PECOTA foreses 74-88

by Eric in Madison on Feb 16, 2008 12:15 PM EST

I won't show everything, since it's subscription, but follow over the jump for some of the highlights.

Here's PECOTA's weighted mean projection:

RS: 709
RA: 779

Playing time is distributed how you would expect, with the assumption that Gomez will get 60% of the CF time and Harris will be the more-or-less regular 2B.

They anticipate Baker, Bonser and Slowey to have the most innings, with Liriano at 110 IP and the 5th spot (and Liriano's missing innings) going to a combo of Blackburn, Perkins, Humber, and Mulvey. This was done before they Hernandez signing, obviously.

They anticipate negative VORP from CF (Gomez and Span (in a little time) are both below replacement, and Pridie (in a little time) is about exactly replacement). They also anticipate negative VORP from SS (Everett, who can't hit).

It's a pretty pessimistic offensive projection, really, with Lamb barely over replacement, significant regression from Morneau, and overall less runs than last season's offensive catastrophe. I trust PECOTA, and don't think this is going to be a good run scoring team, but even I would take the over on 709 RS.

Pitching not too many surprises--Baker, Bonser, and Slowey all in the mid-4s range for ERA, Liriano better but low innings, and the rest of the starters arrayed around replacement level.

0 recs | Comment 35 comments

Story-email Email Printer Print

Comments

This is the third PECOTA thread

Other threads have indicated a distinct disdain for PECOTA projections this year. The consensus seems to be they're overly pessimistic for some players. The PECOTA methodology works for a lot of cases. But if players have no good comparison to past players, PECOTA can be way off. I think that's the case with both Mauer and Morneau this year. It projects neither to have an OPS above 800. Do you think that will happen?

"You're thinking too much. Just have fun." -- Bennie "The Jet" Rodriguez in Sandlot

by cmathewson on Feb 16, 2008 12:51 PM EST 0 recs

Sorry if this is redundant

It projects Morneau just over .800, and Mauer just under.

As I said, I think its rather pessimistic, but oddly, if I accept your reasoning, I should be more confident in Morneau's projection, since he is by no means a unique or even unusual player.

Anyway, sorry if this is rehashing. I would take the over, but I suspect they won't score a lot of runs. 725 is my intuition on where to put the over/under.

by Eric in Madison on Feb 16, 2008 1:00 PM EST 0 recs

Morneau

I actually think Morneau is a bit unique. His first full year was plagued by injuries, after which he has put up an average OPS of 876. If you throw out his injury year and average the other two, considering that he's just entering his prime, that seems a more accurate projection than PECOTA. PECOTA will not take unique or extenuating circumstances into account when it looks at a guy's career. To me, that is its main flaw.

A similar thing happened to Mauer, BTW, at a more severe level. But Mauer's problem is there just aren't any players in history quite like him. The closest ones are pretty far off of his ability.

"You're thinking too much. Just have fun." -- Bennie "The Jet" Rodriguez in Sandlot

by cmathewson on Feb 16, 2008 1:12 PM EST up 0 recs

Marcel

What I do with these projection systems is take several of the better ones and average them. Marcel projects the Twins at over 800 runs, which is the most optimistic I could find. Others are somewhere in between. I think the average is around 750, depending on which ones you include and which you don't.

"You're thinking too much. Just have fun." -- Bennie "The Jet" Rodriguez in Sandlot

by cmathewson on Feb 16, 2008 1:05 PM EST 0 recs

I think

I think that Marcel sounds nice. As we know these different projection systems come out pretty close too each other in terms of correctness in the end. So while PETCOA is very possible, it certainly represents the low end of the spectrum of possibilities for the Twins.

I think 74 is very possible though. Although I expect good competition near .500.

It's kinda one of those things. They can be fighting for .500 all year, and then lose their last 5 games and end up with 74, or they could fight at .500 all year then win their last 5 and end up with 84. Funny stuff...

Baseball is great because you cant take a knee or kill the clock. You gotta put the ball over the plate and give the other guy his damn shot E Weaver abridged

by AdamOnFirst on Feb 16, 2008 4:06 PM EST up 0 recs

I really think these are way off...

and Marcel is a better indication this year.

I think we will score around 800 runs, the best Twins' offense since at least 1996, and we will win around 85-86 games this year.

I really think last year was the absolute floor for this team, and despite losing Johan and Torii, I honestly think we're a better overall team as well.

by djskilbr on Feb 16, 2008 4:30 PM EST 0 recs

This is off...

but not by quite that much. I'd peg us at closer to 750 than 800, but that really depends on how much improvement we expect from Mauer-Morneau-Cuddyer-Kubel. PECOTA not only does not expect improvement, but expects a decline (compared to 2007) for these four. One thing I know for sure, we're not going to be a WORSE offensive team this year than we were in 2007.

One other note, this projection appears to be pre-Livan. Evan with PECOTA's projected 5.57 ERA for Livan, our pitching would improve because this projection assumes 100 IP from Blackburn (5.80), 65 IP from Humber (6.47) and 40 IP from Day (6.54). Replacing this 205 IP with Livan's 5.57 ERA would save a total of 15 runs.

Of course, there's no way I'd see Blackburn pitching 100 innings if he's at a 5.80 ERA. He'd be sent back to AAA before then.

by Adam Peterson on Feb 16, 2008 4:59 PM EST up 0 recs

Maer, Morneau, Cuddyer and Kubel

It's odd they predict all of them to get worse -- I'd say it's more likely for all of them to get better! But if even two better, two worse, would exceed PECOTA, I'd take the over, too.

by by jiminy on Feb 18, 2008 1:00 PM EST up 0 recs

better over all?

I think our offense will be a little better but our pitching will be a lot worse. Santana, Garza, and Silva were our three best pitchers. And over the past four years the Twins had a losing record in games Santana wasn't pitching. Our offense won't be enough better to compensate for losing Santana.

by by jiminy on Feb 18, 2008 1:04 PM EST up 0 recs

I disagree...

1) Garza only pitched 83 innings, so it's not like he's a "massive" loss. Silva's also not that big of a loss with an ERA of around 4.20. One of the pitchers not named Baker/Liriano should be able to at least come close to that. In fact, I would expect the new Boof to do just that.

So that leaves Johan, and while he's very good, his year last year is much easier to replace than his past years. A 3.33 ERA is a whole lot easier to replace than his past 2.60 years. I think Liriano can pitch in the mid 3's for 140-150 innings. So basically you need to replace 70-80 more innings, plus Garza's 83 innings in the high 3's. Now you can't exactly do that, but you are fairly likely to get improvements from Baker (even if a small one), and Slowey/Livan, especially over the likes of Ponson/Ortiz innings from last year. And the pen should be better too, as we had 4 key guys injured for significant stretches from the bullpen last year.

The pitching will probably be worse this year than last, but not as much as the offense will improve IMO. I really think this offense will improve by about 80 runs over last year. I don't see nearly that decline for the staff. We probably won't reach our 2007 ranking of 4th best staff in the AL, but I doubt we'll be worse than 6th.

by djskilbr on Feb 19, 2008 12:34 AM EST up 0 recs

Cherry pickin'

You can't just pick and choose among the projection systems. You can compare them and analyze the differences, but it's ridiculous to only accept the projections who agree the most with your own unscientific evalution.

PECOTA is a much, much more sophisticated tool than Marcel, basically incorporating all Marcel's features and a lot more for added precision. White Sox fans were infuriated by PECOTA's predictions last year - go see how it turned out.

I'm definitely not saying that PECOTA will be 100% right. I hope that the offense will be able to muster more than 709 runs. But expecting 2006-type seasons from both Morneau and Cuddyer is way too optimistic.

by PhoenixV on Feb 19, 2008 3:49 AM EST up 0 recs

We're not saying that

What we're saying is that it's highly unlikely that every single Twins hitter will regress this year, as PECOTA predicts, especially since all of them are on the good side of 28. I can see some regressing. But a realistic projection is that the average of all the Twins better hitters (Mauer, Morneau, Kubel, Cuddyer, and Young) will improve some this year considering that they are still developing at the major league level. ZIPS, for example, predicts such an event.

"You're thinking too much. Just have fun." -- Bennie "The Jet" Rodriguez in Sandlot

by cmathewson on Feb 19, 2008 8:37 AM EST up 0 recs

Can't agree with that

No, that's not a "realistic" projection - that's an unscientific projection. Now, I'm not saying that such a projection is necessarily useless, but PECOTA is based on complicated, thoroughly tested algorithms, and we can't just discard that because it looks wrong compared to our own gut feeling.

Cuddyer's baseball age is 29; he can hardly be considered a developing player. Young's skills are (so far) more useful in fantasy baseball than in real baseball. I agree that both Mauer, Morneau and Kubel COULD go out and have much greater years than their PECOTA weighted means, but that feeling is due to hometown bias and a very human hope for a "good story" - not due to careful research.

It's nice that ZiPS forecasts a more interesting season for the Twins, but PECOTA has historically been better than ZiPS at offensive projections.

by PhoenixV on Feb 20, 2008 1:11 PM EST up 0 recs

We'll just have to agree to disagree

I think PECOTA has some fundamental flaws, which I have outlined in other posts. It's an interesting system, especially in relationship to others. But I don't think it is inherently better than others. So I don't have your faith in it. Not from any homerish bias, but from some real doubts about the methodology.

"You're thinking too much. Just have fun." -- Bennie "The Jet" Rodriguez in Sandlot

by cmathewson on Feb 20, 2008 1:27 PM EST up 0 recs

The doofus system

The doofus system projects an 85-90 win season. This is based on an emotional analysis of doofus's favorite team, grading the moves to upgrade the team through the most optimistic means available. There is some realism involved as the Doofus system had the Twins as a 90 -92 win playoff contender before Johan was dealt.

The reality is that a lot of these games will be decided by inches. We have more depth than any team in the majors at this point for the daily grind. Do we have enough difference makers? especially in the rotation, the one area thats really in question?

by doofus04 on Feb 17, 2008 9:48 AM EST 0 recs

One run games...

...will be the deciding factor, as always, this year. In 2006 the Twins won the majority of their one run games(24-16), last year they split 50-50 (22-22). If they win most of the one run games they could possibly get 85-86+ wins.

by clutchhit20 on Feb 17, 2008 11:59 AM EST 0 recs

One run games

If our bullpen is back near the top of the league, I expect we'll be great in one run games again.

However, with a weaker pitching staff, do we expect a similar number of one run games?

by Adam Peterson on Feb 17, 2008 1:40 PM EST up 0 recs

Check that

(20-10) in 06...

by clutchhit20 on Feb 17, 2008 12:14 PM EST 0 recs

Maybe

Maybe we won a lot more one run games in 2006 because we were a lot BETTER in 2006. Did you notice how our winning percentage in one run games in 2006 and 2007 was pretty much the same as our overall winning percentage? There isn't much trick too it...

Baseball is great because you cant take a knee or kill the clock. You gotta put the ball over the plate and give the other guy his damn shot E Weaver abridged

by AdamOnFirst on Feb 17, 2008 3:59 PM EST up 0 recs

No correlation

If you look at the whole league you can see there is no direct correlation between one run game W-L record and overall record. Sure one run game W-L
record works for some (D-Backs, Orioles, Mets, Pirates, Cards, Angels, Tigers to name a few) but check out the Red Sox, Rangers, Phillies, Royals. Reds, Rays, Nationals, Yankees, Padres...a person cannot make the generalization that as goes the one run game record, so goes the team.

National League
W L
Arizona Diamondbacks 32 20
Atlanta Braves 18 25
Chicago Cubs 23 22
Cincinnati Reds 26 24
Colorado Rockies 19 19
Florida Marlins 22 24
Houston Astros 25 26
Los Angeles Dodgers 28 20
Milwaukee Brewers 22 21
New York Mets 22 15
Philadelphia Phillies 14 23
Pittsburgh Pirates 16 22
St. Louis Cardinals 16 20
San Diego Padres 23 26
San Francisco Giants 24 28
Washington Nationals 27 24

American League
W L
Baltimore Orioles 13 31
Boston Red Sox 22 28
Chicago White Sox 16 26
Cleveland Indians 29 24
Detroit Tigers 27 19
Kansas City Royals 21 22
LA Angels of Anaheim 25 19
Minnesota Twins 22 22
New York Yankees 18 21
Oakland Athletics 25 24
Seattle Mariners 27 20
Tampa Bay Devil Rays 22 21
Texas Rangers 26 18
Toronto Blue Jays 29 25

by clutchhit20 on Feb 18, 2008 1:07 PM EST up 0 recs

Exactly

Those numbers supported my point. For the most part, a team's record in one run games was pretty near the same percentage as their record in all games. Obviously there is always some statistical variance, being it a pretty small sample size especially, but the point is there usually isn't any big deal about a team's record in one run games.

So the Twins were 24-16 in one run games in 2004. So they won about 60% of those games. Well, they won about 57% of their games overall, so that's about right where it should be. No magic. Just statistical likelyhood.

For the record, I do believe that teams can have some affect on record in one run games. Bullpen, good managing, and possibly things like good base-running/speed can make a difference at those times. But these things only make a big difference one way or the other a few games either way, so it mostly just comes down too just your regular rate of winning.

Baseball is great because you cant take a knee or kill the clock. You gotta put the ball over the plate and give the other guy his damn shot E Weaver abridged

by AdamOnFirst on Feb 19, 2008 3:32 PM EST up 0 recs

Sample size

Yeah i wish i could find more years but i tried and failed. With this size though it seems only some teams follow the pattern of having the same overall record as their one run game record. Im not sure how the Red Sox could have lost so many one run games though...

by clutchhit20 on Feb 20, 2008 10:35 AM EST up 0 recs

Small ball

But perhaps it does work for the Twins because of the fact they are a "small-ball" club (i dont know if you can quite classify them that after last year though) because the 24-16 was from 2004..

by clutchhit20 on Feb 18, 2008 2:29 PM EST up 0 recs

What's Delmon Young's PECOTA projection?

I think their offense will be better than last year too. The biggest question mark to me is how fast Delmon Young develops. What did PECOTA predict for him?

by by jiminy on Feb 18, 2008 1:01 PM EST 0 recs

Young

PECOTA has his weighted mean at .294/.329/.458.

His 2007 numbers: .288/.316/.408.

Other systems project Young at:
Marcel: .298/.334/.433
Bill James: .301/.331/.446
CHONE: .290/.326/.438
ZiPS: .294/.324/.421

Interestingly, considering the discussions elsewhere re: PECOTA, the other systems are generally more pessimistic on Young in 2008, primarily on his slugging.

PECOTA pegs his Breakout 40%, Improve 65%, so a real good chance he may end up above these numbers. To contrast, Breakout/Improve rates for our big four:
Mauer: 10% / 31%
Morneau: 7% / 36%
Kubel: 33% / 66%
Cuddyer: 10% / 42%

by Adam Peterson on Feb 18, 2008 1:25 PM EST up 0 recs

very interesting

so Young is still seen by everyone as below .800 because of low OBP. Well he's still young (as they say).

The weird thing to me is projecting Morneau as only having a 36% chance to improve, when his only other full season was way better than last year. What's that all about?

Mauer I can understand being pessimistic about if your system doesn't take into account the possibility that his dropoff was due to injury. I'd say he has a way better than 31% chance of improving because he was a bit hobbled last year.

But Morneau wasn't hurt. Why is everyone so sure 2006 was a fluke, not last year? Wouldn't you expect him to be somewhere around his career average if not better, which would be better than last year?

by by jiminy on Feb 19, 2008 1:25 AM EST up 0 recs

Improvement

PECOTA's improvement rate is based on any improvement over the 3-year average, which obviously includes 2006.

So basically, an improvement over his 3-year average would be an improvement over his 2007 performance. Maintaining his 3-year average would also be an improvement over 2007.

by Adam Peterson on Feb 19, 2008 8:27 AM EST up 0 recs

PECOTA sucks ass

where did pecota put the rockies last year? i bet everything that they put them towards the bottom of their own division yet alone the national league, and i bet that pecota didnt predict that carlos pena would hit more than 5 homers last year and i bet that they didnt come close to picking him and dmirti young as comeback players of the year did they so pecota can kiss my ASS!

by MauerPower07 on Feb 18, 2008 8:00 PM EST 0 recs

Whoa partner

Family blog here.

"You're thinking too much. Just have fun." -- Bennie "The Jet" Rodriguez in Sandlot

by cmathewson on Feb 18, 2008 8:08 PM EST up 0 recs

Hindsight

And I bet you predicted all of the above correctly? ;)

by PhoenixV on Feb 19, 2008 3:44 AM EST up 0 recs

Actually...

PECOTA put the Rockies at 82-80 last year, only 7.5 games off from how they actually finished. No projection system is perfect, and it's silly to expect them to be, especially in predicting crazy breakout/comeback seasons.

Like PhoenixV mentioned earlier, some of these cries are eerily similar to what White Sox fans were saying about last year's projections (although I agree that 74-88 seems awful low).

by BeefMaster on Feb 19, 2008 11:57 AM EST up 0 recs

Late to the party

Here's my take on projecting the 2008 Twins, using 2007 as a baseline:

Position players
-------------------
We will likely get at least equal production overall to the 2007 club at the following positions: C, 1B, 2B, SS, RF

Will will likely see at least a modest upgrade at: DH, 3B, LF

Downgrade: CF (and admittedly, it could be big)

The bench should be improved, if Punto and Monroe are in proper roles.

Pitching staff
-------------------
Santana > Liriano (he's a big wild-card, obviously)
Silva <= Baker
Bonser <= Bonser
Ortiz+Ponson <= Livan

Presumably, the winner(s) of the 5th spot will be the cream of our prospect crop, and will probably perform similar to Garza/Slowey circa 2007 (choose from Slowey, Blackburn, Perkins, Humber, etc).

And finally, we should get at least equal production from the whole bullpen, if not a tick better with a healthy Crain & Reyes, a possible bounce-back from Rincon, and likely some solid arms filtering into the back-end spots (Perkins? Humber? etc.).

Overall
-------------------
That's two big downgrades, but almost every other position we should be at least as good as 2007, if not better. Not every guy will improve, obviously: some will stay roughly the same, and a few might regress or get injured. But overall, I anticipate enough guys to maintain their production or improve it to reflect overall improvement from the whole group.

We were an 80-82 Pythag team last year, and all things considered, I see no reason why we probably won't be in that neighborhood again this year. To me, PECOTA (-7) represents a pessimistic projection, and an equal swing in the opposite direction (+7) would be the optimistic line.

by spycake on Feb 22, 2008 12:53 PM EST 0 recs

A few nits...pretty darn good overall

Position-wise, I agree with your assessments, other than SS, where I think Everett will be a modest downgrade over Bartlett last year. Agree that the bench should be improved as well, but probably not by too much.

Pitching-wise, you need to be careful to include all 2007 starters. Baker made 23 starts last year, need to include that in your analysis. Including ERA in the comparison:

Santana (219 IP, 3.34) > Liriano - agreed. wild card here. Liriano may match Santana's ERA, but over far fewer innings.
Baker/Slowey (201.1 IP, 4.38) <= Baker - I expect Baker over a full year to improve on these numbers.
Silva (202 IP, 4.19) >= Slowey - I'd be very happy if Slowey repeats these numbers in 2008
Bonser (170 IP, 4.92) <= Bonser - I expect Bonser to get his ERA under 4.50 and pitch nearly 200 innings. A bit of a wild card though.
Garza/Ortiz/Ponson (175 IP, 5.09) >= Hernandez - I'm expecting Livan to be in the general 5.25-5.50 ERA range, would be happy with anything under 5.00.

So that makes one downgrade (Santana), two modest downgrades (Silva, Livan) and two modest upgrades (Baker, Bonser). If the modest downgrades cancel out the modest upgrades, then we are left with replacing about 60-70 of Santana's innings with #5 candidates (Humber, Blackburn, Perkins, etc.). Could be a good-sized difference there. Also consider that our starting pitching was awfully healthy last year (8 total starters, no real injuries, three with over 30 starts), the potential of more injuries this year is a real possibility.

Agree that we should get at least equal production from the bullpen. I think we might get a little improvement, but our bullpen ERA was 3.80 last year...

I pretty much agree with your overall assessment, maybe one or two wins below as the mean for 2008.

by Adam Peterson on Feb 22, 2008 1:46 PM EST up 0 recs

SPs

I essentially included all of the starters, I just categorized them a bit differently.

I think Baker should be able to match Silva's 2007 over a full season, and Bonser should be capable of at least repeating his 2007 (hopefully he's better). And I figured even worst-case, Hernandez should match what Ortiz/Ponson offered (hopefully he's better too). Slowey (or whoever else takes the 5th spot) hopefully matches his and Garza's combined contributions from 2007.

Then the only 2007 starter left is Santana. Big shoes to fill, but at least by my equation, we still have Liriano's 2008 contributions left over, plus whoever else might pick up the slack (Blackburn, Perkins, Humber, etc.).

It certainly hurts to lose Santana, but it also helps to not go into the season with Ortiz & Ponson!

by spycake on Feb 22, 2008 4:37 PM EST up 0 recs

Ah

I see now. I didn't include Baker's 2007. Sorry.

Slowey 2008 could be Baker 2007. Then we just need a guy or two to emerge to replicate Slowey+Garza 2007 (26 starts, 150 IP). That would be from the "pick up the slack" group...

Ah, it's confusing. Losing Santana hurts, but the returning pitchers should be better / pitch more, plus we'll be getting back Liriano.

by spycake on Feb 22, 2008 4:44 PM EST up 0 recs