Computer Says Cubs Favored To Win Division

by cwyers on Jan 23, 2008 7:17 PM CST

This is a bit old (from about the 15th), but I just ran into it today and figured someone might be as interested in it as I am. An early Diamond Mind simulation of the '08 season favors the Cubs. They use Diamond Mind Baseball and the CHONE projections.

Certainly take these with a grain of salt; last season's DMB projections from SG had the Cardinals winning the division by a hair. PECOTA was the only projection system that foresaw the Cardinals having problems last season.

But we'll take what we can get for now. Some interesting data points to discuss:

The division seems to fall off after the Brewers and the Reds significantly. That seems to agree with what most of us around here see when we look around the Central. The projections seem to see Cincy having a (slighly) losing record.
Our lead over the Brewers is slight - about 3-4 games. That's why our odds of winning the division are only 47% to their 24%. (The Reds have a 19% chance, sayeth the simulation.) So even though the Cubs look like strong favorites, there's still a lot of work to be done.
Last year the Cubs scored 752 and allowed 690. The simulation sees us scoring 804 and allowing 765; essentially the Cubs' returns improving the offense are expected to do no more than offset our regression on pitching/defense.
The Astros gain a whole two wins over last season and the White Sox actually get worse. And I had it on good authority from Phil Rodgers that both teams had shown real improvement!

This is a FanPost and does not necessarily reflect the views of SB Nation, Bleed Cubbie Blue, or Al Yellon, editor-in-chief. FanPost opinions are, however, valued expressions of opinion by passionate and knowledgeable baseball fans.

0 recs | Comment 35 comments

Story-email Email Printer Print

Comments

Phil Rogers has no idea what is going on

I didn't think that the article you posted made any sense. He also seems like he is making up trade rumors to just for kicks.

Live is boring until March 31st strolls by..

by Chanman25 on Jan 23, 2008 7:43 PM CST 0 recs

To be honest...

...if Rodgers wasn't a source of reliable rumors (or at least as reliable as you can get with rumors) I wouldn't devote any time to him because I wouldn't bother reading him.

There are worst columnists than him that don't get noticed around here because they're easy to ignore. Rodgers and Muskat get a lot more heat than Marriotti and such because they do some actual reporting, and thus are more difficult to ignore.

I would suggest you learn to truly interrupt all stats before using any selective stats. -- cubswin

by cwyers on Jan 23, 2008 8:04 PM CST up 0 recs

As for this article

I some what agree with that. the Cardinals are going to tank and I have no idea what to expect from the Reds. The Cubs are the favorited, but thats why you play the games!

Although I cannot see the article, where are the Yankees listed? They just broke the payroll record, again, of I believe 218 million? Just goes to show...

Live is boring until March 31st strolls by..

by Chanman25 on Jan 23, 2008 7:45 PM CST 0 recs

At about 95 wins...

...and prohibitive favorites for the AL Wildcard. Still about 2-3 wins behind the Red Sox, though. The AL East looks very tough.

I would suggest you learn to truly interrupt all stats before using any selective stats. -- cubswin

by cwyers on Jan 23, 2008 7:52 PM CST up 0 recs

woah

I'm sure glad we aren't in that division on the AL Central either. Hurray NL Central!

Live is boring until March 31st strolls by..

by Chanman25 on Jan 23, 2008 8:47 PM CST up 0 recs

I would certainly think

that the Cubs should be favored to win the division, considering it's probably the worst in baseball.

by Peoria Matt on Jan 23, 2008 8:14 PM CST 0 recs

Defense???

"Last year the Cubs scored 752 and allowed 690. The simulation sees us scoring 804 and allowing 765; essentially the Cubs' returns improving the offense are expected to do no more than offset our regression on pitching/defense"

How is our defense gonna regress? We upgraded our defense alot since say August or early September of last year. That makes no sense to me how is the defense gonna be much worse with Fukudome in RF then Floyd/Murton? How is it gonna get worse with Soto over Barrett/Kendall? How is Felix Pie over Jacque Jones gonna make the defense worse? I don't see Lee, Soriano, Ramirez or DeRosa having any major regression defensivly either.

by cubsfan25 on Jan 23, 2008 8:41 PM CST 0 recs

seems to me the regression

is probably due more to our pitching then defense. I would guess some regression is expected from at least Lilly, Hill (though I don't see it), and a few members of the bullpen.

Its actually a fairly big differance between last year and these projections. Last year we were +62, while this year projected stats put us only at +39...so somewhere we are losing 23 runs and thus probably 3 or 4 losses. I don't buy it.

by DC Cubbie on Jan 23, 2008 8:57 PM CST up 0 recs

You have to remember...

...that the Cubs finished a few wins below their Pythagorean win percentage last season; the projection sees them treading water with the same record as last season.

Also, when I say "regression of the pitching/defense":

Runs allowed is a function of pitching and defense, so that's what I call it. When figuring out teams wins and losses it doesn't really matter whether it's the pitching or the defense that's changed.
Everything has a tendency to regress to the mean over time. Aramis Ramirez, for example, put up probably the best season of his career on defense, according to zone rating. I would certainly expect some regression there. At the same time, Zone Rating was uncharacteristically down on Derrek Lee's performance at first base; once again, I would expect Lee's performance to regress back toward his career norms - in this case upwards. I don't know if the Cubs come out a net plus or minus in that sort of analysis off the top of my head, but it's a factor.
You can definately expect our pitching to regress - but once again, regression doesn't just mean downward. Most projections seem to see Zambrano regressing toward his career norms, thus showing an improvement over 2007. Lilly is expected to get worse by a bit, and Marquis is expected to get worse by a lot. We've made some solid gains on offense to counteract this, but we could be doing more in several places to improve the team further.

I would suggest you learn to truly interrupt all stats before using any selective stats. -- cubswin

by cwyers on Jan 23, 2008 9:38 PM CST up 0 recs

Net Result

Why don't you just simplify it to a plus-minus system like the professional analysts do.

by okiecubbie on Jan 23, 2008 10:20 PM CST up 0 recs

Marquis?

He had an ERA of 4.60 and a WHIP of 1.37 last year. For his career, he has a 4.56 ERA and 1.42 WHIP. For the last three years, those numbers are 4.85 and 1.40 (as a starter). Why do you expect him to regress by "a lot?" Seems to me that last year was, in fact, pretty close to the mean for Marquis.

I honestly don't see where we'll be 60 runs worse as a pitching/defense next year. We should be better defensively in CF, RF, at C, and we should balance out at 3B/1B. We should get better results from Zambrano, maybe slightly worse results from Lilly (maybe not - have to factor the change in leagues and reduced quality of hitters). I'd argue we should expect more of the same from Marquis (good and bad). The question is Rich Hill - is he better or worse than last year?

In the bullpen, Eyre should be better, Marmol should be worse, Wood should theoretically be better (at least healthier earlier if nothing else), and we don't have Ohman's ineptitude or Cotts struggling.

There may be some variation, but 62 runs is fairly substantial.

by SouthernCub on Jan 24, 2008 6:57 AM CST up 0 recs

if i remember correctly- some of the regression

should come from HR's allowed

last year was a down year for Wrigley in terms of HRs allowed and that could be a reason for the overall increase in runs on both ends

I also would like people citing the Cubs pythagorean record to take a deeper look at their Run Differential, you didnt mention this but Cwyers did and others have suggested that our record was worse than it should've been repeatedly

I've posted this many times but all runs are not created equal when the division you play in is as inferior to the rest of the league as the NL Central. In general Run Differential and Pythagorean Records are tremendous tools but in this example digging a little deeper is needed

In division, Cubs record = 45-34 , run differential = +67

out of division Cubs record = 40-43, run differential = -5

The Run differential suggests the Cubs should be one of the better teams in our division certainly, but in an era of unbalanced schedules teams playing in tougher divisions deserve more credit. Thus one run in the NL Central is not equal to one run in the AL East

This is one of the reasons people citing the DBacks Run Differential vs. the Cubs last year were overexaggerating the differences between the two teams.

The Cubs should be good within their division this year and probably should improve to be a better than average club against the rest of the league, but to consider them World Championship contenders right now seems like its pushing the envelope a bit too far

by DartmouthCubsFan on Jan 24, 2008 8:02 AM CST up 0 recs

I never talked about W/L...

I was merely commenting on runs allowed.

As for the run differential discussion - if I WERE to use that argument, I wouldn't use it to say we're better than the D-Backs (or anyone outside our division). I'd use it to say that we should have had a better record last year against the competition we faced. Given our run differential, we should have won more games last year in theory. Conversely, the D-Backs had a negative run differential. Therefore, they were lucky to have the record that they had. They won a lot more games than they "should have" given their differential.

The run differential argument was never meant to say that the Cubs should be better than the D-Backs. It simply means that the Cubs should have won more games than they did win, and that the D-Backs shouldn't have won as many games as they won. Doesn't mean that the Cubs should be better than the D-Backs. It just means that maybe the D-Backs weren't the best team in that division. I agree that comparison across divisions is faulty based on Pythagorean numbers, but it doesn't mean run differential isn't relevant.

Obviously, that's an oversimplification. Run differential and Pythagorean wins don't account for bullpen strength. A team with a strong bullpen is likely to win more close games than average because their setup men and closers are better. Therefore, a team with a good bullpen is more likely to outperform their Pythagorean, whereas a team with a bad bullpen (like the Cubs early last year) is likely to underperform compared to their Pythagorean.

by SouthernCub on Jan 24, 2008 8:16 AM CST up 0 recs

i'm not saying run differential is irrelevant

i'm simply saying runs are not created equal across divisions, which i think you agree with

Perhaps i shouldn't have commented directly below your post, but i wanted to comment on a point i remember looking at distinctly last year that HR's were down in Wrigley.

I haven't looked at the numbers since sometime last year so i don't know if that is indeed fact but i think that plays into some of the Run projections, especially since our pitchers all have FB tendencies

there's a tremendous article on how the DBacks essentially beat the Pythagorean Record last year i believe on Hardball Times

http://www.hardballtimes.com/main/article/no-mirage-in-arizona/

by DartmouthCubsFan on Jan 24, 2008 9:02 AM CST up 0 recs

Baseball Prospectus...

...models league difficulty in their adjusted standings, as the third-order win percentage, or W3 and L3. (Actually strength of schedule, which is even better for these purposes.) If you go by that, we were only an 84 win team instead of an 85 win team. I'm not going to lose all kinds of sleep over that, given that I don't see any evidence that we face a much more difficult schedule this season.

As far as that Hardball Times article - there's a hell of a leap of faith in the middle of it. He goes ahead and advances a theory and shows support that his premise is true, but does absolutely nothing to show that his premise proves his theory. There's a step missing there.

Not that it really matters - the Diamondbacks have Haren in the fold now, and the improvement in their young offense combined with that should lead to a string of ill-considered articles declaring the Pythagorean win theory disproven by this plucky band of gritty grinders or whatnot.

As far as home runs - Wrigley's park factor for home runs was 1.150, slightly above the league average. In '06 it was 1.212, so seemingly they were down somewhat, but back in '05 it was only 1.052. Because of the wind conditions it's hard to predict how Wrigley will play for home runs next season.

I would suggest you learn to truly interrupt all stats before using any selective stats. -- cubswin

by cwyers on Jan 24, 2008 9:35 AM CST up 0 recs

proves and disproves

some of my strings of thought there

i wasn't trying to disprove the Pythagorean Record, I was simply trying to show there's some flaws in it. I think the DBacks in particular are an extreme example because of how vast the difference was between their best relievers and their worst. This is an unusual example because that difference is SO vast, but one that should be recognized for those who just live and die by Run Differential

moreso than anything i thought the article was a good read because i believe too often theories are just assumed and set in stone without analysis

I also would've thought the HR differentials were more severe than that just based on the HR/FB numbers i saw on some Cubs hitters and Cubs pitchers last season, but admittedly i hadn't checked it in a while

of all the most important thing i was trying to get across was the 3rd order BP projection that shows we were essentially an 84 win team last year. That's a pretty staggering difference from some of the "true" title contenders and shows how we're essentially taking advantage of a terrible division and hoping to get lucky come playoff time and we're still fairly far away from being a "true" contender, consistent 90-95 win team

by DartmouthCubsFan on Jan 24, 2008 9:43 AM CST up 0 recs

ERA has an exceedingly poor...

...year to year correlation. FIP ERA correlates better with future performance, and Marquis' FIP ERA was 4.92. Ergo, I expect a higher ERA next season than this season. For its part, CHONE sees Marquis posting a 4.71 ERA next season.

DGU makes the correct point below - probably the best explanation for the Runs Allowed is the expectation that the Cubs' pitching health will regress toward the mean, and that we won't repeat the rare feat of having a rotation 100% healthy (if you ignore the distasteful Wade Miller Experiment.)

As far as the bullpen goes - I absolutely refuse to get into anything more than a generic estimation of bullpen efficacy. Relievers are junk bonds, and trying to project them is an utter crapshoot.

I would suggest you learn to truly interrupt all stats before using any selective stats. -- cubswin

by cwyers on Jan 24, 2008 10:00 AM CST up 0 recs

You missed the point...

I agree ERA is not the best measure. That explains why Marquis' ERA has varied so wildly over the years despite most other factors remaining fairly consistent.

But despite this season-to-season ERA variation, Marquis' ERA has hovered around 4.60 over time and a LOT of innings. Also, his WHIP has also been fairly consistent. Last year, he was right around the mean in both. That's my point: he pitched last year like he has tended to pitch throughout his career, and his ERA and WHIP were pretty similar to his career norms. That indicates to me that he's not likely to get substantially worse.

There are pitchers who simply defy statistical analysis. Marquis appears to be one of those. His results in ERA and WHIP just tend to be better (on average) than they should be.

Also, while the point about health is a valid concern, I'd be SHOCKED if those computers take injury risk into account. Therefore, I don't think that's a relevant argument in explaining why the computers predict 60 more runs allowed this year.

by SouthernCub on Jan 24, 2008 12:38 PM CST up 0 recs

Diamond Mind Baseball...

...is a very complex baseball simulator, and it certainly simulates injuries. You can read up more on it here, but I assure you that DMB is modelling injury risk to some extent.

You have to remember that this is the aggregate of 100 simulated seasons; in some seasons played out in the simulation the Cubs sustained injuries to the pitching staff, and some did not. But using DMB is more sophisticated than the Monte Carlo simulations BP uses for the PECOTA preseason playoff odds. I don't know if that significantly adds to the accuracy of the simulation at hand.

Now, without knowing all the specifics of how they did the trial, I can say that I don't think that the model fully reality (I don't know if DMB knows that Ben Sheets is more likely to have an injury than Ted Lilly, other than what past playing time would indicate). But modeling injury risk is exceedingly difficult, so I don't know how much you want to ding the simulation for that.

As for career norms - eh. I think that looking at weighted averages of three (or maybe four) seasons is a lot more instructive, especially as players age. Aging curves for pitchers are a lot less certain than for position players, but that's something to consider as well.

I would suggest you learn to truly interrupt all stats before using any selective stats. -- cubswin

by cwyers on Jan 24, 2008 12:53 PM CST up 0 recs

Which is why I added the 3-year stats...

which are ALSO very similar to (a) Marquis' career norms and (b) his 2007 results. If you look at the 4-year numbers, they're even closer (4.54 ERA, 1.40 WHIP).

As for aging, since Marquis is 29 and should be entering his prime right now, I'd argue that his numbers shouldn't be getting worse due to age yet.

As for taking injury into account in the simulations, I'll say this. Considering that injuries are, as you say, VERY difficult to predict, I'd trust a ratings system much better if it didn't try to predict them. The point of mathematical modeling is to best simulate reality. To do so, we have to make assumptions. I'm not entirely sure that trying to make assumptions to consider injury risk are any more valid than models that assume no injury risk, as I just don't think you can really predict something like injuries.

by SouthernCub on Jan 24, 2008 1:14 PM CST up 0 recs

Further...

I wouldn't call having a 4.71 ERA to be getting worse by a lot, as you suggested in one of your previous posts. The 4.71 estimate from CHONE is, in fact, well within reasonable deviation of the 4.60 he posted last year and the 4.56 he's posted for his career.

by SouthernCub on Jan 24, 2008 12:47 PM CST up 0 recs

ZiPS sees him with a 5.06 ERA...

...which is what stuck with me, I think. ZiPS was the best system for predicting pitchers in '07, edging out PECOTA.

I would suggest you learn to truly interrupt all stats before using any selective stats. -- cubswin

by cwyers on Jan 24, 2008 1:00 PM CST up 0 recs

What were Zips for the Cubs starters last year?

I can't see them being very close to the results the Cubs got last year from their SP.

by cubsfan25 on Jan 24, 2008 3:41 PM CST up 0 recs

Dugg...

...for both the Phil Rogers bash and the use of "sayeth".

Dan

Evey Hammond: Vi Veri Veniversum Vivus Vici. V: By the power of truth, I, while living, have conquered the universe.

by dtpollitt on Jan 23, 2008 9:22 PM CST 0 recs

I don't see it

how the Cubs pitching staff will allow 75 more runs then a year ago. It's pretty much the same rotation, Lilly should regress a little bit but Zambrano should be better. I don't see Hill pitching any worse then he was a year ago, and he should actually get better. I don't see Lieber, Marquis, and others being much worse then the 4.60 and 5.10 era's we got from the 4 or 5 starters a year ago. The bullpen looks great right now as well.

So I could understand giving up 20-40 runs more ispossible. But to give up 75 runs more is just way to much of a regression IMO.

by cubsfan25 on Jan 23, 2008 10:09 PM CST up 0 recs

On the brink of regression

Now part of that projection came before Lieber was signed. I think if you replace Marquis with Lieber, things look better.

Lilly's regression should come not just in him pitching worse, but also pitching less. That said, he cut down on his walks last year and if he can repeat that, maybe he can keep his performance and innings up.

Hill is a big question mark - is he an improving pitcher or is he a pitcher who fooled 'em once and has been figured out by the league?

If we stick with Pie in CF, the defense should be improved but not by that much. If Cedeno beats out Theriot for SS, that's another possible improvement.

In the pen, you can't expect Marmol to repeat that performance, but he should still be good. You have to wonder if Howry can keep it up another year, but I think the pen will work itself out to be as good as last year.

So, there's plenty of room to think the Cubs shouldn't regress as much as the computer thinks - except that the Cubs' feat of keeping their main SP off the DL is hard to repeat. And if you have to replace Zambrano/Lilly/Hill innings with Gallagher/Hart innings, maybe they step up and do alright, but maybe they struggle and you find yourself in a regression.

*** On Sabermetric Probation ***

by DGU on Jan 23, 2008 10:27 PM CST up 0 recs

Worries about the Brewers

3-4 games is better than I expected. That may be enough of a margin that we don't have the unsavory temptation of rooting for Sheets and Gagne not to stay healthy.

What's your take, cwyers, on whether or not CHONE adequately credits the Brewers pitchers for the defensive improvements in Milwaukee?

*** On Sabermetric Probation ***

by DGU on Jan 23, 2008 10:14 PM CST 0 recs

The first thing to note is that CHONE...

...itself doesn't forsee Sheets being 100% healthy; it's only a computer but it's not stupid. It sees him putting up a 4.01 ERA over 137 innings.

As far as the improvements on defense - I don't know that CHONE accounts for defense specifically enough to make that sort of adjustment. It is regressing to the mean for all of these pitchers; whether or not that's enough to account for that is up in the air - first you'd have to try and project the Brewers defense and then see how that looked.

The real wild card is - how will Ryan Braun perform as a left fielder? I don't know if there's a good answer for that.

I would suggest you learn to truly interrupt all stats before using any selective stats. -- cubswin

by cwyers on Jan 23, 2008 10:31 PM CST up 0 recs

I should clarify...

...that I don't know what playing time assumptions were made in the simulation itself, only in the raw CHONE projections. Playing time estimates are tricky things anyways.

I would suggest you learn to truly interrupt all stats before using any selective stats. -- cubswin

by cwyers on Jan 23, 2008 10:34 PM CST up 0 recs

Now that I'm home and have Excel handy...

...we can play around with this a little bit. Fair warning: this is simply playing around with the data; I wouldn't read too much into the specific numbers here.

Last season, the Brewers' team FIP ERA was 4.18; their team ERA was 4.41.

Taking the CHONE projections of all pitchers on the Brewers (as of the last CHONE update) that have pitched in the majors before, we get a FIP ERA of 4.44 and an ERA of 4.47.

So the CHONE projections seem to be modelling the Brewers as though their offense was (roughly) league average. [The Cameron signing and the chaining off of that certainly helped matters in that regard, but I don't know if they helped that much.]

This seems to bear out in the Diamond Mind sim; the Brewers gave up 776 runs in '07 but are only expected to give up 749 runs next season. That's countervaled by a regression on offense.

I would suggest you learn to truly interrupt all stats before using any selective stats. -- cubswin

by cwyers on Jan 24, 2008 12:10 AM CST up 0 recs