Part of CSNPhilly.com


« Collier, Abernathy ink minor deals, receive invites | Main | Report: Zito, Giants agree on monster 7-year deal »

Wednesday, December 27, 2006

Comments

Take a look at the projections for the Mets. Lets hope zips is right.

The starting rotation look pretty deep but the Phils need at least one of their starters to win 16-18 games next year. Is that Myers, Garcia, or Hamels?

good thing we have those projections.....otherwise me might actually have to PLAY the season out!

MG- to answer your question- all 3 will win 16-18 games. Hello playoffs!

I'm glad to see someone has exceeded me their Phillies optimism - my only question is, who do you envision handling the 8th, carson, and how many saves would that leave Mr. Gordon with - 50+?

me in their

I've never seen this site before. I'm a little mystified by the projections if Ruiz is listed at 111 games and Barajas 87. Seems to me a more sophisticated program would have a limiting factor that would gauge playing time in the context of a 162-game season, rather than what seems to be the case here -- merely projecting from what last season's playing time was.

I agree with Alby - it's like he got a good start and then just dropped the ball.

As for George S.'s question from this morning: "With the talk of PG dropping the ball on getting bulpen help, I'm curious which relievers people think he should have signed and how much he should have paid for them ($$ and years). Who exactly did he let get away?... Where should PG be focusing his efforts to fill the bullpen holes? Which relievers are the answer to our problem?"

I think David Weathers, rhp (2 yrs., $5M), Guillermo Mota, rhp (2 yrs., $5M), Steve Kline, lhp (2 yrs., $3.5M), Tanyon Sturtze, rhp (1 yr., $750,000), and even Jamie Walker, lhp (3 yrs., $12M) were good deals that the Phillies could have matched with their budget, if they can get Schoeneweis for Walker money it would be a good deal (which is why he will probably cost more).

This is the first time I've seen this site as well. The word that comes to mind is "absurd."

I hate to disagree, V, but I don't see a single name on that list that has ever fulfilled the role of 8th-inning guy, which I believe is the primary unfilled job in the Phillies bullpen. Mota would come closest, and he hasn't had a year like that for a while. My guess -- and I emphasize, it's only a guess -- is that Gillick is trying to find someone like that via trade.

I think he is as well, unless he is planning to put Lieber, Eaton, or Moyer in there. But, I think these are all reasonably priced relievers compared to the rest of the market - granted, the Phils might not have been able to get any of them for the same exact money. Kline, Walker, or SS could balance a pen sorely in need of a good lefty and Weathers or Mota both should be good setup guys which would make me much more comfortable to have Madson and Geary manning the 6th/7th.

I wouldn't have touched Mota. He is suspended for the first 50 games of the season and he is a steroid user. And I've mentioned I like the local lefty Schoeneweis, but I would never give a 3 year deal to a middle reliever. Hey Arthur Rhodes is still unsigned.

hey i posted a week weeks ago that Lieber would make a great reliever. From what we've seen of him, we all know that once the 4th / 5th innings roll around the guy begins to lose it and normally pretty fast. That being said, i've always watched him and said that he could make for a good back of the pen guy. But given his salary and probably unwillingness to do it, it will never happen.

Good point on Mota - it would have been like siging Dotel last year. His lower-than-market price is the result of that. As for the smart-aleck point at the end Everyday Eddie Guardado and Troy Percival are former closers who are still available as well.... I just hope they don't bring back the one guy from last year that they might - Rick White.

I don't doubt that PG would be willing to reach out to see if Lieber or any of the others is willing to move to the pen - that is probably a conversation that would not even happen until mid-Feb if we still have all six, but since they couldn't get Torres or Gonzalez out of Pittsburgh for Lieber, you have to really decide if the quality of the arm you get in return is really worth giving up Lieber. Heck, the Cubs are going to start using Wood in the pen....

I'm with Benjah stat projections are for smucks

I've traded emails with the creator of zips in the past, and he will tell you himself, there is a lot of wiggle room in the stats. They are only meant to give one a rough idea, I doubt even the creator would place bets based on the system.

Just to change things up...my take on some of the phils top prospects....

1. RHSP Carlos Carrasco (if he's really 19 then he's at least 3 years away from CBP)

2. RHSP Scott Mathieson (profiles as he Phillies are notorious for falling in love with toolsy outfielders come draft day (so muceither a future starter or relief ace)

3. RHSP Kyle Drabek (attitude/makeup are question marks)

4. LHSP J.A. Happ (will be on this team at some point this year)

5. OF Michael Bourn (kenny lofton clone but without the pop)

6. 3B Michael Costanzo (he's a good looking kid and local so the phils will give him every opportunity to prove he CAN'T be the next everyday 3rd baseman in CBP)

7. LHSP Josh Outman (love the skills and name)

8. RHSP Edgar Garcia (honestly haven't seen him pitch)

9. OF Greg Golson (too many raw skills to say he's a total dissapointment but this year he has to hit .275 with 15-20 hrs and most importantly lower his K rate)

10. INF Adrian Cardenas (will be a major leaguer in the next 3 years)

11. LHRP Fabio Castro (will somehow play a part on this team maybe as a reliever after some experience in AAA).

12. RHRP Zack Segovia (been my darkhorse all offseason as a guy who will make this team out of Spring Training much like Carlos Silva in '02).

13. OF D'Arby Myers (could be a stud from everything i've heard)

14. LHSP Matt Maloney (the guy was a gamer in the playoffs for Lakewood last yr)

15. RHSP Justin Germano (YOUNG journeyman who could play a role this year much like Condrey / Sanches did in '06)

16. C Jason Jaramillo (needs to take his AFL success into this season or else his career could be as a backup)

17. RHRP Joe Bisenius - (throws hard and slower to develop thus far)

18. LHSP Daniel Brauer (???)

19. RHSP Kyle Kendrick (another young gun for the future)

20. OF Jeremy Slayden (has some ability and wouldn't be surprised if becomes a platoon guy in the future)

21. RHSP Jarrod Freeman (too young and not enough experience yet to comment on).

22. RHSP Drew Carpenter (see #20)

23. SS C.J. Henry (will always be considered a disappointment just based upon trade for Abreu was ludicrious on paper)

24. C Lou Marson (has power and can catch but can't hit the ball enough to be a regular catcher in the bigs)

25. C Jesus Sanchez (young but needs to hit alittle more)

26. 3B Welinson Baez (once was a top 5 prospect now nothing can be expected of this bust).

27. OF T.J. Warren (should rise fast considering his age and experience in college)

28. RHRP Pat Overholt (see #20)

29. INF Jason Donald (projects as utility player at best)

30. INF Brad Harman (hit well in '05 and stunk in '06)

31. RHRP Brett Harker (another arm and can't have too many of them.

32. OF Dominic Brown (6'5 200lbs originally signed with Miami to play football)

The ones who got away include 15th rounder Riley Cooper (now playing WR for the soon-to-be National Champion Loser Florida Gators...it pains me, a loyal PSU fan to say it, but it's inevitable), 31st rounder Bruce Billings, and perhaps the most painful of them all, 34th rounder Josh Thrailkill.

Names worth storing away from the Phillies championship VSL squad: IF Redne Fuenmayor , C Francisco Murillo (.299/.402/.502 in 221 at bats), and pitchers Moises Melendez and Mauricio Romero. The VSL is as far as the majors as can be, but it doesn't hurt to just remember a couple extra names in the back of your mind...especially when the farm system is as bad as the Phillies is.

Just finished perusing the ZIPS projections - what a load of crap!

Most of the posters on this site would have just as good of a success rate if they sat down with nothing more than a pencil & paper.

Do you actually have evidence of your own (or anyone else on the site's) projecting skills VoR, because the creator of ZIPS does? What is the standard deviation of your projection methodology last year? Can you send me your spreadsheet containing this year's projections?

Oh...that's right, you didn't do one.

I agree that the projections are a shaky game becuase there are too many variables, but this is a hell of a lot better than you or I could come up with.

And why, exactly, is it a load of crap?

I propose a VoR vs. ZIPS contest for this year, you can just send in your projections to the site and we can compare them at the end of the year. OK?

Keith Foulke could possibly be an 8th inning man with the ability to step into the closer role should Gordon go down again. He is close to signing with Cleveland though, so Gillick would have to get things moving if he had any interest in him.

http://mlb.mlb.com/NASApp/mlb/news/article.jsp?ymd=20061226&content_id=1767846&vkey=hotstove2006&fext=.jsp

I looked at the Zips and read their comments. I was in a graduate level statistics course in college and our Professor, a Harvard Ph.D., asked what is the best predictor of the future? The class gave a bunch of intelligent sounding answers, things like running regression analysis, and other high tech language solutions. After the class was exhausted, the Professor said that the best predictor of the future is the last real data. If you hit 58 homers last year, the best prediction is 58 homers for next year. To be realistic, we need to believe these players will do roughly what they did last year. To me the Zips predicted mediocrity.

Do you actually have evidence of your own (or anyone else on the site's) projecting skills VoR, because the creator of ZIPS does?
I’ve seen the evidence for the creator of ZIPS for the 2006 Phillies and the results are mixed. While some are impressive, there are as many, if not more, that are way off the mark.

Can you send me your spreadsheet containing this year's projections?Oh...that's right, you didn't do one.
Excellent prediction, kdon, almost uncanny!

I agree that the projections are a shaky game becuase there are too many variables, but this is a hell of a lot better than you or I could come up with.
And what exactly do you consider “a hell of a lot better”?

And why, exactly, is it a load of crap?
For all of the above reasons: mixed results and too many variables for consistently accurate projections.

"After the class was exhausted, the Professor said that the best predictor of the future is the last real data."

If this is true, then from a statistical standpont it seems the prediction for Howard and Victorino are really poor because there's too little data for the prediction to be reliable.

In the case of Moyer, there's ample data on which to base a prediction. Hamels? It would seem to be almost a crapshoot.

Why would you want someone who gives up a ton of hits and homeruns in the bullpen? Why not just re-sign Ryan Franklin?

ZIPS and VORP aside, there's practically no way Lieber is as bad as he was last year. I think he's going to be good for 12-15 wins on a decent team, with an ERA in the mid-fours. The Phillies ought to be able to score a solid reliever for him.

AWH, you don't know what you you are writing about. You are just like the students in that class as you write "from a statistical standpont it seems the prediction for ... are really poor because there's too little data for the prediction to be reliable." You think you sound smart, but you miss the point. This has nothing to do with sample size. If you saw the sun rise today at 6:10
AM for the first time in your life, 6:10 AM would be a good prediction for tomorrow, despite your sample size of one observation.

This site is a learning site, not a put down site.

Actually, Fred, the chances of the sun rising at 6:10 a.m. tomorrow are almost nil. It will either rise a minute or two earlier or a minute or two later, depending on what season we're in. So the professor was right as far as he went; what you failed to appreciate was that the "last available data" extend beyond just the time of the sunrise.

As with the sunrise, so with projecting stats for players next year. It helps to know if they're on the upside or downside of their careers. My questioning of the ZIPS data was based on a projected 198 games for just two of next year's probably catchers. Lo and behold, the site acknowledges that it is based not on predictions of actual playing time -- it even gives projections for players who will never sniff the major leagues. I have to wonder how useful such data is. I don't know much about this site, but it looks less sophisticated than computer program Baseball Prospectus runs.

In fact, Fred, if you think about it you'll realize the glaring error in your Howard example -- I'm willing to bet you any amount that he hits some number of home runs other than 58 in 2007. Why? Because virtually no player hits the same number of home runs two years in a row, let alone his first two full seasons. That, too, is a bit of data that should be taken into account in predicting next year's performance.

Alby, that's not the point. I'm not saying that the predictive number will be exactly like its predecessor, but it will be close. The sun may rise at 6:08 AM tomorrow, not exact, but close. The sun surely won't rise at 3:10 AM.

The Howard example of 58 homers is probably not a good example, because only a tiny percentage of ball players have ever hit 58 homers in a season. Next year Howard could hit 45 homers, or he could hit 65. I would think that Burrell's numbers will be similar to his 2006 numbers, roughly 30 homers and 95 RBIs. He may do better, he may do worse, but he should be similar to what he does.

I found it odd that the Zip starting pitchers all winning 12 or 13 games. I tend to agree with Voice of Reason that the zip stats were a load of crap, albeit well intentioned and mathematically modeled, but still crap.

People are taking to heart the ZIPS numbers way too seriously. This is just a guy running a series of regressions and outputting some answers. More of a casual glance type thing than a thing of serious statistical discussion.

As for regression or any advanced stat analysis, "garbage in, garbage out" generally is a pretty good rule of thumb to follow.

Also, you have to be careful about you massage the data. Can always get a regression analysis that explains a higher percentage of the variation but a simpler regression formula is almost always more useful for actually interpreting the data.

Fred, If you felt put down by what I wrote you misinterpreted it. I meant to be supportive of what you had written, particularly the quote from the professor, and that the ZIP projections are "a bunch of crap. I was not focusing on anything else.

MG and Alby are correct.

Take the stock market as an example. Multiple prognosticators try every year to predict what the market will do based on various data, valuation and statistical models. People who are smarter than I am. They are almost never right because of the unpredictable variables that come into play.

Interest rates are also an example. Every year the Wall Street Journal asks fourty leading economists to predict where interest rates will be at the end of the year. Not only do they not get the level right, the majority of them don't even get the DIRECTION right. Too many variables and unknowns for even these PhDs and their staffs to handle.

Baseball is no different. Take Utley and Howard. Suppose in a fit of passion the Mets overpay for Barry Zito and trade for another pretty good left hander before the start of the season. The Phils lineup would have to face them plus Galvine multiple times next year. How will that affect their numbers?

These "predictive" models are of limited usefulness right now, as we don't even know what the final lineups will look like.

Are injuries factored into these models, even the nagging kind?

Projections are fine and dandy to look at and take as a ballpark figure as to what the player will actually do. Then again you could look at his career trend/stats and make as nearly as solid prediction. Either way it's a bunch of baloney and doesn't really amount to a hill of beans...just play the season out and stop trying to win games with a strat-o-matic.

AWH, everything is cool. I am normally a member of the anti-combative wing of the Beerleaguer posters.

Carson, I was ready to write a post using the word strat-o-matic and you beat me to it. I would rather be playing with strat-o-matic numbers than these zip predictions. At least strat-o-matic used last year's stats. I don't know how strat-o-matic works today, or even if it still exists, but I do remember wasting (but really having fun) many hours rolling those dice and looking at those player cards during my first year of college. (Probably explains my poor calculus and physics grades that year.)

We had a ten team league with a player pool from mixed eras. We had some current teams mixed with the 1950 Whiz Kids Phillies, a Phila. A's team from around 1930, and another old team I forget. We mixed up the teams' players and held a draft. I had Del Ennis (Whiz Kid) in the outfield, George Earnshaw from the 1930's as one of my starters, and Jack Akers (he had a low ERA) from the 1969 Yankees (I think) as my closer. The good thing is your strat-o-matic players never got tired. You could start a short inning reliever and pitch him all day. He never got tired and the batters never figured him out.

Carson - I was intriqued by the Burrell, Lieber, and Bourn trade rumor which generated on your site. I am very skeptical but since it is a slow day I thought about who could possibly be involved.

Pretty Boy Burrell has publicly agreed to be traded to only the BoSox, Yanks, and Arizona (thought there was another like SF on Left Coast but cannot remember for sure).

With Randy Johnson on the trading block the YANKS would appear to be the key to anything of this magnitude to happening with the shuffling of their current roster. Perhaps we could get MATSUI and some relief help or a strong minor leaguer with some luck.

ALSO: Does everyone realize the EXTREME MAKEOVER the Phillies have gone through in the last year? Burrell and Jimmy Rollins are the longest tenured Phillies with 6 full years of experience. The next........Brett Myers who is 26 yrs old & has 4 Phillies years of experience. Following by Utley, Geary, and Madson with 3 years.

The ZIPS stuff is interesting but it's completely ridiculous.

Ryan Howard will have 90 walks by the All Star break. And if he doesn't have a OBP over 400 I'll be completely surprised.

Does ZIPS take into account that Lieber will show up 20 pounds overweight this spring?


A few things...
1) ZIPS isn't just some guy messing around. It is a sophistocated computer program that has a track record of success. Last season, it beat Baseball Prospectus and every other major prediction program at projecting pitching, and I think it did okay at hitting too. This is especially encouraging considering, as it has a more rosy view of Phillies pitchers than hitters.

2) With the exceptions of the Howard and Myers projections which come with 15th, Mean, and 85th percentile projections, the other players are only mean projections. They project the average result based on previous results. It's not that ZIPS projects no Phillies pitchers getting more than 13 wins-- it's that it projects each individual pitcher to have around 12-13 wins. Probably someone will be unlucky and get 10 and someone will be lucky at get 16. The Myers projection is a good example.

3) The best predictor of the future is not just the past year. There are tons of studies that explain year to year variations. Constant themes emerge such as regression to the mean (Howard hitting a slightly less beastly 51 homeruns), Hitters with very high batting averages on balls in play compared to their career levels lowering their batting average (Helms), and pitchers with injury histories getting injured even if old injuries are healed (141 IP for Eaton).

4) The point about Barajas playing 87 games and Ruiz playing 114 games is a good point, but the system projects individual players based on previous years played by similar players. It's done by a computer, so it basically explains what similar players have done with similar histories. It eliminates some bias while sacrificing some accuracy. The idea is to see how similar players have performed based on similar historical performances.

5) It does take into account injuries (as someone asked). Rowand is listed as playing 116 games. That means he'll miss an average of 46 games. Probably that means that about 1/2 the time he'll miss a lot more and 1/2 the time he'll miss a lot less.

Fantastic post, Matt.

Dittos, Matt, Good Post! One question, on the pitching wins and losses, does it take into account the team that the pitchers are pitching for? A pitcher pitching for the Phillies, a team that scores a lot of runs, will win more games than the same pitcher pitching for a team with anemic bats.

When considering the ZIPs numbers, and all other statistical analysis of any kind, just remember one universal truth: 67.5% of all statistics are BS.

Does the ballpark impact these statistics (i.e. do the Barajas/Helms numbers reflect the fact that they're now playing in a band box)?

Obviously, these are just projections and should be taken with a grain of salt. Hell, 3 or 4 of the Phillies players (along with many others from other teams) could be indicted on 'roids charges, a factor that I'm sure Zips didn't take into account. Global warming could mean more HR's this year. Let's keep this all in perspective. If years and years of scouting has taught us nothing else, it is that these predictive models have some merit, but cannot be used for anything more than message board discussion. I doubt highly that Zips would have predicted Coste's season last year, and if so, I'd love to know who's going to do it this year so we can have Gillick trade for him.

Hmm...I'm not sure that they take into account what team the pitcher plays for. I'm pretty sure some projection systems do. I think ERA was what ZIPS got 1st place in projecting last year. Seeing as how they are used for fantasy predictions, I imagine they are.

Willard Preacher: outside of the fact that it's unlikely that 100 players will be indicted on 'roids charges (most likely it would be used to figure out who lied to a grand jury), you completely miss the point of projections. No one is able to predict the next Chris Coste. What we can learn is that cases like that happen occassionally but rarely. Predictive models have some merit, so of course they can be used for more than message board discussion--real baseball teams use this with success. They don't say anything with 100% certainty, but they are an excellent guess. It's like a die roll-- you can't say definitively that a die will always land on 3.5 (it never will), but you can say that if you want to be as close as possible and that you win more money the closer you are, you beter guess 3.5 because that's the average value.

Maybe we can trade Chris Roberson for relief help. ZIPS makes him look like a superstar in-the-making. How do you project stats like those when he was a pinch runner/defensive replacement (and that's all he is)?

Great post Matt! You said pretty much everything I wanted to say. Just a few additional points.

I understand why people would be skeptical of some of the stats, but I don't understand calling it "crap" or "a hill of beans" stuff. This is such a reactionary position. What I mean is , SPECIFICALLY, which projections do you all find fault with?

ZiPS, or any other projection system, is a fairly useful tool to evaluate the upcomming year, nothing more, nothing less.

For example, some posters might be alarmed to see that of all the bullpen arms returning, only Geary and Gordon are projected for an ERA below 4.90. (The Rhodes projection of 4.20 is not happening, I think we can throw that out) This is VERY scary and fits in line with the general consensus that the team needs to add bullpen help.

Concerning the "mediocrity", Matt correctly points out that this is the 50th percentile prediction, so of course you are going to get a lot of 12-3 game winners and .280 BA. You would have to be crazy to actually predict around 58 HR for Howard next year...the guy had one of the greatest seasons in Phillies history last year, you can't just assume he will repeat that.

Would your professor have predicted 73 HR for Bonds in 2002, LF?

As to ballpark, ZIPS definately does factor in the ballpark and league which is one of the things that makes it valuable. Helms is a nice example of a player that will benefit from going from large stadium like Florida to CBP.

What are you talking about Reed? ZiPS projects Roberson for a .703 OPS, 6 HR and 38 RBI.

over 400 abs, 20 doubles, 17 sbs? Even the 6hrs & 38 ribs.
I am being a little sarcastic in my other post. There is no way in hell Roberson is capable of that.

Matt, I actually agree with you. You cannot predict who will have an out of this world year. I've always been steadfast that a player's career stats are the most fundamental predictor of future performance. Granted, when looking at players like Coste or Victorino, you must take into account the fact that their MLB sample size is incredibly small, so you have to discount some of that past performance. However, if I have a gun to my head and have to predict what kind of year Pat Burrell is going to have next year, I'd sign up for .258, 27 HR's, 92 RBI (.362/.479). This is slightly less optimistic than Zips projections, however, I believe that a player is what he is over time. I think that errors in overvaluing players based off of a single season have cost many teams a lot of money - see David Bell.

Fundamentally, we're saying the same thing here, as Zips appears to be in line, for the most part, with career stats.

Zito to the Giants according to mlb.com.

better than the Mets or Yanks, would've preferred an AL team though

ZIPs predicted Nunez would bat .279 last year.

I think my work here is done.

Met fans everywhere are starting to panic.

Remember, Zito is a Scott Boras client, and Boras could be leaking disinformation to ratchet up demand. I wouldn't bank on anything until there's a dotted line with Barry Zito's name on it.

I've never heard of an agent leaking information that a client would actually sign with a club. Definitely that a club was interested or had made an offer, but not actually agreed to terms. An agent starts doing that and he loses credibility with teams and, eventually, prospective future client base.

Seven years, $126 million for Zito, according to MLB.com & ESPN. Yikes.

So is this slightly more in depth than say auto playing a season in Madden on PS2?

The Mets must be fuming! The Giants could be an upcoming force in the West.

Preacher - you have a higher probability of being right than I do based on passed performance, but Burrell has a history of being up and down. I'm hoping(praying, actually) that it's more up than down, and he turns in a season of .280, 31 HR, 120 RBI, .385/.505. He was at .389/.504 in '05, and .388/.502 last year, so his OBP?SLG was pretty consistent. His strikeout rate was almost identical both years. What dropped was his batting average, and hence his HRs and RBIs. He also lost 100 ABs because of the drop.

His BABIP went down. Was he a victim of bad luck? Will he have a better year if his luck improves? Or did he just not hit the ball as hard in '06 because of his physical problems?

The comments to this entry are closed.

EST. 2005

Top Stories

HardballTalk

Rotoworld News

Follow on Twitter

Follow on Facebook

Contact Weitzel

CSG