tag:blogger.com,1999:blog-23241716.post1226978272249150269..comments2007-10-26T18:16:15.212-04:00Justinhttp://www.blogger.com/profile/07697776280178146413noreply@blogger.comBlogger4125tag:blogger.com,1999:blog-23241716.post-13908712494959213772007-07-06T13:11:00.000-04:002007-07-06T13:11:00.000-04:00Hi Joel,<BR/><BR/>I probably didn't attach enough qualifiers to my previous post, as I wasn't disagreeing at all with your assessment of the data you provided.<BR/><BR/>As for WPA, I see it's value more an indication of what happened, rather than a tool that can be used to figure out something that will happen. David Appleman showed that it does have some predictive value, but then again, so do RBI's. I sort of treat WPA as an informative measure of performance in situations, so in that way it is somewhat like RBI totals (though I'd argue it's more informative).<BR/><BR/>If someone has high or low WPA relative to their other stats, I don't get particularly hyped up about that as far as what I expect from that player in the future. But I do find it useful to understand how specific players have contributed to games over the course of the season (historically). It helps me ground past player performance in the reality of the games that were played. So that's why I like it. But I agree that you have to be careful in how you use it.<BR/><BR/>There are some things that could be improved about it, but I still think it's useful. You don't have to agree, of course. :)<BR/>-jJustinhttp://www.blogger.com/profile/07697776280178146413noreply@blogger.comtag:blogger.com,1999:blog-23241716.post-50241104284117786412007-07-06T08:37:00.000-04:002007-07-06T08:37:00.000-04:00A couple comments. First, they are in the 1 RS and 2 RS bucket more than we would like, but that is balanced by being in the 0 RS and 3 RS bucket less. I know this isn't right, but in my head, I tend to group everything under 4 runs together as those are games that I don't expect the team to win. And the Reds have only lost one game where the pitching staff has given up less than 3 runs, so you can pretty much move the numbers around in those 4 buckets and at most it gives the Reds one win. That's why I didn't think it was a big deal that the 1 & 2 buckets were a little high.<BR/><BR/>Secondly, on WPA, I don't think I'll ever be comfortable with it as any more than a way to look at the ebb and flow of a game. I've never been able to get through that "leap of faith" required to accept the assigning of WPA to individuals, let alone to start aggregating it over games. I can see how it can be useful to identify something like why they fall short of their pythag, but it doesn't necessarily give me any information as to what needs to be fixed. Does that make sense?<BR/><BR/>I appreciate that people like yourself keep looking at it though because I see potential there, but I'm just not sure it's been properly fleshed out yet.Joelhttp://www.blogger.com/profile/08493717107262478265noreply@blogger.comtag:blogger.com,1999:blog-23241716.post-52507304316793238842007-07-05T14:03:00.000-04:002007-07-05T14:03:00.000-04:00Joel, thanks for that. It's one of those cases where I should have checked this out more carefully before making the claim. The Reds do seem to be a bit higher than usual in the 1-2 runs scored bins, but you're right that they don't seem to be in the upper bins more often than expected, especially for a team that is scoring as many runs as they have. <BR/><BR/>The issue of any production on offense being influenced by the performance of pitching (and vice versa) is a significant limitation of WPA. But WPA also gets at something that simply looking at runs scored does not, and that's the historical tendency to score (or prevent) runs when it counts. This can help us figure out why the Reds, for example, are five games below what is expected by Pythagoras, as I <A HREF="http://jinaz-reds.blogspot.com/2007/05/why-have-reds-deviated-from-pythagoras.html" REL="nofollow">toyed with here</A>. Yes, "when it counts" for the offense will be determined in part by the defense, but that's the way games work. <BR/><BR/>Supporting WPA's critique of the Reds' offense, we can compare their overall performance versus performance in key situations (via ESPN):<BR/>Overall OPS: 0.756 (5th in NL)<BR/>Rnrs Scoring Pos: 0.722 (13th in NL)<BR/>Scr Pos, 2out: 0.696 (10th in NL)<BR/>Close and Late: 0.595 (last in NL)<BR/><BR/>To be clear, I'm not claiming that the Reds offense needs a major overhaul, or that it's the major problem with the Reds' team. It is a good offense, which is why it's scored the 5th-most runs in the league. I'm just saying that it's production hasn't matched the game situations very well this year.<BR/><BR/>If it makes you feel any better about WPA, teams that show great pitching and poor offense can show negative offensive WPA's and yet still show positive pitching WPA's. The San Diego Padres, for example, have starter WPA = 6.97, reliever WPA = 6.78, and yet have offensive WPA = -7.26. <BR/><BR/>WPA isn't a perfect stat, but I think it's worth a fair bit more than a grain of salt.<BR/>-jJustinhttp://www.blogger.com/profile/07697776280178146413noreply@blogger.comtag:blogger.com,1999:blog-23241716.post-88277134470787266672007-07-05T13:20:00.000-04:002007-07-05T13:20:00.000-04:00One minor disagreement, J. The Reds don't tend to score in bunches, at least not compared to the rest of the league. Here is the run distribution compared to the NL average in parentheses:<BR/>0 RS - 2 (4.5)<BR/>1 RS - 10 (8.7)<BR/>2 RS - 11 (9.7)<BR/>3 RS - 9 (13.2)<BR/>4 RS - 12 (10.7)<BR/>5 RS - 12 (10.3)<BR/>6 RS - 7 (7.9)<BR/>7 RS - 7 (5.3)<BR/>8 RS - 3 (4.3)<BR/>9+ RS - 11 (9.4)<BR/><BR/>I'd say they're pretty close to average, scoring in the higher buckets slightly higher than normal and scoring below 4 runs at a fairly normal rate. <BR/><BR/>I think the bigger problem is that while the rest of the league is winning almost 80% of their games when their offense scores 5 runs or more, the Reds have only won 55% of such games. You touched upon that in your second point about the offense, but I think it is actually the major reason, as usual, as to why this team is still losing.<BR/><BR/>This is why I'm still not comfortable with WPA for this type of evaluation. It seems to me like it punishes one side for the mistakes of the other side. If the offense scores 6 runs in the first 5 innings, but is being matched along the way by the other team, the WPA value of those runs isn't that high. Then if the other team takes the lead early enough, the offense gets nothing but negatives for the rest of the game if they fail to score more runs, even if they've scored a reasonable amount already to win.<BR/><BR/>I may be missing something, but it seems like the offense gets punished because the pitching staff can hold a lead for very long.Joelhttp://www.blogger.com/profile/08493717107262478265noreply@blogger.com