opponents times their nine opponents) results changing.
• For each school whose OOWP is being calculated, each of their opponents’ opponents is changing, so that is potentially 729 (nine opponents times their nine opponents times their nine opponent’s opponents) results changing.
• So truly, it is a constantly moving calculation that could change with every game played.

How are the components of the formula specifically calculated?

• For a complete review of the calculations, including the link to an extremely detailed 2018 example, see the post at https://khsaa.org/rpi-calculation-steps/

What changes have been made since the RPI concept was adopted?

• None. Absolutely none. Initially, there was information distributed that gave, in concept, the RPI and constantly referred to the Colorado example.
• Most of that information came from the initial RPI review during 2017-18 when it became apparent that their system was proving to be a good method to rank teams solely based on results, with no incentive for blow-out games, etc.
• Original discussions involved various factors including combining the existing computer ratings service results, etc., however it was apparent that all of those inherently rewarded teams for “running up the score”, which was never desired.
• There were some media types who tried initial calculations based on unverified formula, etc., and those individuals whose results differ from the final product have made claims about changes from the original system. Such has never occurred.
• Frankly, had those media types inquired about the original formula, they would have been referred to the KHSAA’s constant referral, the CHSAA website, for details as the model in Colorado has always served as the prototype.
• The system ALWAYS factored in the class of the teams being played as this is nearly essential, especially in football when there are less games played than most sports meaning many less observations.
• It is important mathematically to be able to differentiate the schedules played by different teams and reward those who play up (although the reward is ever so slight with only a 15% difference between classes and the baseline for the 15% increase starting at 1.0.
• Several iterations of the calculations were reviewed to be sure the formula was correct, but the concepts, adopted values for things like out of state game, class factors, etc. have never changed.
• The system will be reviewed after 2019-20 to determine if any revision is necessary, but no revision is anticipated.

How will out-of-state opponents be handled?

• All out-of-state opponents will be handled in the following manner:
  • All out of state teams will be assigned a .500 WP for all calculations. Were this not the case, schools would be chasing tens of thousands of opponents of out-of-state opponents over the course of a season, and there is no way to ensure the accuracy of that data.
  • The .500 figure was selected because it is the average value of opponents’ opponents winning percentages across all sports.
• This same value of .5 is also assigned to any in-state home school competing against a member school.

How will international opponents be handled?

• Due to the difficulty involved with obtaining accurate data on a consistent and timely basis from schools outside the U.S., international opponents will not count toward the RPI calculation.

What happens if two teams are tied in the final RPI standings?

• The tiebreaker will sequentially be as follows for this unlikely scenario. It is as follows:
  • The RPI is calculated to infinity by the computer system, however, only the first three decimal places are displayed on the website. Should there be a tie, the first tie-breaker will be the complete calculated RPI, first at five decimal places, and then beyond five decimal places.
  • Head-to-head result between the two teams
  • Winning percentage
  • Opponents’ winning percentage
  • Opponents’ opponents winning percentage
  • Highest-rated win (according to the final RPI standings)
  • Next-highest rated win (exhaust all possibilities)
  • Coin flip – The only reason for the coin flip is as a last result if all other scenarios happen to be tied.

What happens if three teams are tied in the final RPI standings?

• The tiebreaker will be as follows for this unlikely scenario. It is as follows:
  • Step One
    • The RPI is calculated to infinity by the computer system, however only the first three decimal places are displayed on the website.
    • Should there be a tie, the first tie-breaker will be the complete calculated RPI, first at five decimal places, and then beyond five decimal places.
    • Head-to-head result between the three tied teams to determine if one has defeated the other two
    • If one has defeated the other two, that team wins the tiebreaker and the two team tiebreaker shall break the remaining tie
  • Step Two
    • The RPI is calculated to infinity by the computer system, however, only the first five decimal places are displayed. Should there be a tie, the first tie-breaker will be the complete calculated RPI, beyond five decimal places.
    • Head-to-head results between the three tied teams to determine if one has lost to the other two
    • If one has been defeated by the other two, that team is eliminated from the tiebreaker and the two team tiebreaker shall break the remaining tie
  • Step Three
    • If Steps one and two do not break the tie, the following steps shall be used in sequence until one of the steps breaks the tie, and then the two-team tiebreaker shall break the remaining tie
    • Overall Winning percentage
    • Opponents’ winning percentage
    • Opponents’ opponents winning percentage
    • Highest-rated win (according to the final RPI standings)
    • Next-highest rated win (exhaust all possibilities)
    • Draw – The only reason for the Draw is as a last resort if all other scenarios happen to be tied in which case the draw team would prevail in the tiebreaker and the two team tiebreaker shall break the remaining tie.

How should teams be scheduling?

• The main thing to remember with the RPI is it takes an entire schedule into account, and in general, a single game will not result in a significant change in the calculations.
• Do not fret over scheduling one game, but instead, see the entire schedule as a whole and try to judge if it will be tough or not.

Does the score of the contest matter in the RPI formula?

• Only in that, it gives a winner and a loser (or results in a tie).
• There is no factor for score differential in the RPI formula.
• A 1-0 win counts the same as 100-1.

Can and how can, an RPI be greater than 1?

• When game wins have a value of 1.000 then the RPI will always be from 0.000 to 1.000 using the formula RPI= (0.35 * wp) + (0.35 * owp) + (0.30 * oowp). This is the way it is at the KHSAA for sports other than football.
• In football we modify the value of game wins based on whether a team is playing up or down in class. If playing up in class a win will have a value of greater than 1.000; if playing down in class, a game win will have a value of less than 1.000.
• For example, if a Class 1A team beats a Class 6A team the value of the game win is 2.011. If a class 6A team beats a Class 1A team, the value of the game win is only. 0.497.
• This isn’t a problem since RPI is a way to compare teams in the same sport in the same season.

How can a winless team have an RPI?

• RPI is made up of three components: WP, OWP and OOWP. WP is the winning percentage of the team and, if they don’t have any wins will be 0.000. OWP is the winning percentage of the opponents playing team. For OWP to be 0.000 all of the team’s opponents would have to have a record of complete loses. Finally, OOWP is the winning percentage of all of the opponents of the team’s opponents. For OOWP to be zero all of the opponents of the teams’ opponents would have to have a record of complete loses.
• Hard to imagine that scenario where RPI is zero. Someone has to win somewhere.
• This points out the advantage of RPI over a simple win-loss statistic. RPI takes into consideration so much more!

What happens if a game is canceled and can’t be rescheduled?

• Because the RPI system works off of averages, it won’t make a difference in the final formula if a game cannot be rescheduled.
• It would not penalize, nor benefit, any team involved in that scenario.

How do schools that drop programs affect the RPI?

• If a school drops a program prior to the start their competitive season, no forfeits will be involved.
• Instead, their opponents now have open dates in the sports where they were scheduled to play the teams who dropped their program, and they are free to try and find another game, if possible.
• If a school drops a program after their competitive season has started, that team shall forfeit their remaining games to their opponents.
• In this instance, the competitive season is defined as the first day a team plays in a contest.

Where should we be reporting scores?

• Continue reporting scores to Riherds.com/KHSAA scoreboard.
• The official RPI feed will be calculated off of results entered into that platform.

How often will the RPI standings be published?

• The RPI standings will not publish for the first time until a season nears the midway point.
• Once they do, feeds will be updated nightly during the regular season, with future consideration to a change each time a game result is reported.

My team didn’t play this week, how come it’s RPI changed since last week?

• Your opponent or one of your opponent’s opponents played a game or its score was changed

How often will you calculate RPIs?

• Daily

Do RPI calculations include postseason games?

• No. Calculations stop the last day of the regular season.

Why do postseason games not count?

• If we were to factor in postseason, the impact of seeding skews the data.
• In approximately 2/3 of districts in all sports except football, and in all football districts, postseason play is seeded.
• The teams in a higher-seeded position could be inherently disadvantaged in that they would almost always play a four-seed (or lower) whose WP and probably OWP would be lower, making further comparisons inaccurate.

Other people do RPI calculations and they differ from yours. Who do I believe?

• The KHSAA RPI is the official RPI. As you can tell by the detailed example, since the KHSAA has the only comprehensive score database with required reporting, it is incredibly accurate

Print