Monday, December 17, 2007
Consistency is inconsistentPosted by Sal Baxamusa
I screwed up when I presented the Weibull plots and the consistency numbers last week! Based on reader feedback, I changed the model that I used to compute the Weibull parameters. You can read about the changes in this article. Those changes did impact (pretty signifcantly) the lists I presented previously, so I wanted to present the revised lists here.
Scored 0-2 runs more frequently than expected LAN +9.6 SLN +5.9 NYA +4.5 ... ARI -5.7 PHI -9.6 MIL -11.7 Scored 8+ runs more frequently than expected SLN +6.2 SFN +5.8 KCA +5.5 ... CLE -5.5 CHA -5.7 BAL -6.7 Offenses, ranked by most consistent to least consistent MIL 15.6 BAL 9.4 PHI 8.0 FLO 7.4 CHN 7.0 SEA 5.5 CHA 4.0 CLE 3.8 OAK 3.7 TBA 2.9 SDN 2.0 CIN 1.4 MIN 0.8 ARI 0.7 PIT 0.6 MON -0.1 HOU -0.1 NYN -0.8 ANA -1.0 COL -1.6 DET -1.9 ATL -1.9 NYA -2.1 SFN -2.1 TOR -3.1 BOS -3.8 TEX -4.4 LAN -5.8 KCA -6.4 SLN -12.7
It also affects the conclusion that consistency is not correlated with overall offensive prowess. That is incorrect; it turns out that good offenses are less consistent, but this correlation is quite weak (r2=0.10, if you're a stat wonk). In other words, you shouldn't intentionally build a weak-hitting team in search of a more consistent offense - you're not guaranteed to get one and but you'll definitely end up scoring fewer runs! You can download all the revised plots and reports by clicking here.
Sal Baxamusa is a graduate student in chemical engineering. He can be reached here.