Pecota 2013

[jjdman]

Do they have scoring as a whole going up dramatically?

If they are predicting the overall level of offense as being basically consistent from last season then that prediction is very odd.

Looks like 4.59 avg for the AL

[jjdman]

Looks like 4.59 avg for the AL

4.75 avg for AL East

NY 805

Boston 810

TB 736

Toronto 787

O's 711

[Can_of_corn]

4.75 avg for AL East

NY 805

Boston 810

TB 736

Toronto 787

O's 711

Compared to actual numbers from 2012

Yankees

805 804

Boston

810 734

TB

736 697

Toronto

787 716

Baltimore

711 712

So the only AL East team they are predicting to score less runs is... Baltimore. I would like them to explain how they think the Yankees offense is going to score more runs then last season. I am also wondering how often a team RS goes up by 76 or even 71 runs?

[DrungoHazewood]

I am also wondering how often a team RS goes up by 76 or even 71 runs?

I would assume it happens at least several times each year. Just looking at the Orioles they improved by 70+ runs from '10-11, from 03-04, from 02-03, from 95-96, from 92-93, and from 88-89. If you make the leap that the O's are representative of MLB a 70 run jump seems to happen roughly 1/5th of the time.

[Can_of_corn]

I would assume it happens at least several times each year. Just looking at the Orioles they improved by 70+ runs from '10-11, from 03-04, from 02-03, from 95-96, from 92-93, and from 88-89. If you make the leap that the O's are representative of MLB a 70 run jump seems to happen roughly 1/5th of the time.

Thanks, it still seems very unlikely for it to happen twice in one division. That seems to be the sort of thing that happens despite predictions, not something that should be predicted.

[DrungoHazewood]

Do they have scoring as a whole going up dramatically?

If they are predicting the overall level of offense as being basically consistent from last season then that prediction is very odd.

I don't know, but... I think I've come up with a partial answer without even asking the question. Mostly it's as simple as regressing everyone to their component ERA models, i.e. FIP/xFIP/SIERA. The '12 Orioles pen out-performed their FIP by about 3/4ths of a run, SIERA by half a run. That would explain part of the regression - on average they'll say an O's reliever should have an ERA in '13 about .50-.75 higher than last year.

What that doesn't explain is why their projections are much worse than that. My first thought was career averages, and that may be part of the explanation. But maybe not all.

What it really looks like is the O's have assembled a pen mostly out of pitchers with a history of waaaay outperforming their component ERAs. I'm not entirely sure what that means. But Johnson has a career ERA more than half a run better than his FIP/xFIP. O'Day is closer to a full run. Ayala over 10+ years is a good half a run better. So the projection systems rely on pitchers to regress to their component values... but the O's have a bunch of guys who have a 1300-inning track record of significantly outperforming their FIP/xFIP/SIERA.

It's an interesting test case. Oh, and I checked for other teams that had a similar trait. The A's pen ERA had an even bigger split with their component projections. And PECOTA has a number of Oakland relievers absolutely collapsing, too. Like Jordon Norberto going from a 2.77 to a 4.78, Jerry Blevins 2.48 to 3.70, and Pat Neshek from 1.37 to 4.34. And they have the A's finishing just over .500 despite the addition of the worst team in baseball to their division.

[DrungoHazewood]

The more I ignore work and dig into this, the stranger it gets. The other projections Fangraphs lists (James, Steamer, Oliver) all are pretty consistent: they project pitchers to somewhere around their career FIP or xFIP.

PECOTA doesn't do anything like that, at least not consistently. The Orioles relievers are uniformly worse than their career SIERAs (which is BP's own component ERA estimator). Sometimes a lot worse, like Ayala and Patton are projected to be a full run worse than their career SIERAs. My guess is they are still working in historical comps, and the O's happen to have a lot of frightfully bad comps to work with.

Another observation... their weighted means spreadsheet has improve/breakout/collapse/attrition numbers for chances of big step changes in performance. And at least in some cases they don't make a lick of sense. For example, Troy Patton is projected to see his ERA nearly double, but he's listed with a 58% chance of improvement. Jim Johnson is projected for an ERA a run and a quarter higher, but a 25% chance at a breakout season, and a 47% chance of improvement. The math don't work, folks. Hmmm... unless they're defining improvement as their component ERAs getting better, reflecting underlying performance instead of ERA fogged up by defense/park/luck/sequencing/etc.

[Can_of_corn]

The more I ignore work and dig into this, the stranger it gets. The other projections Fangraphs lists (James, Steamer, Oliver) all are pretty consistent: they project pitchers to somewhere around their career FIP or xFIP.

PECOTA doesn't do anything like that, at least not consistently. The Orioles relievers are uniformly worse than their career SIERAs (which is BP's own component ERA estimator). Sometimes a lot worse, like Ayala and Patton are projected to be a full run worse than their career SIERAs. My guess is they are still working in historical comps, and the O's happen to have a lot of frightfully bad comps to work with.

Another observation... their weighted means spreadsheet has improve/breakout/collapse/attrition numbers for chances of big step changes in performance. And at least in some cases they don't make a lick of sense. For example, Troy Patton is projected to see his ERA nearly double, but he's listed with a 58% chance of improvement. Jim Johnson is projected for an ERA a run and a quarter higher, but a 25% chance at a breakout season, and a 47% chance of improvement. The math don't work, folks.

Could this just be a case of them releasing the work early? Before it was ready and had been properly reviewed?

I am also unsure what a "breakout" season for Johnson would look like. Aside from K rate I don't think there is huge room for improvement.

[DrungoHazewood]

Could this just be a case of them releasing the work early? Before it was ready and had been properly reviewed?

I am also unsure what a "breakout" season for Johnson would look like. Aside from K rate I don't think there is huge room for improvement.

Based on this glossary, I'd conclude BP has decided these breakout/improve percentages don't really work at all and are best ignored.

[DrungoHazewood]

Actually, the Improvement Rate definition gave me a bit of insight. They say a 50% improvement rate is what you'll get if you're expected to be just the same as last year. Which makes sense. It's 50/50 whether you'll be just above or just below last year's line. So Jim Johnson's 47% improvement rate suggests he'll do just a bit worse.

That tells me that they think Johnson will continue to pitch just like last year - from a "what the pitcher can control" level. What their projections are suggesting is that the Orioles pitchers have pitched like pitchers with much higher ERAs (in terms of LD% or HR% or BB and K/9, etc), so even if they continue to pitch exactly the same their high-level results should be much worse.

And thus we're back to the question of whether the O's pitchers have a skill at out-pitching their components, or if they're just a huge confluence of good (for lack of a better term) luck.

[Can_of_corn]

Actually, the Improvement Rate definition gave me a bit of insight. They say a 50% improvement rate is what you'll get if you're expected to be just the same as last year. Which makes sense. It's 50/50 whether you'll be just above or just below last year's line. So Jim Johnson's 47% improvement rate suggests he'll do just a bit worse.

That tells me that they think Johnson will continue to pitch just like last year - from a "what the pitcher can control" level. What their projections are suggesting is that the Orioles pitchers have pitched like pitchers with much higher ERAs (in terms of LD% or HR% or BB and K/9, etc), so even if they continue to pitch exactly the same their high-level results should be much worse.

And thus we're back to the question of whether the O's pitchers have a skill at out-pitching their components, or if they're just a huge confluence of good (for lack of a better term) luck.

That in turn makes me wonder if the O's specifically targeted those types of players.

[DrungoHazewood]

That in turn makes me wonder if the O's specifically targeted those types of players.

And the follow up if the answer is "yes" would be "why"? I know analytics has jumped ahead by leaps and bounds, but is it really to the point where there's a market inefficiency there? Teams will only sign guys with a good xFIP, so the O's are taking advantage by signing guys with track records of out-doing their xFIPs?

That would certainly be a change from a few years ago when it appeared the organization couldn't spell FIP.

[El Gordo]

Thank goodness the O's pitchers don't really have to pitch without 8 guys with gloves behind them. Especially when 5 of those gloves. are GG.

[Camden_yardbird]

What I would ask is "what allows a pitcher to consistently out perform their FIP/xFIP?" Is it what happens before the pitch (i.e. intelligence, composure, pitch selection) or what happens after the pitch, or has sabermetrics just proven some people are lucky at life?

And also, if anyone wants to send me in the direction of a summary of ERA component parts, i would be greatly appreciative.

[Can_of_corn]

This stuff is filler after all. Speculative and abstract. Baseball is played on the field,projections mean little but fodder for web sites and water coolers. Im not ANTI STATS just so you know, but

I think the likelihood of Nate Silver(one of the founders of PECOTA) calling an election and comparing that to calling a pennant race is just not wise. Too many things can happen. And BTW,

politicians dont usually get injured like baseball players.

Nate cheated.

You can't cheat baseball predictions.