I don't think they'll finish last, but I'd be surprised if a statistical projection system didn't have them with the fewest number of wins. Touting it as "last place", though obscures the fact that they're probably about ten games behind the Blue Jays, not Marlins or Twins bad.

fake edit: Now that I look, they are projected almost 20 games (!) behind the Yankees (!!). And the Blue Jays are in 4th place, not 1st.

The Yankees' injuries haven't moved their W/L total at all which doesn't make sense to me, but PECOTA's projections look off to me on a few other teams as well.

"Me? I'm for wasting sportswriters' time. So I'd like to see if we can hang around and give 'em all a nice big s#*tburger to eat."

I say the same thing in all these threads, but again -- it's one thing to project a regression to something closer to Pythag. But the Orioles Pythag record last year was 82-80. PECOTA (and a lot of other projection systems too) are picking them to win 74 or 75. That's what I don't get. I don't see a ton of players on the 2012 Orioles who were playing way over their heads. Indeed, Lindbergh says in the interview, "So, I don’t think there are a ton of guys that are in for a huge step back this year. I don’t think there were too many guys who were just kind of mirages. It just seems like they probably won’t be quite as fortunate in when they score their runs this season."

So if the 2013 Orioles are putting a team on the field as good as the 2012 Orioles, and the 2012 Orioles were a .500 team on talent, why project the Orioles to win 74 or 75 and not in the low 80s? Has the rest of the AL East gotten _that much_ better?

Has the rest of the AL East gotten _that much_ better?

I think the Blue Jays and Red Sox each should project to 10-15 more wins than last year. That is about a 6 game swing right there.

I don't see the O's bullpen performing as well as last year, but otherwise they should be pretty good. 80 wins seems about right.

It seems like this year's AL East is one of the hardest divisions to predict in recent years.

But I think inside the game really there is not a team that doesn’t look at stats and take stats seriously and do advanced things with stats that we don’t even know about. I think the whole stats vs. scouts debate has tended to be overblown probably from the start and especially now.

A debate between stats guys and scouts is a wonderful idea. Has that ever happened before?

5.

That's what I don't get. 

Projection systems do not and should not care what the team dud last year. They have no knowledge of that as it is irrelevant. They look at the team as it is constructed today and how the players on the team project and that is that.

The fan methodology of startibg with last season's total and adding and subtracting wins does not apply to statistical projections.

A debate between stats guys and scouts is a wonderful idea. Has that ever happened before?

I think a round table would preferable to a debate.

Projection systems do not and should not care what the team dud last year. They have no knowledge of that as it is irrelevant. They look at the team as it is constructed today and how the players on the team project and that is that.

I guess that's what I thought I was doing. The team as it is constituted today consists in large measure of the team as it was constituted in the second half of 2012; that group of players hit and pitched at a somewhat over .500 level, and are projected to play about equally well this year (at least according to Lindbergh.) I think JJ1986's suggestion has merit -- if the Red Sox and Blue Jays have each added so many good players that they're likely to improve by 15 games each, then the 36 games the Orioles play against those teams could lead them to end up with 75 wins, even if their players are exactly as good as players who would have finished .500 against the 2012 AL.

Well, the Yankees look about 10 games worse than last year which nullifies half of the Red Sox/Jays' improvement.

I think the big thing is that many of the the Orioles' players who played so well in the second half just don't project to play that well again in 2013.

I've got the O's around 79-80 wins right now, and the only 90 win team I have in the East is Toronto. I think they'll be in the mix this year.

and made a lot of us look wrong.

Missed projections will do that.

I guess that's what I thought I was doing. The team as it is constituted today consists in large measure of the team as it was constituted in the second half of 2012; that group of players hit and pitched at a somewhat over .500 level, and are projected to play about equally well this year (at least according to Lindbergh.)

I think I would take the under on quite a few of their pitchers.

#9 Bill James had 7 indicators for whether a team was likely to improve or decline:

1. Team age
2. comps between expected runs scored and actual (teams which score a lot more than expected tend to decline the next year)
3. comps between pythag wins and actual (teams which win more games than their pythag tend to decline)
4. W/L record (good teams tend to decline, bad teams tend to improve)
5. change relative to the last year. A team which improved considerably this year is probably not quite as good as its record indicates, because the good performance was probably caused by some
players having better-than-expected years.
6. Record after August 1
7. Record of AAA team (which serves as a proxy for organizational depth. Of course it's perfectly possible to have a good AAA team with no talent of short term major league consequence or have a young impact player on a lousy AAA team)

He never published anything on the relative weights and any of these indicators are likely to be truly meaningful only at the extremes. For that matter, it's worth checking whether these indicators are still valid. He did the work in the early 80s after all.

You can score the Orioles at home, but as I see it #6 is a positive, as is #1 I think. #5 is the big issue and probably carries more weight than anything else.

It seems like this year's AL East is one of the hardest divisions to predict in recent years.

Agreed! I can really see the Yankees, Blue Jays and Red Sox with any win total. I think TB is 85+ wins and I don't think Baltimore is that good (84 or less wins) but I can make a case for the other three to have a wide range of wins.

I guess the numerical negativity this winter is just that they had that best record ever in one-run games, which is not something that teams have shown any ability to sustain from season to season.

I've seen this asserted in pretty much every thread on the Orioles we've had in the last 8-10 months, but has a serious study been done on year-to-year correlation of team record in one-run games? I would assume that there has been and I just haven't seen it.

#17 There have been a bunch of studies on one run games. Don't have any links handy but I can summarize.

One run games are not precisely random, but they are one step removed from it. There's a very weak correlation between team quality. There's a stronger one to bullpen quality (but again it's not that strong) and nobody has been able to find any offensive indicators that tells you anything of use.

I know I specifically looked at base stealing and found nothing (base stealing is quite a bit higher in the late innings of close games but I couldn't find any evidence that this moved the dial in predicting what teams will do well). I also looked at bunting and couldn't find any evidence that team that bunted a lot (or a little) played well in one run games.

Again, teams bunt quite a bit more in the late innings of close games, but I couldn't find any evidence that this mattered.

It seems like this year's AL East is one of the hardest divisions to predict in recent years.

I think the entire league is pretty wide open. I think the only locks are the Twins and Astros to finish last, and the Tigers are a good bet to finish first. I guess the Angels and Rangers are a decent bet to be pretty good. All the other teams could finish anywhere.

I've seen this asserted in pretty much every thread on the Orioles we've had in the last 8-10 months, but has a serious study been done on year-to-year correlation of team record in one-run games? I would assume that there has been and I just haven't seen it.

Not sure about one-run games, but haven't Scioscia's Angels consistently outperformed their pythag?

I'll say again that I don't see the logic of looking at one-run games as an indicator of anything.

If you want to say that the distribution of run differential for the O's was strange given their RS-RA that's fine but look at the whole distribution. Basically, you need to argue that the O's had more 1-run wins than expected but not at the cost of 2+ run wins ... and that their few one-run losses was not due to having more 2+ run losses than expected. This essentially follows from pythag vs. actual for the O's but deviations from pythag don't have to be about 1-run wins.

If you want to look at record when tied after 7 or something, that might well largely be a factor of luck and is presumably what people are trying to get at with 1-run wins. The O's dominated in extra-inning games last year and that looks like luck (with some relation to bullpen) but they also seem to have had a lot of extra-inning games which might be good or bad luck (depending on how they got there). But in those latter two, it's got nothing to do with whether the final differential was 1 or 2 or whatever. An extra inning 2-run win is presumably every bit as lucky as an extra-inning 1-run win.

I understand it's easier to talk to the public about 1-run games. But it's not hard for the public to understand that if a team is ahead 4-2 at the end of 8 and wins the game 4-3 that this was not the outcome of good luck for the winning team and bad luck for the losing team. It is not hard to understand that, to be tied at the end of 9, some luck was probably involved and that, now that it's a series of 1-inning games, luck could be involved in winning it. Instead, scoring 1 in the top of the tenth is luck, scoring 2 in the top of the tenth is skill?

I've seen this asserted in pretty much every thread on the Orioles we've had in the last 8-10 months, but has a serious study been done on year-to-year correlation of team record in one-run games? I would assume that there has been and I just haven't seen it.

Not sure if it qualifies as "serious," but here's a recent one.

Not sure about one-run games, but haven't Scioscia's Angels consistently outperformed their pythag?

Bobby Cox's first 14 full seasons with Atlanta: +32 vs pythag
Bobby Cox's last 6 full seasons with Atlanta: -25 vs pythag

You never know when it's going to turn. Some will say Bobby started to lose it at the end, and having watched it, I can't really say they are wrong.

Not sure if it qualifies as "serious," but here's a recent one.

And was discussed at BTF shortly thereafter.

I'll say again that I don't see the logic of looking at one-run games as an indicator of anything.

If you want to say that the distribution of run differential for the O's was strange given their RS-RA that's fine but look at the whole distribution. Basically, you need to argue that the O's had more 1-run wins than expected but not at the cost of 2+ run wins ... and that their few one-run losses was not due to having more 2+ run losses than expected. This essentially follows from pythag vs. actual for the O's but deviations from pythag don't have to be about 1-run wins.

But it's not like their pythag and 1-run record are saying anything different. Even if you chop their record in one run games by 10 wins to .500, that still doesn't account for the 11 wins they beat their pythag by.

R-squared for all of baseball history for a single team from one year to the next for one-run games is .0137.

For 1980 to 2011 it is 0.008.

It's possible I butchered the calc, but I'm pretty sure that is correct.

Can you tell me what "R" stands for above? I didnt read the article so let me know if it's there. Also is that figure of .0137 signifcant in your opinion? Jes trying to follow the discussion, thanks for any input.

Can you tell me what "R" stands for above? I didnt read the article so let me know if it's there. Also is that figure of .0137 signifcant in your opinion? Jes trying to follow the discussion, thanks for any input.

r-squared is a general statistical method for determining how predictive past results are for future results. Basically the closer it is to 1, the more predictive it is, the closer it is to zero the more random it is.
.0137 means it's extremely random.

For 1980 to 2011 it is 0.008.

So the question is - what can a team do to improve their fortunes in One-Run games? Is that the million-dollar question?

What should we forecast a team to perform like in 1-run games? .500?

So the question is - what can a team do to improve their fortunes in One-Run games? Is that the million-dollar question?

Be a better team.

[quote[What should we forecast a team to perform like in 1-run games? .500?
Very close to it, especially for a near .500 team. Since you cannot win a game 0 to -1, a team that never scores will have a .000 record in 1 run (and all other) games. So if run scoring is very low, then there will be a (small) effect from getting shut out. So throwing lots of shut outs in a low run environment should get you a better than random 1-run game record. And also >1 run game.

Be a better team.

Doesn't that just avoid 1-run games? It doesn't help you in 1-run games?

#29 (I see it's already covered in 30) Yeah. expect .500. Maybe move the dial a tech if the team has a really good (or very unbalanced -- a team with studs and duds in the bullpen should do well in one run games if the manager does a good job of identifying the studs. Tricky to study though) bullpen.


As you know, what makes for a good overall record is usually the record in blowouts. If there's a secret sauce for 1 run games nobody's found it yet.

(or very unbalanced -- a team with studs and duds in the bullpen should do well in one run games if the manager does a good job of identifying the studs. Tricky to study though) bullpen.

True, though it's worth noting that this doesn't improve your record in close games, it just adds runs to the backs of blowout losses (which isn't a goal). Having an unbalanced bullpen doesn't get you more extra wins, it just lowers your pythag record.