Silent James speaks his Mind

[SilentJames]

An interesting bit of investigation:

What were the Orioles as a team rank in offense and pitching at the start of that streak, vs. the end of it.

[Flash- bd]

An interesting bit of investigation:

What were the Orioles as a team rank in offense and pitching at the start of that streak, vs. the end of it.

Really not trying to be a d*** here, but I think you're missing the point.

[SilentJames]

Really not trying to be a d*** here, but I think you're missing the point.

what am I missing?

[Flash- bd]

what am I missing?

Well, among other things: you can't selectively exclude data from a data set that is designed precisely to measure a large sample of data--and which relies on the fact that extreme data points like the 23 game stretch you are so obsessed with will be offset by other extreme points in the overall data set--and then make assertions from that modified data. It just doesn't make sense, and it defeats the initial purpose of the statistic.

The reason why pythag has been so good over, you know, 125 years of data, and why there are so few exceptions is precisely because the kind of data you are trying to exclude is quite often (if not always) a part of the data sets it's measuring. Simply put, every team has stretches like the one you are fixated on. Teams that aren't quite as good have more of them, and when they have them their more extended (like 23 games) and they give up a lot of runs and don't score that many in that time.

I shouldn't really have to tell you this, though....this kind of qualification is made so often here on the OH that I would think people would be aware of it when posting.

EDIT: Basically, your point is as valid as this: if Ellsbury, Crawford, and Ross had been healthy all year, Beckett had been his normal self and Buccholz hadn't had that horrid first month, Boston would've been a lot better. So they are really a super good team that has just played like an average one.

ie, it has worth only in the most limited, quixotic, wide-eyed and naive sense.

[Moose Milligan]

Well, among other things: you can't selectively exclude data from a data set that is designed precisely to measure a large sample of data--and which relies on the fact that extreme data points like the 23 game stretch you are so obsessed with will be offset by other extreme points in the overall data set--and then make assertions from that modified data. It just doesn't make sense, and it defeats the initial purpose of the statistic.

The reason why pythag has been so good over, you know, 125 years of data, and why there are so few exceptions is precisely because the kind of data you are trying to exclude is quite often (if not always) a part of the data sets it's measuring. Simply put, every team has stretches like the one you are fixated on. Teams that aren't quite as good have more of them, and when they have them their more extended (like 23 games) and they give up a lot of runs and don't score that many in that time.

I shouldn't really have to tell you this, though....this kind of qualification is made so often here on the OH that I would think people would be aware of it when posting.

Yeah...yeah, that pretty much puts an end to what was a one-sided debate.

[SilentJames]

His blog is down, but even a peek at his original analysis shows that he has no idea what he's talking about.

He looks at 10 blowout games and says that our 3-7 record, in games with a -28 run differential, should be the focus. But our pythag for that period is 2-8. In other words, it's pretty much in line.

He then says:

How is four games over .500 while having a negative differential "much more acceptable" in terms of the pythag? At a minimum, it's really no different. In some ways the exact opposite.

SJ needs to give us the dates of this 23 run stretch, the runs scored and runs against and the expected pythag versus the actual record.

Actually I just re-did the math, using this calculator (Bill James' equation)

http://www.crashburnalley.com/pwl.html

our pythag on blowouts, with that run diff should be 1-9

our non blowout pythag in June should have been 8-8.

I was wrong to say MUCH better, in fact even in my notes I didn't write that. I wrote : "Still beating the run diff, but all of June was weird and I am willing to concede luck there. All 10 non-blowout wins in June were either by one or two runs"

The basic point of my post really doesn't change though. That our pythag has been thrown off due to losses by an extreme number of runs in a short period of time.

That one mischaracterization, which I will concede, really doesn't alter my overall point.

This isn't about bringing down pythagoras here the theorem works, but the Orioles' data has been screwed up by a remarkably bad stretch that I do not feel is indicative of the talent level of this team going forward. Therefore pytahgoras may very well be moot for this team going forward.

If I'm right then this team will begin to take a chunk out of that number and stay in this race.

[weams]

If I'm right then this team will ...stay in this race.

I think we can all agree that these are the results we are all desiring.

[CA-ORIOLE]

Yes, but we lost a lot of games by A LOT of runs, and it has skewed our overall run differential south.

We may have been outperforming our differential before/ after that, but in the other 4/5ths of the season the Orioles are only up 6, not 10. 6 is a lot - but its not the eye-popping there is no way they will keep this up kind of number.

My contention from the get go is that the run of games in question were losses that were so bad, (again almost 30% of the year's total runs allowed were surrendered during those 23 games) threw off our numbers significantly.

This isn't a bad team getting exceptionally lucky, rather the opposite - it is a good team that went through an unbelievably bad stretch, as far as run differential goes.

So how does a lot of losses (which you admit to a net loss of 9 over 23 games I think) equate to a loss of runs (minus 72 I guess). Like LJ said, we were down 5 before your run. Whether the losses were over a singular 23 game period or not, how does the distribution and volatility of high/low run games correlate on a pro-rata basis over an entire season (or this point in the season) with other teams?

1. All good teams don't win 79% of one run games. Why have we?

2. If it's the BP, then how do exceptional BP's relate to winning close games?

Your hypothesis seems to be that that unusual run volatility and BP performance are contributing to us outperforming our Pythag and that we have been a much better team than Pythag suggests. Maybe you have something. You just haven't shown anything imo. Why not provide some actual research?

[SilentJames]

But my point is this - they haven't been offset. The Orioles' offense has not been good enough to offset the -72, yet.

Going forward - if I'm right we should see the Orioles start to balance that out.

Once again, I'm not saying pytahgorean isn't good - I am not denying the science of it. But I am saying that 6-17 stretch was so bad that it has poisoned the well, in a way.

I mean really - do you really look at this team and agree that they should be a 46 win team? What is more likely: a team is beating their pythag by THAT much - or something weird has happened in the data set so far that has skewed the numbers in a negative direction?

[Lucky Jim]

Actually I just re-did the math, using this calculator (Bill James' equation)

http://www.crashburnalley.com/pwl.html

our pythag on blowouts, with that run diff should be 1-9

our non blowout pythag in June should have been 8-8.

I was wrong to say MUCH better, in fact even in my notes I didn't write that. I wrote : "Still beating the run diff, but all of June was weird and I am willing to concede luck there. All 10 non-blowout wins in June were either by one or two runs"

The basic point of my post really doesn't change though. That our pythag has been thrown off due to losses by an extreme number of runs in a short period of time.

That one mischaracterization, which I will concede, really doesn't alter my overall point.

This isn't about bringing down pythagoras here the theorem works, but the Orioles' data has been screwed up by a remarkably bad stretch that I do not feel is indicative of the talent level of this team going forward. Therefore pytahgoras may very well be moot for this team going forward.

If I'm right then this team will begin to take a chunk out of that number and stay in this race.

I'm using this site: http://www.had2know.com/sports/pythagorean-expectation-win-percentage-baseball.html

I get two wins. When I use the 32/60 split over those 10 games through your site I get two wins.

For the 6-17 stretch where we were outscored by 72 runs, the expected W-L was 5-18 according to my site's calculation and according to your site's calculation.

In other words, that terrible stretch of baseball wasn't deeply reflected in deviation from the pythag because it largely mirrored the pythag.

You're not bringing down the pythag because your analysis makes no real sense.

[TonySoprano]

Better to remain silent and be thought a fool than to blog and remove all doubt.

[CA-ORIOLE]

But my point is this - they haven't been offset. The Orioles' offense has not been good enough to offset the -72, yet.

Going forward - if I'm right we should see the Orioles start to balance that out.

Once again, I'm not saying pytahgorean isn't good - I am not denying the science of it. But I am saying that 6-17 stretch was so bad that it has poisoned the well, in a way.

I mean really - do you really look at this team and agree that they should be a 46 win team? What is more likely: a team is beating their pythag by THAT much - or something weird has happened in the data set so far that has skewed the numbers in a negative direction?

I think we have guys that can perform better than what they have, a lot better. In that sense I think we're a better team than Pythag suggests.

In the sense of our past actual performance, I think we have been pretty lucky to be where we are right now and should probably be a lot closer to our Pythag than our actual win total.

[SrMeowMeow]

This thread is just disappointing.

Just believing something doesn't make it true. And there has been not one iota of actual work done to demonstrate any fraction of a suggestion that the O's have over-performed their Pythagorean in some sustainable way.

[CA-ORIOLE]

I'm using this site: http://www.had2know.com/sports/pythagorean-expectation-win-percentage-baseball.html

I get two wins. When I use the 32/60 split over those 10 games through your site I get two wins.

For the 6-17 stretch where we were outscored by 72 runs, the expected W-L was 5-18 according to my site's calculation and according to your site's calculation.

In other words, that terrible stretch of baseball wasn't deeply reflected in deviation from the pythag because it largely mirrored the pythag.

You're not bringing down the pythag because your analysis makes no real sense.

Pretty much the bottom line I would say.

[Hank Scorpio]

To be honest with you, at this point I don't really give much of an "S" about why they're winning or whether they'll make the playoffs, etc...

It's just nice to be having a conversation about it for once in August. I've been chanting "talk to me in August" since early June and they've managed to hold up pretty well thus far.

Plus, I'm not overly convinced that the rest of the division is *that* much better right now.