i wonder, however, whether it’s scope is too broad and you need to define “significant” (ie does that mean 5% improvement? in what? hr?). the most common claim i hear is that the hitting of major league regulars improves in september (measured by hr or ops). find a group of players over the last 5 years who had played 100+ games pre-september as well as 20+ games in september and compare. obviously, there will be more error in the second group, but seeing whether anecdotal reports are, on surface, correct or wrong would be a good start IMO.
Comment by joshcohen — September 7, 2011 @ 3:36 pm
Comment by joshcohen — September 7, 2011 @ 3:38 pm
An idea for testing this could be seeing how well a player’s season-september numbers correlate to their projection (well rate stats anyways), and then seeing how well the whole season correlates. obviously you would need many years of data/ player seasons because of the small sample of one month.
another test could be the run environment in september vs the rest of the year. is it higher (scrub pitchers), lower (scrub hitters) or the same (evens out)? basically does september call up season make any appreciable impact on the run environment? this would also answer a more specific question of, if one does discount september stats somewhat, whether pitchers or hitters are a greater concern. again you would probably want to use multiple seasons of data.
just a couple of ideas to mull over. i dont know how feasible this would be, so im just spitballing here.
Comment by phoenix2042 — September 7, 2011 @ 3:45 pm
One item is that in the last few days, walks go way down and balls in play go up. Players seem like they want to end the season swinging.