Sunday, January 16, 2022

Value Add Basketball Game Provided Accurate for Every Era

Since inventing the Value Add Basketball Game a couple of years ago, we have played 175 games between the 136 all-time great teams. We track all the game scores on this google sheet. If you are not technical, you may just want to read that the following teams are having their dunk ranges adjusted based on the study detailed below. We are in progress of making this updates on this google sheet of the 136 great teams. 

TeamYearPreviousAdjustCurrenct
Arizona199751-60add 251-62
Arkansas199451-57add 251-59
Auburn201951-56subtract 251-54
Baylor202151-61subtract 251-59
Colorado202151-60subtract 251-58
Connecticut199951-57add 2 51-59
Creighton202051-56subtract 2 51-54
Dayton202051-56subtract 2 51-54
Duke199251-57add 2 51-59
Georgia Tech1990 51-53add 2 51-55
Iowa2021 51-57subtract 2 51-55
Kansas1997 51-55add 251-57 
Kentucky1996 add 2 
Loyola Marymount1990 add 2 
Loyola-Chicago2018 subtract 2 
LSU1992 add 2 
North Carolina1998 add 2 
Purdue2018 subtract 2 
Texas Tech2019 subtract 2 
UNLV1991 add 2 
USC2021 subtract 2 
Utah1998 add 2 
Villanova2018 subtract 2 
Virginia2019 subtract 2 
Wake Forest1996 add 2

Technical Explanation for Calculating These Adjustments

If you study basketball analytics and want to see how accurate the game is and why these adjustments capture the only outliers continue to read - but if not you can stop here.

With 175 games played we now have enough games to answer the question everyone who plays college board games wrestles with - do you need to make adjustments to account for eras that used or did not use a 3-point shot. Over time we have written about how we estimate how many 3-pointers players from earlier eras would have made and adjusting the overall team cards to make players from different eras equal.

With 175 games played, we have enough data to see if those adjustments worked and if not make further adjustments to player cards in the game.

To do this we tracked the Simple Ratings System calculation for each team in the game, which measures how many points better or worse than the average team each team was. The SRS on the sheet is actually 21 lower than the actual SRS, because we want to compare them to the average team in the game or great teams, which is actually a +21 since the average team in the game is 21 points better than the average team in basketball.

Second, we track a "Game Score" for each team in each game played, which measures how many points better or worst than the average great team the team was in that game. The following table tracks the results:

Ranges of SeasonsSRS-21Ave. Game ScoreToo Good?
1942 to 1959 (no adjustment)-4.8-5.2-0.4
1960 to 1969 (no adjustment)-0.40.91.4
1970 to 1979 (no adjustment)0.30.70.3
1980 to 1989 (no adjustment)-2.4-1.50.9
1990 to 1999 (add 2 to dunk ranges)2.5-0.6-3.1
2000 to 2005 (no adjustment)0.6-1.3-1.8
2000 to 2009 (no adjustment)0.00.70.7
2010 to 2015 (no adjustment)0.2-1.5-1.7
2016 to 2018 (no adjustment)1.92.60.7
2019 to 2021 (subtract 2 from dunk ranges)1.24.02.8

The good news is ...

The Numbers Indicate the Game is Accurate for Teams of all Eras!

No era has averaged being off by more than 3 points when playing other eras.

Teams from the past few years, 2019-2021, when 3-pointers have continued to increase dramatically, have done 2.8 points better against other eras than they should have averaged based on the SRS ratings of both teams in the game.

Therefore, each team from 2019 to 2021 in the game now has a dunk range 2 lower than it was. On the other extreme, the teams of the 1990s when teams were starting to use 3-pointers a little bit, but making the adjustment, is the era for which we do the opposite and will add 2 to each dunk range.

The cards for all other eras have proven so accurate that no adjustment are made.


No comments:

Post a Comment