PRIG (Pitchers Rate Index Grades) is a "report card" that provides park adjusted/league relative A+ through F grades for the rate statistics that characterize pitching performance.The PRIG grades are derived from park adjusted, league relative statistics for pitchers. Statistics are park adjusted to even out the differences that result from pitching in a shooting gallery like Coors Field versus pitching in a graveyard like the Astrodome. Statistics are league relative to balance out differences in league batting (DH vs. no DH) and year to year offensive levels.
- Using play-by-play data, the outcomes of each pitcher's opponent's plate appearances (strike out, non-strike out, walk, single, etc.) are split by game site and by starting/relief.
- The plate appearance outcomes for each location are adjusted by the specific park factors for the specific location for each player. The adjusted starting splits are combined for each pitcher, and the adjusted relief splits are combined for each pitcher.
- A pool of regulars is formed for each league. There are separate pools for starters and relievers. The pool selection criteria is based on innings pitched (IP). The players in each league are sorted by number of IP. The lowest number of IP in the first N players in each league becomes the minimum number of IP for a regular. In the AL and the NL, N is 5 times the number of teams in the league. Typically for a full 162-game season this works out to be ~80 IP for starters and ~40 IP for relievers.
- The park adjusted batting rates (AVG, OBP, SLG, ISO, HR/PA, BB/PA, SO/PA, KBB, RC25, ER and RA) are calculated for each pitcher.
- The mean and standard deviations for each pitching rate are calculated for each league using the pool of regulars for each league.
- The pitching rates for each batter in each league are converted to the number of standard deviations from the league mean.
- The standard deviations for players who have changed leagues during the season are combined (weighted by the number of PAs in each league).
- The combined deviations for all players (both leagues) are sorted by descending RA.
- Grades are assigned according to the following scale:
Grade Deviations from Mean A+ 1.50+ A 1.25 A- 1.00 B+ 0.75 B 0.50 B- 0.25 C+ 0.10 C -0.10 C- -0.25 D+ -0.50 D -0.75 D- -1.00 F -INFINITY
PQ stands for "Pitching Quality". The intent of PQ is to try to determine how well a pitcher pitcher is pitching from his peripheral statistics (AVG, OBP, SLG, ISO, HR/PA, BB/PA, SO/PA, KBB, RC). A pitcher's value is determined by how many runs he allowed, but there's always that temptation to try to read the tea leaves (or goat entrails) represented by the peripheral stats.Why? Because luck, both good and bad, can play a part in run scoring. Have you ever seen a pitcher "scatter" nine hits and four walks in seven innings and only give up one run? Have you ever seen a pitcher allow 1 hit and 2 walks in nine innings but give up three runs? Most people would feel that the second pitcher pitched better, but you wouldn't know it by RA. There are other factors like defense and the pitcher's ability to pitch from the stretch that play a part in this.
PQ is calculated as a weighted sum of the standard deviations of the peripheral statistics. The weights have been derived using a differential evolution genetic algorithm that minimized the squared error between the weighted deviation sum and the pitchers' RA deviations. The algorithm was applied to the stats of every pitcher who had at least 80 IP in the period 1982-1996.
When a pitcher's PQ differs significantly from his RA chances are good that his RA will be headed towards his PQ in the future, but that's an entire article in itself.
PRI is a numeric representation of the RA standard deviation grade. League average is 1000 which corresponds to a grade of C.