rokhsar@lasspvax.UUCP (Dan Rokhsar) (08/30/85)
For those of you who don't have access to "Earned Runs Prevented" statistics, (especially those St. Louisians who seem to think that being the first to win 20 games by a few days is preferable to being consistently the best pitcher in baseball), here's the top 10 in the national league as of Tuesday, August 27 ERA Innings ERPRev. W L Gooden, NY 1.78 218 40.2 20 - 3 Tudor, St. Louis 2.12 208 31.4 15 - 8 Valenzuela, L.A. 2.37 217 26.8 16 - 8 Dravecky, San Diego 2.16 175 25.5 11 - 8 Hershiser, LA 2.37 175 21.5 13 - 3 Garrelts, San Fran. 1.61 90 18.7 7 - 3 Andujar, St Louis 2.75 219 17.7 20 - 7 Hesketh, Montreal 2.49 155 17.1 10 - 5 B. Smith, Montreal 2.64 181 16.9 15 - 4 Burke, Montreal 1.77 87 16.5 8 - 2 (note that these statistics do not include Wednesday nights game against Cincinnati, when Andujar failed to hold a 6 run lead; this game will probably knock him off the top 10 list next week). I do not claim that any of the above statistics are perfect (for more discussion on this point, see any article with the heading "Carter vs. Pena"), but clearly Andujar is NOT the premier pitcher in the league by any standards; even with the four man rotation compared with the five man rotation Dr. K. pitches in, Andujar has not even been more durable than Gooden, which could have been claimed last year. Statistics for strikeouts and complete games were not included, because it didn't seem necessary to completely embarass those who believe Andujar to be the MVP! A word about Earned Runs Prevented; in the search for more perfect statistics, it would be nice to weight the runs given up by a pitcher to take into account the inning in which they were allowed - runs allowed late in a game are more costly since there is less time to catch up. This would penalize starters who had poor bullpens behind them, but would raise ace relievers like Garrelts and Burke in the "standings"; relievers are after all often called in to preserve a one run game, and if they allow a run it is more damaging to the teams effort. Batting average vs On Base Average: If we knew the fraction of the time that an average player comes up with (a) a man on first (b) an man on second (c) a man on third (d) man on first and second ..., and the number of outs at the time, we could compute (1) the probability that a run scores later in the inning if the player at bat walks, assuming "average" players following him (2) the same for a hit (perhaps Slugging percentage and Batting Average can be turned into a rough guess at how many of each type of hit the player will get) (3) The probability that a run will score in the inning (maybe expected number of runs scored) if the player makes out (here perhaps strikeouts/at bat could be used to estimate the probability of a "useful" out, like a sacrifice) The resulting function of Batting Average, Slugging Percentage, On Base Pct. will yield the average number of runs scored which the player at bat participates in by either prolonging the inning, moving a runner to scoring position, RBI, etc. If we subtract from this what the "average" player with average Batting Avg., etc. would do, we get a sort of opposite of "Earned Runs Prevented", a RCT (runs contributed to). As a start, does anyone have the probabilities that a man will be on first with no outs, etc? This would be an interesting but not, repeat not perfect statistic. One Last Question: If a team with a .500 winning percentage plays another .500 team, each will win roughly half of the games, assuming all sorts of things are, well, average. Does anyone know or can anyone guess a form for the probability P(x,y) of a team with winning percentage x beating a team with winning percentage y? Some "boundary" conditions are P(x,y) + P(y,x) = 1 P(x,.500) = x (since x measures the winning percentage against an average team) P(0.500,x) = 1-x (combining the two equations above) P(1.000,x) = 1.000 (a team that always wins always wins (x<1.000) P(0.000,x) = 0.000 (a team that always loses always loses (x>0) Sorry if I seem so statistics oriented, but the idea of quantifying such a qualitatively beautiful (sans DH, of course) game is intriguing. Dan Rokhsar rokhsar@lasspvax "Meet the Mets, meet the Mets Come on down and greet the Mets Bring your honey, bring your wife (is this right? it seems pretty Bring your kids sexist today) They'll have the time of their life Because the Mets are really socking that ball Hitting those home runs Over the wall East side West side (something) all around the town Oh its th (something) M-E-T-S Mets In New York town" Can any of the Met faithful fill in the blanks?