They aren't just beating their Pythag-they have also scored more runs and allowed less than their peripherals would suggest (which would, according to BBPro, drop their run differential from +80 to +30 (they have a meager +13 points of OPS over the opposition). And, on top of all that, they are also ahead of their preseason projections (Pecota et al.), which is a fancy way of saying tons of guys are having career years (some of that is youth developing I'll admit). We see articles like this every year (Dbacks '07 anyone?), and the next year this seemingly "magical" ability of a team to outplay their base talent level typically vanishes.

We see articles like this every year (Dbacks '07 anyone?), and the next year this seemingly "magical" ability of a team to outplay their base talent level typically vanishes.

Except the Angels won 100 games last year with the run differential of an 87 win team. I don't know what to make of it.

have also scored more runs and allowed less than their peripherals would suggest

There is baserunning. One example is they've scored 124 times from 2B on a single, no other team even has 100. They make more baserunning outs, but the net effect looks to be positive. Detailed baserunning stats can be found on baseball reference now.

It's conventional wisdom that baserunning doesn't matter a lot, and that may be true for most teams, but the Angels are a real outlier in how often they take extra bases.

on another thread someone pointed out that the Angels are very agressive at taking the extra bases(baseball-reference backs this up) so that may help a little.

or what AROM said.

AROM, those numbers also show the Angels as having the most opportunities to take extra bases. They hit a lot of singles.

The base running could certainly account for the run improvement relative to their peripherals but that shouldn't explain the improvement over PR. Last year I thought maybe there was some sort of effect from playing in a division with three sub-.500 teams but obviously no such issue is happening this year.

They DO have the 2nd best record in the AL in one run games though at 22-10. Not surprisingly the Mariners (also +7 vs. their PR) are first.

They have beaten their pythag for five years running....

It is a curious thing and one I am not equipped to analyze. I was firmly in the luck camp over the last couple of years, but after this year I really don't know. Mike Scioscia drives me crazy with regard to lineup construction and other little things, but I'm starting to think that he is a magic garden gnome or something because this team is mind boggling. How can you explain Matt Palmer, Shane Loux, Trever Bell, Sean O'Sullivan and Anthony Ortega getting 31 starts (out of 118 or about 26% of games) and this team being 28 games over .500? Those guys are something like 14-4 (I didn't go through the box scores to see all the outcomes were) in those games.

But, their offense is just off the charts this year. In retrospect, the Juan Rivera signing is looking very smart - $4 million per year for 3 years for this: 344/506/850?

Not signing Teix seems like a good move too - Morales is putting up this line: 351/574/925...

Their starting catcher is rockin' with a 131 OPS+; Abreu is being himself; Torii Hunter is having his best year; Erick Aybar has finally turned into a two-way player; Figgins is having career year #2; and the good Izturis filled in wonderfully while Howie Kendrick was in AAA remembering what that stick in his hand was for.

I wrote this the other day on Halos Heaven in a comment - so it didn't get any play - but I am really curious what the 2009 Angels would look like if: Scot Shields were healthy, Nick Adenhart didn't die, Joe Saunders and Kelvim Escobar had better luck with their shoulders, Lackey didn't miss a month and a half, and if Santant and Jose Arredondo didn't have a minor elbow things holding them back until late August. More simply put - I'd like to combine the 2008 pitching of the Angels with the 2009 offense of the Angels and see what would happen.

116 wins? Seriously, it might have happened.

It's been neat to watch a wonderful pitching year followed by the best offensive year for this club since 2000 or maybe 1995.

I dunno, why, but the how of this year's achievement has been that the Angels have gotten the holy hell beat out of them when they lose. They have lost games this year by scores of 17-3, 13-1, 11-1 (twice), and 10-1. By contrast, when they've allowed 0, 1, or 2 runs, they've scored 10 runs only once.

I don't have the slightest idea what that means, but it has to be a big component of that real/Pythag discrepancy.

One thing I have never comprehended, is the faction that wants to attributed winning more than pyth to the manager? why?

Has anyone charted which teams, playing under the same manager for more than a few years in a row, either overachieve or underachieve their Pythag nearly every year? IIRC Torre did that with the Yankees***, and Scioscia may be having the same sort of effect.

Of course the Whys and Hows would still be up in the air, but it'd still be interesting to see such a list.

***I just checked that, and between 1996 and 2008 the Yanks outdid their Pythag in every year except 1997 and 2008, in some cases by double digits.

One thing I have never comprehended, is the faction that wants to attributed winning more than pyth to the manager? why?

I think it makes some sense. If a team is beating it's PR then one reasonable argument is that the manager is utilizing his weapons to maximum efficiency.

For example, every team is going to have a couple of relievers with ERAs in the 6.25 range, usually not for long but they'll have them. If the manager is able to ensure that these pitchers NEVER pitch in a close game, rather than rarely, he's going to be better off. Using Joe Stiff for three innings at the end of an 8-1 game might turn it into a 15-1 game but in reality has zero impact on the likelihood of winning that game. It messes up the PR but doesn't impact the actual W/L record.

Bob D.

Those games were started by: Anthony Ortega, Ervin Santana (3x - battling through injury) and Matt Palmer.

Santana has been just awful most of the season. This team will have to succeed in the postseason just like they did in 2002 - by bashing the crap out of the ball.

Using Joe Stiff for three innings at the end of an 8-1 game might turn it into a 15-1 game but in reality has zero impact on the likelihood of winning that game. It messes up the PR but doesn't impact the actual W/L record.

That would be Shane Loux in 2009 - this week especially.

I can think of a couple of other notions. The Angels could be hitting the sweet spot in runs scored frequently. Rather than looking at pythags a look at the flip side of SNWL (I called it PNWE -- pitching neutral wins expected). And then do the same on the pitching side.

If the Angels are consistently avoiding diminishing returns in their run scoring and have some extra blowouts that would go a long way to explaining their beating the pythags. Still wouldn't answer the issue of luck or skill but we'd know where to dig deeper.

Also, years ago Dave Smith argued that base stealing is at least semi-discretionary and occurs with greater frequency in high leverage situations. Pete Palmer took this argument a step further and awarded greater weight to stolen base runs than his regressions showed. The logic being since they occur with greater frequency in important situations they should have greater value in team wins and losses. Problem being that further investigation did not bear this notion out.

But that's the general case. The Angels could be winning close games (at least in part) because of baserunning.

Just throwing it out there. I'd start by looking at the distribution of runs scored and separately the distribution of runs allowed.

AROM, those numbers also show the Angels as having the most opportunities to take extra bases. They hit a lot of singles.

It's a lot more than opportunities. The average teams scores from 2nd on a single 58% of the time. If the Angels were average baserunners, they'd score 101 times. Instead they've scored 124.

Wouldn't effective bullpen use account for a lot of this?

I don't have time to pound the numbers (I'm anti-crunch), but the Angels typically have a strong bullpen, a strong bullpen wins close games, and maybe when they lose they have a habit of throwing in scrubs and getting dinged a lot more than they could be getting hit if they played at max ability.

Has anyone charted which teams, playing under the same manager for more than a few years in a row, either overachieve or underachieve their Pythag nearly every year? IIRC Torre did that with the Yankees***, and Scioscia may be having the same sort of effect.

Pyth is like a simple regression tool. If you win over half your games you will overshoot your pyth. If you lose more than half your games you will undershoot your pyth. So teams that are constantly above .500 will constantly overshoot their pyth. Teams that constantly win lots of games will constantly overshoot their pyth by bigger margins.

I don't have time to pound the numbers (I'm anti-crunch), but the Angels typically have a strong bullpen, a strong bullpen wins close games, and maybe when they lose they have a habit of throwing in scrubs and getting dinged a but more than they could be getting hit if they played at max ability.

That happened a fair amount of times with the Yankees under Torre. For a team with as good an overall record as they always had, they sure seemed to be on the short end of a disproportionate number of blowouts. Which would hurt their Pythag but otherwise be of little meaning, since those garbage time pitchers wouldn't be called upon in close games.

The Angels bullpen has been a lot better lately, but was godawful early in the year. They certainly do have an abnormal ratio of blowout losses to blowout wins. I don't quite understand why their pattern of those games is different than most teams, but it is.

I'm sure somebody has looked at this, but it might be interesting to see if teams that take the extra base (use retrosheet so you can look at all bases taken instead of just steals) perform better in close games than you'd expect.

the Angels typically have a strong bullpen

The trend (of the past five years) has survived strong bullpen years and weak bullpen years. And, in fact, the bullpens were much better back when Scioscia wasn't beating Pythag like a drum.

My guess, and it's only that, has been that the way the offense is constructed makes it awfully hard to shut the team out, or limit them to one run. That, plus the relentless emphasis on putting the ball in play, and focusing esp. hard on doing that in high-leverage situations, has contributed to their generally above-average performance in that category. But I haven't looked it all up or anything.

They have beaten their pythag for five years running....

Note that if there is a 50% chance of going over pythag, and 50% chance of going under, then the chance of going over for five straight years is 1/32. With 30 teams, it seems not improbable that at least one team at any given moment has such a streak going.

They aren't just beating pythag over the last 4 years, looks like they are at least +1 standard deviation per year. 68% of teams will be within 1 SD, so you'd expect 32% to be more than 1 SD away from the mean (both good and bad), and 16% to be on the plus side.

The odds of any team doing that 4 years in a row are more like 1 in 1500.

The odds of any team doing that 4 years in a row are more like 1 in 1500.

I am not smart enough to agree or disagree with that amazing stat. Assuming for the moment that we are witnessesing a 1 in 1500 event - does that speak more towards the Angels or the Pythag theory as a whole?

I am not smart enough to agree or disagree with that amazing stat. Assuming for the moment that we are witnessesing a 1 in 1500 event - does that speak more towards the Angels or the Pythag theory as a whole?

It really depends, if it's been done a couple of dozen times, it speaks something about Pyth, if it's truly as rare of an occurance as the odds say, then it says something about the Angels.

Note that if there is a 50% chance of going over pythag, and 50% chance of going under, then the chance of going over for five straight years is 1/32. With 30 teams, it seems not improbable that at least one team at any given moment has such a streak going.

What about when it happens 9 years in a row? What are the odds against that? Because that's what Torre's Yankees did between 1998 and 2006, with an average of 5 games over Pythag a year.

In order to account for winning/losing blowouts, we should probably look at the standard deviation of the difference in scores when using Pythag records. What we should expect is a team to have their pythag record, AND a league average SD in the difference in scores. The difference (from average) in the latter should give us an idea how to account for the former. And it'll give us a nice solid number we can compare proposed explanations against. (To see if the explanation does all/part/too much to explain the difference between actual and pythag records.)

The odds of any team doing that 4 years in a row are more like 1 in 1500.

Yes, but the odds of at least one of the 30 teams doing it in a given 4-year stretch is ~2%. So, not nearly so remote.

It's luck.

Edit: In the article he has the over achievement as +1, +2, +5, +4, +12, +7, with the SD being 6.3.

That's only 2 years over 1 SD, not 4.

Those games were started by: Anthony Ortega, Ervin Santana (3x - battling through injury) and Matt Palmer.

Santana has been just awful most of the season

If this fact has some explanatory value, we might conclude that the Angels' real record is not unusually high. They're an excellent team that does everything well. But their Pythagorean record is unusually low, because it suffers from an egregious collection of just awful games thrown by just awful pitchers. (This was a factor for the Diamondbacks a few years ago, as well.)

This has nothing to do with outscoring their own components and peripherals, of course, which is an entirely separate subthread.

It's luck.

Really? That's it? No further questions? When's he going to glean? I asked.

Really? That's it? No further questions? When's he going to glean? I asked.

Edited my comment a little. 2 season in a row over 1 SD among a 30 team sample is literally nothing statistically speaking. You'd expect at least one team to do this 46% of the time.

Why waste your time trying to explain a normal statistical phenomena. If they get up to 4 or 5 seasons, we can revisit.

But their Pythagorean record is unusually low, because it suffers from an egregious collection of just awful games thrown by just awful pitchers. (This was a factor for the Diamondbacks a few years ago, as well.)

A distinct possibility (is there a consensus about the Arizona performance)? If so that would suggest that the Pythag theory needs some tweaking (like #26 suggests) rather than taking the vanilla Runs Scored - Runs Against aproach.

It would also suggest that it is not fact not "just luck".

Why waste your time trying to explain a normal statistical phenomena. If they get up to 4 or 5 seasons, we can revisit.

Fair enough. AROM's number was the one that sparked my question. If you believe 1 in 1500 then we (well not me but someone smart in math like yourself) should find something if you turn over a few more rocks.

What about the Yankee run a few years ago (see #25 above) is that statistically significant?

For the Yankees, it's Rivera.

There's some luck, in that they never fell enough short of pythag by luck that Rivera couldn't bail them out. Note that the endpoints, 1997 and 2007, were down years for Rivera. 2007 he had his worst ERA, and in 1997 he blew 9 saves (career high).

EDIT: same could probably be said of K-Rod and the Angels. Good bullpen, especially the closer and top set up guys, is how you beat pythag.

The odds of any team doing that 4 years in a row are more like 1 in 1500.

Yes, but the odds of at least one of the 30 teams doing it in a given 4-year stretch is ~2%. So, not nearly so remote.

It's luck.

But again, what if it happens 9 years in a row? Is that also luck?

A distinct possibility (is there a consensus about the Arizona performance)? If so that would suggest that the Pythag theory needs some tweaking (like #26 suggests) rather than taking the vanilla Runs Scored - Runs Against aproach.

It would also suggest that it is not fact not "just luck".

Well, you can say the same think about the Yankees this year with Chien-ming Wang's starts. Yes, the formulas can probably be improved by capping the margin of victory/defeat.

But, it's still "just luck" b/c no team can plan to have an inordinate amount of runs allowed concentrated in five blow outs. It's a random fluctuation.

“Mike Scocia is a genius”

You didn't even try, did you?

But again, what if it happens 9 years in a row? Is that also luck?

Probably, or the formula breaks down around 100 wins. Or, the distribution of runs scored/allowed isn't precisely normal.

You also have to exclude anything in the +/-3 wins area. Given the small number of observations (games) that's just noise.

I think everybody is giving way too must scientific credence to a formula that's a rough approximation. The real world doesn't follow nice distributions. Look at the financial crisis. 1 in 1000 year events seem to happen every 50 years.

There must be some way (even if manually) to take care of #11/#16 in the data: the effect of having, say, a particularly awful mopup guy. He pitches in blowout losses, never affects the W/L, but always lowers the Pyth. winning %. I would imagine this (and the opposite effect of having a good mopup guy) could be responsible for a lot of pythagorean offsets, without being of any real interest to anyone.

But, it's still "just luck" b/c no team can plan to have an inordinate amount of runs allowed concentrated in five blow outs. It's a random fluctuation.

Right - I get that. But, as you seem to be open to, would you say it is possible the calculated Pythags for each team are possibly inaccurate based on groupings of runs and therefore a more specific formula might create a more "true pythag" for each team? In which case some teams are not overperforming their true pythag at all but possibly meeting it?

There must be some way (even if manually) to take care of #11/#16 in the data: the effect of having, say, a particularly awful mopup guy. He pitches in blowout losses, never affects the W/L, but always lowers the Pyth. winning %. I would imagine this (and the opposite effect of having a good mopup guy) could be responsible for a lot of pythagorean offsets, without being of any real interest to anyone.

Yes, cap the margin of victory/defeat at say 6 runs. If you win or lose 10-1, ignore the last 3 runs scored/allowed.

Right - I get that. But, as you seem to be open to, would you say it is possible the calculated Pythags for each team are possibly inaccurate based on groupings of runs and therefore a more specific formula might create a more "true pythag" for each team? In which case some teams are not overperforming their true pythag at all but possibly meeting it?

Oh definitely. The formula is only meant to be a rough approximation. It's a "toy" not a full-blown predictive model. I'd assume anything within 1 SD is probably "meeting" the pythag.

I'm sure you can come up with a better regression model by looking at distribution of runs scored/allowed, cappinh the margin of victory, and so on.

But again, what if it happens 9 years in a row? Is that also luck?

What snapper said in #37. The odds of out-performing Pythag for nine straight years is 1-in-512 (.5^9). With 30 major-league teams, that means you'd expect to see one such streak about once every 17 years or so. So, the fact that we've seen one such run in the last 11 years isn't all that remarkable. Plus, the formula tends to break down at the extremes, so it's probably even a little less remarkable than that in the case of a Yankees team that won 114 games one of those years and broke 100 at least a couple of other times.

This isn't to diminish the Yankees of that era or the current Angels who are great teams, of course (and who likely deserve some of the credit for "beating" Pythag at least to the extent that part of the reason for it is that Pythag breaks down at the extremes). The best analogy I've heard is that it's not surprising when somebody wins the Lotto, but it's shocking if you do.

But, it's still "just luck" b/c no team can plan to have an inordinate amount of runs allowed concentrated in five blow outs. It's a random fluctuation.

Actually they can. By employing a rigid bullpen philosophy where they never use any of thier top 3 bullpen arms unless they're leading or the game is tied, they can insure that the Chris Bootchecks and Shane Louxs get plenty of innings- virtually all of them when games are already "over." Employing such a strategy will increase the number of blowouts (and aggregate RAs) because the manager isn't doing anything to stop them. To his mind, losing 11-3 (like last night) is better than losing 6-3 and burning up your best bullpen arms. By taking the blowout- they increase their odds of winning the next game if its a close one.

For this reason, among others, if you want Pythag to be useful you would need to have an adjustment (much like a golf handicap) where at a certain point, additional runs in a game aren't really counted, or they're counted "less" than regular runs.

EDIT- Coke to Snapper

This isn't to diminish the Yankees of that era or the current Angels who are great teams, of course (and who likely deserve some of the credit for "beating" Pythag at least to the extent that part of the reason for it is that Pythag breaks down at the extremes).

I realize it's a bit of a chicken or the egg type of situation, but doesn't it seem a bit too fortunate that the teams with all the pythag luck are teams that have the best record over the respective time frame?

I'm sure you can come up with a better regression model by looking at distribution of runs scored/allowed, capping the margin of victory, and so on.

OK Cool. So assuming that a team has .5 chance of beating or missing their "true pythag" and not knowing exactly what that is (though we must be pretty close) some of the teams like the Angels, Yankees and Diamondbacks may not have been as statistically unusual as they appeared to be?

OK, so the +4/+5 seasons don't get to 1 SD. But the +12 in 2008 is 2 SD. The average of 2006-2009 is about 1 SD per year. Probably won't get the same 1 in 1500 as before but could be close.

Yes, cap the margin of victory/defeat at say 6 runs. If you win or lose 10-1, ignore the last 3 runs scored/allowed.

I like that. There's also figuring pythag based on individual game scores, and totalling them up. So if you shut someone out you get a 1.00, a zero if you get shut out, and a 9-8 game has the winner getting .55 credit, the loser .45. If you lose 15-1 it's hardly worse than losing 8-1, but your season pythag will take a much greater hit.

A guy on Statspeak who posted as SabrMatt came up with this a few years back, don't know if anyone tried it before him.

Actually they can. By employing a rigid bullpen philosophy where they never use any of thier top 3 bullpen arms unless they're leading or the game is tied, they can insure that the Chris Bootchecks and Shane Louxs get plenty of innings- virtually all of them when games are already "over." Employing such a strategy will increase the number of blowouts (and aggregate RAs) because the manager isn't doing anything to stop them. To his mind, losing 11-3 (like last night) is better than losing 6-3 and burning up your best bullpen arms. By taking the blowout- they increase their odds of winning the next game if its a close one.

Isn't that the accepted, or at least the popular explanation of the 2007 D'Backs? They had a bunch of lights out relievers and a bunch of drek that gave up a lot of runs in blowouts.

To his mind, losing 11-3 (like last night) is better than losing 6-3 and burning up your best bullpen arms. By taking the blowout- they increase their odds of winning the next game if its a close one.

This makes sense, but not wholly satisfying. Every team should be employing similar strategies, not just the Angels. Plus, the first guy out of the pen last night was Matt Palmer. He threw some gasoline on the fire but he's the same guy who has been instrumental in the middle innings of a lot of games lately, where he held the other team down and gave the offense time to do it's job.

Actually they can. By employing a rigid bullpen philosophy where they never use any of thier top 3 bullpen arms unless they're leading or the game is tied, they can insure that the Chris Bootchecks and Shane Louxs get plenty of innings- virtually all of them when games are already "over." Employing such a strategy will increase the number of blowouts (and aggregate RAs) because the manager isn't doing anything to stop them. To his mind, losing 11-3 (like last night) is better than losing 6-3 and burning up your best bullpen arms. By taking the blowout- they increase their odds of winning the next game if its a close one.

Unless of course you have a really good bullpen, and guys 4-6 are decent.

What I'm saying is nobody plans before the season starts to have 3 RP with sub-3.00 ERAs and 3 with over 6.00 ERAs, it's a fluke. Just like having Wang give up 90 runs in 4 starts.

Palmer is the number 5 guy in the pen (assuming he's still ahead of Arredondo) when he failed to shut down the Tribe in his inning (thereby letting the game really get away) he got removed, despite the fact that he easily could have finished the game. MS went with Loux so as to save Palmer for when he could be useful in terms of winning- and he sacrificed total RA to do it.

As far as every team doing it- I'm not so sure. The thing abut MS is how rigid his philosophy is. As an Angels fan, I'm sure you've scratched you head a couple times when the bottom of the pen came out in a 2 run Angel deficit. MS (seems) to place a premium of protecting the key bullpen guys in terms of workload. This allows him to lean on those guys heavily when the Angels get a lead. I would certainly think that the strategy is a good deal less effective if your key bullpen guys aren't particularly good- a problem the Angels haven't really had in some time. Even this year, the Angels have done pretty well as far as protecting leads when they get them- despite the fact tht this is the worst bullpen they've had in some time.

Doesn't change much. From the B-ref game logs, They have 73 wins and 66 pythag wins. Totalling up the pythag on game to game basis gives them 67.4.

If I cap the margin of defeat (or victory) at 6 runs, they have a +102 run differential, instead of the actual +88. So optimal use of good/bad relievers in blowouts is only explaining 1.5 wins of a 7 win difference.

But hey, it's a start.

Isn't that the accepted, or at least the popular explanation of the 2007 D'Backs? They had a bunch of lights out relievers and a bunch of drek that gave up a lot of runs in blowouts.

It's the popular explanation, but I don't think it really works. For one thing, the back of the bullpen wasn't that bad; Dustin Nippert (85 ERA+), Brandon Medders (109 ERA+) and Edgar Gonzalez (94 ERA+) were sixth, seventh and eighth in relief appearances for them. I haven't done a systematic comparison, but I have a hard time believing that most teams get a lot better performance out of the last men in the pen.

They DBacks had a good, deep bullpen. I think that's a better explanation for their Pythag record than Bob Melvin's wizardry in deploying his bad relievers.

Unless of course you have a really good bullpen, and guys 4-6 are decent.

That would seem to be a lot more "lucky" than just about anything else.

Obviously no one plans to have a lousy mop-up guy (or 2 or 3) but once you realize you've got one, the issue is do you keep that guy off the field as much as possible (protecting your total RA) or do you let him throw a bit more so as to protect the guys who are more useful. It may very well be that to exceed your pythag for a stretch- the fluke is really the delta between your good and bad bullpen arms- with the bad ones getting a good bit of work. Was that the Yankee scenario as well?

Angel record by score difference:

1 run: 22-10
2 runs 5-14
3 runs 14-5
4 runs 9-4
5 runs 7-2
6 runs or more: 16-11

I don't have a point, don't know where I'm going with this, just searching for ideas.

Good bullpen, especially the closer and top set up guys, is how you beat pythag.

Yeah but that's generally in the two win range. Leverage can only take you so far.

If indeed Scoscia is dogmatic about reserving his best bullpen arms for games in which they are tied or lead, that he doesn't care if a 7-3 deficit becomes 11-3, at the expense of using his best arms for naught, then he is making an additional assumption about his offense.

He must assume that the chances of his offense being able to score even four more runs at the point that the score becomes 7-3 are very low...or else he would find relatively high marginal value in preventing the 8th run from scoring.

Because the team is known for investing outs in one-run strategies more than many teams (is this still true, particularly with Vlad's decline?), would this be a case of Scoscia acknowledging that, for any strengths of investing in one-run strategy outs, a real weakness is the inability to come back from deficits of more than a couple of runs?

And, by the way, Bobby Abreu is having a very good year again - on base all the time, lots of SB at a high %...how many more years of this (his performance the last several years is extremely similar to 2009) before he becomes a legit HOF candidate?

22-10 in one-run games, but 5-14 in two-run games? How is that not luck?

It's the popular explanation, but I don't think it really works. For one thing, the back of the bullpen wasn't that bad; Dustin Nippert (85 ERA+), Brandon Medders (109 ERA+) and Edgar Gonzalez (94 ERA+) were sixth, seventh and eighth in relief appearances for them. I haven't done a systematic comparison, but I have a hard time believing that most teams get a lot better performance out of the last men in the pen.

At this year's SABR convention, Vince Gennaro did a presentation that noted in passing that in either the 2007 or 2008 the D-backs had the biggest gap in quality between their first four relievers and the rest of the 'pen. Now, I don't recall exactly how he determined best relievers - I think it was innings, and he may have just said "main four" and I'm mis-recalling it as best. Anyhow, here is the gap in quality for Arizona's bullpens (top four vs rest) in both 2007 and 2008:

2007
Main 4: 2.94
Rest: 5.40

2008
Main 4: 3.64
Rest: 4.66

Actually, I think he said '08, but looking at the numbers makes me wonder.

Anyhow, the gap in quality between the top four and the rest was substantial in the ARI bullpen in 2007. If someone can find another 'pen with a bigger gap between their top four and the rest, I'd be interested to see it.

Also, it wasn't just reliever quality, but how used. Melvin went out of his way to avoid using his main relievers in garbage time, reserving them for only key situations whenever possible. To that end, he even used positions players on the mound on at least two occasions.

Note: the 2007 Big Four threw 58.9% of the teams relief innings that year.

He must assume that the chances of his offense being able to score even four more runs at the point that the score becomes 7-3 are very low...or else he would find relatively high marginal value in preventing the 8th run from scoring.

I'm not sure how much his offense has to do with it vs. the general probability of winning a game you trail 7-3. For instance, the pattern has remained this year despite the fact that the offense is scoring like crazy. I think the general idea is that sacrificing your best guys in the 7-3 game costs you wins over the course of the season, no matter what your offense is capable of.

In any case, the notion that teams treat all runs allowed the same (a foundation of pythag I believe) just isn't so.

Also, it wasn't just reliever quality, but how used. Melvin went out of his way to avoid using his main relievers in garbage time, reserving them for only key situations whenever possible.

Well, I'll just say I'm not convinced. I'm not exactly sure who the Big Four are, since Arizona had five quality relievers that year, but Juan Cruz was really good, and he pitched more often in losses than wins. He pitched in games Arizona lost 11-4, 11-6, 11-5, 7-1, 7-1, 8-1, 11-3, 14-5, 7-1 and 11-1.

Doug Slaten was better than Cruz, and he pitched the ninth inning in a 7-1 win, a 13-3 win, a 13-0 loss, a 7-0 win, a 14-0 loss, and an 11-4 win.

22-10 in one-run games, but 5-14 in two-run games? How is that not luck?

When Fuentes blows it, he REALLY blows it.

Actually, the reliever thing works both ways. Using crappy guys in a game when trailing 8-2 is likely to lead to a 12-2 loss, understating the pythag. But it will do the same when you use crappy guys when leading 8-2, turning it into a 8-6 win. This has happened more than a few times with the Angels this year, or at least it seems like it.

Of course, using those guys at those times is also pretty intuitive, so you think it'd be a wash with the rest of the league.

ETA: Oops, good catch Randy. Dammit, I'm on vacation, I can't be expected to think straight.

But it will do the same when you use crappy guys when leading 8-2, turning it into a 6-2 win.

If the Angels are using pitchers that can retroactively remove runs from their own team, they have some really, really bad pitchers...

The Angels have scored about about .25 more runs per game than expected. From a regression, the runs per game in the AL this year can be estimated by

R/G = 5.96*SLG + 24.96*OBP - 6.01

It says the Angels should score 5.50 R/G but they actaully have 5.76. So over the whole season, that is about an extra 40 runs. But they are doing it because of their phenomenal hitting with runners on base (ROB). The overall AVG-OBP-SLG this year are 0.290-0.354-0.451. But with ROB, they are 0.307-0.378-0.464. So their differential in all three with ROB are .019-.024-.013.

The AL league averages for AVG-OBP-SLG are .266-.335-.430. With ROB, they are .270-.345-.430. The differences are .004-.010-.000. So the Angels ramp it up alot more with ROB than most other teams.

Then I ran a regression with SLG and OBP broken down by ROB & NONROB. The equation was


R/G = 6.94*NONROBOBP + 4.46*NONROBSLG +17.18*ROBOBP + 1.72*ROBSLG -5.98

This predicts that the Angels would score about 5.63 runs per game. Over a whole season, it means they are scoring about 21 more runs than expected. So taking their ROB hitting into account, we reduce their differential by about half. When I did OBP for both regressions, I only included walks and hits. So the OBPs used are slightly different than what I report (from ESPN)

I'm not exactly sure who the Big Four are, since Arizona had five quality relievers that year, but Juan Cruz was really good, and he pitched more often in losses than wins.

Arizona had three relievers that Melvin relied on in high-leverage situations - Valverde, Lyon, and Pena - plus Slaten, who was his LOOGY but who also pitched in many lower-leverage situations (as do most LOOGYs, actually). Excluding extra innings (where usage is often dictated by who's available), the Big Three came into games 81 times in high-leverage situations, just 34 in low-leverage. Not counting Slaten, the next five relievers (Cruz, Nippert, Medders, Gonzalez, and Peguero) came on 10 times in high-leverage situations, 96 times in low-leverage situations. Slaten was 15 high/26 low.

Looking at the Angels a year ago, you see much the same pattern. KRod, Shields, and Arredondo combined for 100 high-leverage appearances and just 27 low-leverage appearances (13 by Arredondo). Oliver (the closest thing they had to a LOOGY) was 15 high, 12 low. Speier had a lot of high-leverage appearances early before Arredondo's emergence, but then he pitched most low-leverage innings after that. O'Day and Bulger were used almost exclusively in low leverage.

-- MWE

The Angels hitters have an OPS of .804 this year and their pitchers allow a .791 OPS. But in close and late situations, the Angels hitters' OPS is .853 while their pitchers only allow a .710 OPS. So overall their OPS differential is .013. But in close and late situations it is a whopping .143.

Interesting Cy, gives me something to go on. Pitchers showing the biggest clutch difference than the hitters, so that's where I'll focus next. Is it the same pitchers suddenly turning into supermen with the game on the line? Or use ofdifferent of pitchers? We'll see.

AROM

Go to baseball reference to see how their individual pitchers do in close and late situations. don't know where else you can get it

Cy

Correction, the hitting is the much bigger deal, I forgot for a second that every team's pitchers are better in close & late, league is .713 vs .759 overall.

For the Angels, Fuentes has 132 BFP in close and late, with a 669 OPS, vs 698 overall. Jepsen, Dondo, and Oliver are pitching better in those situations (60-70 BFP) while Bulger is pitching worse.

But the hitting rising 49 OPS points in situations where the league drops by 46 is the key.

To make a long story short, I estimate that the Angels should have about a .553 winning pct instead of about .520 just taking close and late data into account. Over a whole season, that is about 5.4 wins. I will provide some details shortly

I did a study a few years ago called Does Team Clutch Matter in Baseball?. The equation for pct was

PCT = 0.49 + 1.27*OPS - 1.26*OPPOPS

Using only walks and hits in OBP, the Angels have an .804 OPS and a .787 OPS allowed. It predicts they will have a pct of a .519.

But if you break it down by close and late situations and non close and late situations it was

0.501 + 0.918*NONCLOPS + 0.345*CLOPS - 0.845*OPPNONCLOPS - 0.421*OPPCLOPS

For hitting, their OPS was .796 in NONCL and .852 in CL. For pitching, it was .799 & .707. So they get predicted to have a .553 pct. That .033 gain over 162 games is 5.41 wins.

With 685 runs scored and 605 allowed, their Pyth pct is .561, good for 91 wins in a seasons. But they actually have a .613 pct, good for about 99 wins. So the gap is projected to be about 8. But 5.4 (or about 2/3) of that is due to their close and late performance

One thing I should clarify is that I really should be comparing the Angels' actual pct of .613 to the .519 predicted by PCT = 0.49 + 1.27*OPS - 1.26*OPPOPS. That would give them about 84 wins. So the gap is about 15 wins. So the 5.4 explains about 1/3 of the gap. But how they do with runners on base will shrink that a little more. Since it was about 20 runs, it might be 2 wins (but probably less since there is an overlap between ROB and CL)

Has anyone charted which teams, playing under the same manager for more than a few years in a row, either overachieve or underachieve their Pythag nearly every year?

Atlanta Braves 1991-2005 (the run of division titles) :
+2, +4, 0, +1, +6, +2, -2, 0, +5, +5, -2, +5, +5, +1, -1

Net +31 in 15 seasons. Presented without commentary.

So far this year, Fangraphs has the Angels as +4.79 clutch for their hitters and +4.18 for their pitchers. So that is 8.92 extra wins due to clutch performance.

Gotta agree with Cyril, here--looking at those batting splits, holy ####! a team .364 BABIP in high leverage situations, which, surely, cannot be anything else than luck.....or, they put the ball in play and other teams' defenders choke? help me out here, I'm entering the Vortex of Clutch Thinking....

To make another long story short, and using formulas from my team clutch study linked above, taking RISP performance into account would add another .023 to the Angel's pct. That amounts to 3.7 additional wins. It is probably not that much, since some RISP situations happen when it is close and late and I already did a calculation that took that into account. But if about 25% of PAs are with RISP and about 15% when it is C&L;, then maybe 3.75% are both. Not sure if is that simple, but I think most of that 3.7 could be added to the 5.4 I got before and get us close to 9 wins, 60% of the differential of 15.

I'm not sure why my initial post didn't get any play but if you want to know the odds of somebody beating their pyth over a certain number of years than all you really need to know is how often a team will win 90+ games a year year in and year out because that is how you overshoot your pyth year in and year out.

If you win lots of games you will almost always overshoot your pyth, if you lose lots of games you will without a doubt undershoot your pyth. You could give me 100 random teams from throughout the 162 schedule seasons and I can lump them into three pools and be highly accurate in those pools. I can lump the 100 teams in under shoot, over shoot, and too close to call for teams hovering 2 or so games around .500.