BBTF's Newsblog Discussion :: Baker: Bill James doesn’t expect Mariners to improve on 2009’s 85-win season

Baseball Primer Newsblog
— The Best News Links from the Baseball Newsstand

Thursday, March 18, 2010

Baker: Bill James doesn’t expect Mariners to improve on 2009’s 85-win season

That theory states the Mariners were really more of a 75-win squad last year that overachieved through luck and by winning a ton of close games. There are those who dispute the “luck” contention regarding the Mariners and who espouse other statistical formulas that suggest Seattle was truly an 85-win team.

But James is not one of them.

“Everybody agrees and I agree that the Mariners’ front office has done a fantastic job,” he said. “On the other hand, it remains true that they overperformed last year. I think they gave up more runs than they scored and still won 85 games.”

James then added: “Historically, the great majority of teams that overperform by that margin relapse the next season.”

His new book, “The Bill James Gold Mine 2010”, contains a section on the Mariners, as well as his “Young Talent Inventory” and a review of the Cy Young races. In the book itself, James is noncommittal about whether Hernandez or Greinke was more deserving of the award, but he tipped toward the Mariners pitcher in an interview this week.

“The Mariners scored 149 runs in Felix’s starts and went 25-9,” James said. “The Royals scored 117 for Greinke and went 17-16. There’s a difference in the bullpens of course. But then, the Mariner bullpen was struggling for some of the year and the Royals have (Joakim) Soria, so that’s not the whole thing, either. So, I’m not arguing with anybody else’s numbers and I love Zack Greinke. But I’m not certain that Greinke is better than Felix. I think they were on the same level, or that Felix was an inch ahead.”

And make sure to catch Baker’s extensive Interview with Bill James.

Repoz Posted: March 18, 2010 at 02:09 AM | 14 comment(s) | Login to Bookmark
Related News: General, Sabermetrics, Seattle

Reader Comments and Retorts

Go to end of page

Statements posted here are those of our readers and do not represent the BaseballThinkFactory. Names are provided by the poster and are not verified. We ask that posters follow our submission policy. Please report any inappropriate comments.

Page 1 of 1 pages

1. Rough Carrigan

Posted: March 18, 2010 at 02:47 AM (#3481274)

Bill. It's a process.

2. Walt Davis Posted: March 18, 2010 at 04:18 AM (#3481303)

who espouse other statistical formulas that suggest Seattle was truly an 85-win team.

Other than the "statistical formula" that they did win 85 games, anybody got any idea what they're talking about?

3. SteveF Posted: March 18, 2010 at 04:55 AM (#3481313)

I have to assume third order winning percentage. In other words, they gave more runs than the number/type of hits allowed would otherwise suggest and/or they scored fewer runs than the number/type of hits allowed would suggest.

Edit: And looking over at BPro, third order winning percentage has them as an 83 win team. Not too far off of 85.

4. David Cameron Posted: March 18, 2010 at 04:57 AM (#3481314)

Baker's referring to my argument that you can't just stop at pythag and pretend that RS/RA somehow equals true talent level. If you're trying to strip unrepeatable performances out of 2009 data, there's no reason to stop at runs scored and runs allowed, as there is quite a bit of "luck" that goes into translating batting lines into run totals. The Mariners did very poorly with runners in scoring position last year, so their RS total is about 30 runs lower than what it would have been with normal "clutch" hitting. If you drill down to expected RS/RA, you find that the M's actually played like an 83 win team last year, not that much different than their actual record.

Of course, I think the entire premise of projecting a team based on prior season record (or pythag, or third order wins, or whatever) is pretty stupid anyway.

5. Avoid running at all times.-S. Paige Posted: March 18, 2010 at 05:07 AM (#3481317)

Is what James is saying then not true: that historically most teams relapse the following year if they overperform their run differential by a lot? If it isn't, then I agree, it's silly to use the previous season's pythag as any sort of indicator. But if it is true, then the term "stupid" seems harsh.

6. David Cameron Posted: March 18, 2010 at 05:28 AM (#3481318)

It's true, but it's a generalization that is basically just describing regression to the mean. It's also true to say that almost every team that wins 95+ games in one year does worse in the next year, but that doesn't mean that they weren't a 95 win team the year before - it's just normal regression. By definition, teams that over-perform their pythag record in a year are going to be likely to regress because we've selected teams that had good years. It's a lot easier to go from 91 wins to 85 wins than from 91 to 96, regardless of your true talent level, so any time you select a group of teams that did well, you'll find that they did worse as a whole the next year.

As for the projection thing, I just don't see the point of using the performance of a different roster in a one season sample as a starting point for our expectations when its just not that hard to develop a true talent estimate of the current roster using multiple years of data. Teams change, players performances change... there's just no real reason to use pythag or WAR or third order wins or anything else from last year to project this year. Use CHONE or ZIPS. The extra two minutes it takes to look up a true talent estimate is worth it.

7. Avoid running at all times.-S. Paige Posted: March 18, 2010 at 05:37 AM (#3481319)

:) I get what you're saying. I'm sort of defensive because until the advent of readily accessible projection systems, I was rather reliant on the previous season's performances in anticipating what was going to happen next. Found it was a nice, quick and dirty way to "project" next season...at least in conversations with fellow office mates.

8. Rich Rifkin Posted: March 18, 2010 at 07:08 AM (#3481326)

"If you drill down to expected RS/RA, you find that the M's actually played like an 83 win team last year, not that much different than their actual record."

This page shows all of the second and third order wins and losses from 2009. What Dave Cameron says is confirmed by the "third order" AEQR/AEQRA numbers.

This is how the AL West would have finished last year using expected runs (3rd):

Team - AEQR - AEQRA
Angels 86.8 - 75.2
Rangers 85.3 - 76.7
Mariners 83.0 - 79.0
Athletics 81.8 - 80.2

In other words, the AL West was fairly tight. If Sheets does not fall apart and Duchsherer's arm does not fall off, the A's may be a bit better; Seattle should be better once Lee comes back from his suspension and King Felix is King Felix and Chone is Chone; the Rangers could be better with a manager who is not doing lines off the trainer's table; and the Angels should suck.

9. Athletic Supporter leads the nation in drifters Posted: March 18, 2010 at 07:46 AM (#3481329)

If Sheets does not fall apart

Tooooo late.

10. Fancy Pants Handle is the AntAgonizer Posted: March 18, 2010 at 08:00 AM (#3481330)

Is what James is saying then not true: that historically most teams relapse the following year if they overperform their run differential by a lot?

Well, it is true in the abstract. The problem is that in general, the teams that overperform their pythag as a group should* be relatively average in the other "luck" categories, which go into run scoring/preventing, that have allready been mentioned. So when look at a specific example, where two factors were opposing each other, you should expect each of those to regress, and those effects will at least partially cancel each other out.

*I'm guessing, I have no evidence for this

11. Voros McCracken, Human Shield Posted: March 18, 2010 at 08:41 AM (#3481331)

Every time I've run a regression in terms of predicting next year's record, the team's actual record adds no predictive value when used alongside actual win percentage. This is with an intercept so the regression argument doesn't really apply.

That doesn't mean that the differences are all luck or that they don't mean anything, just that in the aggregate that's the result you get. So I wouldn't doubt the existence of exceptions for whatever reason, just that they are indeed exceptions and the general rule is it doesn't mean much. Generally after about three years even the pythags start not to matter anymore.

Something a little interesting is that if you go a few years back, a negative relationship between actual record three or four years ago shows up that sometimes is very close to a statistical significance standard (IE, the better your actual record then, the worse the predicted record in the future). This may be due to behavior from front offices or more likely draft position or it just might not mean anything since it's not a very strong indicator.

12. Ron Johnson Posted: March 18, 2010 at 10:07 AM (#3481337)

#5 Around 3 decades ago James identified 7 indicators as to whether teams would decline or improve. I suspect James mentions the pythag issue because it's the strongest factor at work in the case of the Mariners.

And David, it's actually stronger than regression. Two forms of regression were among the factors James identified. As a group good teams tended to decline. Further, teams which improved a lot in one season tended to decline the next.

13. Tango Posted: March 18, 2010 at 11:09 AM (#3481349)

Walt: look at OBP and SLG on offense and OBP and SLG on defense. And then include SB/CS numbers (offense and defense).

Seattle was in the slight plus.

I don't see any reason why we would look at individual players' OBP and SLG numbers to establish their talent level, and then throw that away when we look at team's talent level and instead focus on the number of runs scored and allowed.

And of course, no need to stop at OBP and SLG, since those things are a manifestation of their actual talent levels. (So, if you can translate a player's talent level visually, then you don't need numbers at all. Not that we're there yet.)

Anyway, the more luck you add in, the less you know. That's why you need to focus alot more on individual OBP and SLG than on team runs scored.

14. Ron Johnson Posted: March 18, 2010 at 11:33 AM (#3481354)

Worth noting that Tom has mentioned a 4th of James' factors. Actually Paul Johnson came up with it first, James adopted it (after checking that it did in fact work)

The other 3 factors were quality of AAA team, record from August 1 and team age.

Page 1 of 1 pages

You must be Registered and Logged In to post comments.

<< Back to main

Thursday, March 18, 2010

Baker: Bill James doesn’t expect Mariners to improve on 2009’s 85-win season

Reader Comments and Retorts

Support BBTF

Bookmarks

Hot Topics