Marcel projections are out.

[dadofandrew]

I was thinking Gross/Rottino would replace Prince's bat, as Hart moved to 1B, as i have Jenkins/Clark platooning in my mind. While both good role players, I see a dropoff of 100 OPS points or more.

[danzig6767]

Pretending there is some huge disparity between JJ and Craig is just incorrect and emotional. If you think the Crew will win 88 games with JJ as the SS, the math puts them at about 86.8 with Craig. That is not much falloff at all, and oodles better than most.

 

But aren't you just pretending to know what kind of year Hardy will have if healthy?I think it's fair to assume that Cousell will put up fairly crappy numbers if forced into significn't playing time,he's old with no upside potential and lots of history to use as a basis.With Hardy though,it wouldn't shock me if he had a poor offensive year or showed alot of his 2005 Post All-Star break form.

 

Either way,if the Brewers are in any sort of contention next year and Hardy gets a significant injury,i can't see how the best thing to help the team win wouldn't be moving Hall back to shortstop.Counsell hits like a backup catcher,anything that puts him in the lineup outside of spot duty is more trouble for an already questionable offense.

[Snuffzilla]

Only Stat I care about is Wins & Losses. & We need more 'W's...

[Team Canada]

The difference between 82 and 84 wins might not be a big deal

 

Although last year that'd be the difference between winning the division and going home.

[rluzinski]

Considering Pujols is something like 6 wins above average (I don't remember exactly), there's no reason to act like a 2 win difference is insignificant.

[Tangotiger]

I replied to the various comments in this thread on my blog:

 

www.insidethebook.com/ee/...ts_2007/#6

[GenoSeligPrieb]

"I replied to the various comments in this thread on my blog:"

 

Wow. In post # 1, screen view traffic is already being diverted away from BF.net to a blog. That's some chutzpah, newbie!

[sbrylski]

Wow, inventor of FIP on our board! Welcome, and thanks for answering our questions.

[whizkid]

Was that meant to be in blue Geno?

[RedbirdRay]

Boy, don't I feel like a jerk.

[Ennder]

Just keep in mind that this is the weakest of all projection systems and thats because its not really meant to be a strong projection system, its a very basic baseline type system.

 

As for the pitching results on Chone's site. Until he looks at a stat other than ERA his correlations are useless since ERA has such a huge variance to begin with. He at least needs to compare some sort of component ERA instead to judge the systems.

[SoCalBrewfan]

Geno, please be nice. Perhaps you don't recognize tangotiger's handle, but he's the guy who developed the Marcel system and calculated the numbers above, as well as something of a celebrity in the stathead world (for whatever that's worth). He's responding to some of the criticism above, which he perceives as unfair and/or ignorant. Given that the criticism is in response to something he posted on his blog, and has made available as a service to the community (thanks, tango, by the way!), I don't think it's unreasonable to respond there by any means.

[1992casey]

Welcome aboard, Tango!

[rluzinski]

Yeh, considering this thread is about his work and already had multiple links to his site as it is, I don't exactly see the conflict. I assure you, he doesn't need our message board to get hits. http://forum.brewerfan.net/images/smilies/smile.gif

 

EDIT: At the link showing the Marcel methodology (I linked to it in the second post), he says:

 

Quote:

---

Note: I do not stand behind these forecasts. These forecasts are the minimum level of competence that you should expect from any forecaster. Do not attach my name to these forecasts in any kind of evaluation experiment. They should only be referred to as Marcel The Monkey Forecasting System, or simply The Marcels.

---

 

That might give people a better idea of what he is trying to accomplish in the first place.

[dadofandrew]

I do projections at Ramblings and they usually are similar to Pecota, Shandler, and others. For the most part, it's just math, with some adjustments for potential breakout seasons.

 

The true measure is how accurately they are able to guess big improvements and declines, and for the most part, none do it very well.

[Ennder]

Quote:

---

do projections at Ramblings and they usually are similar to Pecota, Shandler, and others. For the most part, it's just math, with some adjustments for potential breakout seasons.

 

The true measure is how accurately they are able to guess big improvements and declines, and for the most part, none do it very well.

---

 

I agree in general though I think part of the problem is the sample size of players they look at. Projecting a player like say Pujols is a lot easier and shouldn't be included in the comparison in my opinion. Basically any player who has been full time and on the same team for the past 3 years is a bad data point when comparing systems and since thats a large percentage of players it clouds comparison of systems greatly.

 

The best systems do a good job of showing older players declines, how new teams will affect established players and predicting breakouts for young players. The other 50% of projections aren't really important since every system is capable of projecting established players at about the same rate.

[Tangotiger]

I don't necessarily disagree. At the same time, it shows how little potential for gain there is.

[rluzinski]

Here's a link to a nice Hardball Times article about the statistical regression and how it pertains to Marcel projections:

 

But I Regress?

 

It's written in layman's terms, so it should shed some light on what Marcels is trying to do.

[Brew Man Dan]

Does anyone know how to sort the marcel projections in excel? Also, how reliable are these? Would it be wise to use them for fantasy picks?

[rluzinski]

What kind of sorting do you want it to do? Excel can sort them by whatever is there.

 

As for using them for fantasy picks, they are very nice for established vets but for young guys without much of a track record, not so much. The reliability column will reflect that. The higher that number is, the more likely the players stats are reflective of his actual ability. kepp in mind that park factors aren't considered, so you will have to adjust accordingly.

[Tangotiger]

Right, the question of "how reliable are these" is answered in the "reliability" column! Show me another forecast that does that.

 

And, as mentioned, if a guy has switched in or out of an extreme park (like Coors), just be careful.

[Ennder]

Quote:

---

Right, the question of "how reliable are these" is answered in the "reliability" column! Show me another forecast that does that.

---

 

Shandler does it as well. No clue how reliable his reliability ratings are though.