Points Created

From QBWiki
Revision as of 19:24, 24 January 2010 by fix>QBWikiBot
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search

A statistic developed by Pat Matthews and Clay Davenport, which attempted to account for the shadow effect, shallow tossup-mongering, and other distortions which prevent PPG from corresponding to the actual ranking of the best players in the tournament.

PC, rather than PPG, was used to rank individuals at several Penn Bowls as well as early NAQT events. The statistic fell out of favor because it is time-consuming to calculate and was never demonstrated to actually have predictive value or otherwise be fundamentally necessary. For other attempts at modifying PPG, see PATH.

From a 1999 QB Mailing List post by Samer Ismail, quoting Matthews, this is how you calculate Points Created:

"For PC/G to work best, you will need to know the following things:

  1. Total tossups heard for every player and team
  2. Total tossups answered for every player and team
  3. Total interrupts for every player and team
  4. Total points scored for each team
  5. Total matches for each team (note that if you know how many tossups each player has heard, you do not need to know how many games that player appeared in)
  6. Interrupt conversion percentage for the field (explained later)
  7. For each player, the number of tossups answered or interrupts taken by teammates while the player was in the game

I usually start by calculating the various field-wise averages:

  • Tossup Value (TUV). The average net gain the field achieved by answering a tossup and playing the subsequent bonus. To calculate, sum the total points scored for the field. Add back the points lost on interrupts. Now divide by the total tossups answered.
  • Interrupt Conversion Percentage (INTconv%). The average rate at which the field "converted" opponent interrupts into correctly-answered tossups. To calculate this, I usually take a random sample of 30 scoresheets, count the number of interrupts, andthe number of conversions. Divide conversions by interrupts to get the INTconv%. Note that the value is positive.
  • Interrupt Value (INTV). 5 (the amount of the penalty) plus the product of TUV and INTconv%.
  • Average Questions Per Round (Avg?). Total tossups read for the tournament divided by total matches played for the tournament. Recall that if you sum the total questions heard by the teams that you'll have to divide by two, as each question is asked to two teams. Ditto for total matches.
  • Average Adjusted Tossups Per Round (AvgATU). Total tossups answered by the field divided by the quantity 8 times total matches played.
  • Average Adjusted Interrupts Per Round (AvgAI). Same as above, except using total interrupts.
  • AvgATU Percentage (AvgATU%). Divide AvgATU by Avg?.
  • AvgAI Percentage (AvgAI%). Divide AvgAI by Avg?.
  • Average Opportunities (AvgOpps). Avg? - 3 times (AvgATU + AvgAI). This tells us how many tossup opportunities are available to the average player (and the opposing team) after we remove the contributions his/her three average teammates are expected to make.

Now you're ready to start calculating individual stats. For each player, you will need to calculate two further averages; these averages, in turn, are used to calculate PC/G.

  • Individual Adjusted Tossup Percentage (ATU%). The formula is tricky:

ATU% = (player's TU answered)/(player's TU heard - teammate TU - teammate INT)

  • Individual Adjusted Interrupt Percentage (AI%). Same as above, except substitute player's INT in numerator.

From that, you can calculate each player's PC/G:

PC/G = AvgOpps times ( TUV times (ATU% - AvgATU%) + INTV times (AvgAI% - AI%) )

  • Some adjustments must be made if you have incomplete information. For example, say you know how many tossups each team was asked and how many rounds each player played, but you know neither how many tossups each player heard nor how many teammate tossups or interrupts occurred while a given player was in the game. You could use the following formula for ATU% or AI%:

T[i] or I[i] = (G[i] / G[t]) times (?[t] - (T[t] + I[t] - T[i] - I[i]) ) + T[i] + I[i]


  • G[i] = individual games
  • G[t] = team games
  • ?[t] = team tossups heard
  • T[t] = team tossups answered
  • T[i] = individual TU answered
  • I[t] = team INT
  • I[i] = individual INT

The adjustment is not perfect, but it does save a great deal of time, even if accuracy is sacrificed.

(Note: This paragraph is Samer's.) I will add that in the last year or so, there was a suggestion to delete the average TU% and INT% rates from the PC/G formula, so that it would give absolute points created or |PC/G| (yes, I know that means absolute value :v), which would reduce the number of negative numbers substantially. As an example, instead of ~50% of the field having negative numbers, at PB6, only one player out of >250 would have had a negative |PC/G|."