Posted by burnsy483 on 3/3/2015 1:28:00 PM (view original):

Posted by MikeT23 on 3/3/2015 1:20:00 PM (view original):
Team construction matters.    As I pointed out earlier, good teams that strike out a lot are using players who do something useful when they're not striking out(like an Adam Dunn who strikes out 28% of his plate appearances but does something positive in 36% of his PA).   Bad teams use a Brett Wallace who whiffs 33% of the time while only doing something positive 31% of the time. 

As far as whiffing, they're practically the same guy but Dunn is much more valuable.

This sounds like a reason to use BL's model. If you use tec's model it makes Adam Dunn look like Brett Wallace. No?

Posted by MikeT23 on 3/3/2015 1:32:00 PM (view original):

Posted by burnsy483 on 3/3/2015 1:28:00 PM (view original):

Posted by MikeT23 on 3/3/2015 1:20:00 PM (view original):
Team construction matters.    As I pointed out earlier, good teams that strike out a lot are using players who do something useful when they're not striking out(like an Adam Dunn who strikes out 28% of his plate appearances but does something positive in 36% of his PA).   Bad teams use a Brett Wallace who whiffs 33% of the time while only doing something positive 31% of the time. 

As far as whiffing, they're practically the same guy but Dunn is much more valuable.

This sounds like a reason to use BL's model. If you use tec's model it makes Adam Dunn look like Brett Wallace. No?

No, tec's model gives them equal weight. 

A league average is a league average.   There's no room for smarter GMs, better managers, pitching match-ups, etc, etc.   It's just what happened in baseball.

No one is debating on whether Boston is better run than the Cubs.   And that's what happens when you break it down to teams.


I'm pretty sure tec's entire point is "Scoring is down, strikeouts are up."     Not in Boston or Anaheim but in baseball.   He can correct me if I'm wrong.

Posted by tecwrg on 3/3/2015 1:32:00 PM (view original):
An example: two of the data points that BL would be using would be the 1996 Pirates and the 1996 Cardinals.

Both had pitching staffs that whiffed batters at the MLB season average of 6.5 K/9.  But Pittsburgh surrendered 5.16 R/9, while St. Louis only surrendered 4.38 R/9.

Pittsburgh had a ****** staff that surrendered over 200+ more hits than did St. Louis that season.  You can't compare their K/9 rates to their R/9 rates because you're comparing a good pitching staff to a bad pitching staff at that team level of detail.

Only when you aggregate the data at a MLB-wide level can you start to draw any meaningful conclusions because now you're summarizing and aggregating the talent as a whole and not at discrete and misleading levels of detail.

And when you do that over the past 20 years, you see a very clear and consistent correlation of strikeouts to runs scored.

Well, there you have it.   BL's model is useless when making tec's point.

Which, of course, is what BL wants.

Posted by bad_luck on 3/3/2015 1:34:00 PM (view original):

Posted by tecwrg on 3/3/2015 1:32:00 PM (view original):
An example: two of the data points that BL would be using would be the 1996 Pirates and the 1996 Cardinals.

Both had pitching staffs that whiffed batters at the MLB season average of 6.5 K/9.  But Pittsburgh surrendered 5.16 R/9, while St. Louis only surrendered 4.38 R/9.

Pittsburgh had a ****** staff that surrendered over 200+ more hits than did St. Louis that season.  You can't compare their K/9 rates to their R/9 rates because you're comparing a good pitching staff to a bad pitching staff at that team level of detail.

Only when you aggregate the data at a MLB-wide level can you start to draw any meaningful conclusions because now you're summarizing and aggregating the talent as a whole and not at discrete and misleading levels of detail.

And when you do that over the past 20 years, you see a very clear and consistent correlation of strikeouts to runs scored.

Are you looking at K's by pitchers or by batters?

Posted by dahsdebater on 3/3/2015 1:07:00 PM (view original):
BL's method is unambiguously better.  With the variance in baseball, Tec's sample size is irrelevant.  The fact that BL's data contradicts Tec's also substantially invalidates Tec's numbers.  Even if on the aggregate it can be made to appear that strikeouts correlate to runs, if it doesn't hold on a team-by-team basis but only when large numbers of teams are averaged, it indicates that the correlation isn't real.  Or at the very least, is indirect, which I think is what just about everyone believes anyway.  Except maybe tec.

More Ks > less balls in play > less hits > less runs.  Everyone agrees on that.

The only disagreement is whether a player who hits .300/.360/.500 with 180 Ks is less valuable than a player who hits .300/.360/.500 with the same defense, same baserunning, etc, with 75 Ks in the same number of PAs.

Posted by tecwrg on 3/3/2015 1:44:00 PM (view original):
It works the same for batters, though.

Same season, 1996.  Houston and Seattle batters both struck out at the same league wide rate of 6.5 K/G.  But Houston only scored 4.65 R/G, while Seattle scored 6.17 R/G.

Different hitters, different ballparks, different offensive production despite the same rate of K's by their hitters.

Posted by MikeT23 on 3/3/2015 1:45:00 PM (view original):

Posted by dahsdebater on 3/3/2015 1:07:00 PM (view original):
BL's method is unambiguously better.  With the variance in baseball, Tec's sample size is irrelevant.  The fact that BL's data contradicts Tec's also substantially invalidates Tec's numbers.  Even if on the aggregate it can be made to appear that strikeouts correlate to runs, if it doesn't hold on a team-by-team basis but only when large numbers of teams are averaged, it indicates that the correlation isn't real.  Or at the very least, is indirect, which I think is what just about everyone believes anyway.  Except maybe tec.

More Ks > less balls in play > less hits > less runs.  Everyone agrees on that.

The only disagreement is whether a player who hits .300/.360/.500 with 180 Ks is less valuable than a player who hits .300/.360/.500 with the same defense, same baserunning, etc, with 75 Ks in the same number of PAs.

Is there anyone who takes the 180k player?   

Posted by bad_luck on 3/3/2015 1:46:00 PM (view original):

Posted by tecwrg on 3/3/2015 1:44:00 PM (view original):
It works the same for batters, though.

Same season, 1996.  Houston and Seattle batters both struck out at the same league wide rate of 6.5 K/G.  But Houston only scored 4.65 R/G, while Seattle scored 6.17 R/G.

Different hitters, different ballparks, different offensive production despite the same rate of K's by their hitters.

So you're saying the K rates don't track well to the runs scored rates?

Posted by tecwrg on 3/3/2015 1:51:00 PM (view original):

Posted by bad_luck on 3/3/2015 1:46:00 PM (view original):

Posted by tecwrg on 3/3/2015 1:44:00 PM (view original):
It works the same for batters, though.

Same season, 1996.  Houston and Seattle batters both struck out at the same league wide rate of 6.5 K/G.  But Houston only scored 4.65 R/G, while Seattle scored 6.17 R/G.

Different hitters, different ballparks, different offensive production despite the same rate of K's by their hitters.

So you're saying the K rates don't track well to the runs scored rates?

Using your incorrect way of looking at the data, correct.  Because you're looking at very disparate levels of talent that are also impacted by potentially significant external factors.

That's why you have to aggregate the data and look at it at a higher summarized level.  Which is what I've done.

Posted by bad_luck on 3/3/2015 1:57:00 PM (view original):

Posted by tecwrg on 3/3/2015 1:51:00 PM (view original):

Posted by bad_luck on 3/3/2015 1:46:00 PM (view original):

Posted by tecwrg on 3/3/2015 1:44:00 PM (view original):
It works the same for batters, though.

Same season, 1996.  Houston and Seattle batters both struck out at the same league wide rate of 6.5 K/G.  But Houston only scored 4.65 R/G, while Seattle scored 6.17 R/G.

Different hitters, different ballparks, different offensive production despite the same rate of K's by their hitters.

So you're saying the K rates don't track well to the runs scored rates?

Using your incorrect way of looking at the data, correct.  Because you're looking at very disparate levels of talent that are also impacted by potentially significant external factors.

That's why you have to aggregate the data and look at it at a higher summarized level.  Which is what I've done.

Ok, so, based on your awesome 20 data points, would you say that league-wide run scoring would increase if players turned some of their strikeouts into other types of outs?