I found this statistical study of historical box score data on a blog (http://statspeakmvn.wordpress.com/2008/11/24/ah_so_we_meet_again/) which hints that the pitcher-batter confrontation initially favors the pitcher very slightly, and then trends towa…

your table has some errors, and your 50 game minimum leaves some seasons on the table, although that doesn’t explain some of the players you missed.

Your article says players in the last 15 years, but you pick up scattered players before 1994 (counting 15 back from 2008 – latest year available (i assume) in bbdatabank): Tim wallach (1987 but not 1989), Paul O’Neill 1987, Kevin Seitzer 1993, Darrin Jackson 1991, although there were about 50 to find between 1987 and 1993.

You have the right pitching line for Derrick Bell, but associate it with 1996 instead of 2000.In general you seem to list the wrong year for the pitching performance, which probably has something to do with your misbehaving selection criteria.

Your list leaves out Mike Aldred 1996, Mike Benjamin 1997, Wade Bogg’s 2nd season with an appearance (1999), a 2nd year for john Cangelosi 1995 (or 1997? – the pitching lines are similar), Chris Donnels 2001, gary Gaetti’s 2nd and 3rd seasons (1998,1999), shane halter 2000, Tim Laker’s 2nd season (2004), John Mabry’s 2nd season (2001), Frank Menechino’s 2nd season (2004), Kevin Osik’s 2nd season (2000) , Andy Tomberlin (1994), Todd Zeile’s 2nd season (2004).

here is a mysql query with a very compact and reasonably good criterion for distinguishing players who mainly were used as hitters or position players and not as pitchers. I haven’t seen it give any false positives even with low #’s of games pitched and played.

select p.playerID,

p.yearid,

p.bfp-p.hbp-p.ibb-p.so-p.bb-p.hr BIP,

p.h-p.hr biph,

p.hr,

b.g,

p.g

from pitching p

left outer join batting b

on p.playerid=b.playerid and p.yearid=b.yearid and p.stint=b.stint and p.teamid=b.teamid

where ipoutscoalesce(p.g,0)) and p.yearid>1986;

Anyway for 1994 on, position players have allowed 60 non-HR hits on approximately 206 BIP, along with 10 HR. The BABIP is .291, but I have always disagreed with the premise that subtracting out HR helps you measure a distinct skill.