I think that this is a fun exercise.

five guys that you think the projection systems are wrong on?

That Dana Eveland projection is a knee-slapper.

I think Miguel Cabrera's absolutely going to #### #### up this year. I'd take the over on .962.

i'll take the over on Cano in a heartbeat.

hell, i'm so convinced Cano is gonna have a big year that i'd take the over on most of *defensive* projections.

Harang, Justin Upton, Liriano, Helton, Khalil Greene: I'll take the overs

Bedard, Derrek Lee, Posada, Mike Lowell, Jurrjens: unders

i'll take the over on Cano in a heartbeat.

Cano is one I would probably put on my list.

I'll definitely take the over on Justin Upton.

I'll take the over on your mom.

I have no idea what they projected for Wieters, but obviously he'll beat it.

What will Wieters' OPS of 5.000 translate into for OPS+?

PECOTA projects Wieters to be the best player in the American League next year. To be fair, as Colin and others have pointed out, there are some very bizarre things about this projection, not the least of which is his PECOTA page having a similarity index of 0.

Delgado had a crazy second half last year, and he still barely cleared .853, posting an .871 OPS. In 2007, his OPS was .781. Frankly, I don’t see where this projection is coming from, and I certainly don’t believe it is right.

Delgado has posted an OPS of .848 or better in every year of his career except for 2007, when he was coming off wrist surgery. His slow start in 2008 might have been due to him breaking his hand late in 2007. Is it really that hard to see "where this projection is coming from?" It could be as simple as a guy being healthy after an injury riddled year.

I'd take the .853 OPS from Delgado.

Hi David

Just reading your article on 2009 THT projections that you thought might be wrong.

"What follows then is a list of 15 hitters and 14 pitchers who I think will either over- or underperform their projections, with my reasoning explained. I formed this list without looking at other projection systems, since the idea here is to figure out if human intuition can beat a computer-based system."

I'm not sure that cherry picking 29 players after the computer has done it's work, is exactly beating the system. I'm sure that if you were to independently do projections for 1,500 major leaguers this season, the computer could accurately find quibbles with 29 of your projections. It's always nice to pick the 29 we want to challenge, than let the other 1,471 go unchallenged. To truly beat the system you would need to project all the same players the computer did and then be more accurate using whatever comparison methodology one deems appropriate.

Let's look at the players you chose:

Justin Upton .771

other projections - KW 836, HQ 819, ZIPS 811, CHONE 827, BP 867

- so clearly everybody agrees with you; THT appears to be out to lunch.

Chipper Jones .972

other projections - KW 980, HQ 961, ZIPS 981, CHONE 942, BP 1007

- not sure I follow your logic here. Your criteria has been achieved 15 times in the past and Chipper is coming off a 1.044 OPS season at age 36, making him a prime candidate to be #16 in my mind. Other than a little support from CHONE you seem to be on your on on this one.

Miguel Cabrera .962

other projections - KW 956, HQ 974, ZIPS 941, CHONE 941, BP 896

- can't disagree with your points....but you don't seem to consider that Cabrera has not yet reached his peak (he will be only 26 in 2009) and did post a .951 OPS in the 2nd half last season. Perhaps it just took him a couple of months to adjust to the AL. Looks like BP agrees with you but nobody else sees the same decline that you do.

David Ortiz .951

other projections - KW 950, HQ 963, ZIPS 957, CHONE 956, BP 879

- because of the uncertainty of the health status he is a hard guy to project, but other than BP there seems to be considerable agreement. He did post a .914 OPS after returning from his wrist injury. So I'm not sure how that equals "did not look very good". I guess if you compare to earlier seasons you could say that, but nobody is projecting a 1000+ OPS. I suspect that part of his early season problems were maybe trying to play with the injury.

Alex Gordon .794

other projections - KW 785, HQ 833, ZIPS 785, CHONE 789, BP 799

- I tend to agree with you here. He posted a .965 OPS over the last 32 games. I think I went too low. A break through season could well happen in 2009. His much improved plate discipline in 08 is also very encouraging. Shandler also agrees with you.

Mike Napoli .875

other projections - KW 887, HQ 902, ZIPS 830, CHONE 805, BP 826

- most projectors tend to agree with you to some extent, but he did have a .960 OPS last season, so how surprising would .875 really be.

Delmon Young .731

other projections - KW 747, HQ 736, ZIPS 745, CHONE 775, BP 745

- he did OPS .789 after June 7th, after going .669 over the first 60 games, but he still didn't walk. Nobody seems to agree with you. His horrible batting eye does make significant improvement unlikely, but maybe his last four months is his true talent level.

Carlos Delgado .853

other projections - KW 858, HQ 860, ZIPS 820, CHONE 819, BP 841

- ZIPS and CHONE agree with you to some extent. Why are you so quick to discount his 1.018 OPS after June 26th. It's not as if he hasn't done it before. No matter how you slice it the man has a .929 career OPS (much higher if you only take his past 11 seasons) and is coming off a 1.018 OPS for the last 90 games last season. How about this run from 1998 to 2006 - .977, .948, 1.134, .948, .955, 1.019, .907, .981, & .909. What makes .853 such a stretch ?


Robinson Cano .796

other projections - KW 804, HQ 790, ZIPS 808, CHONE 808, BP 742

- so much for using "human intuition", and even Carty's conclusion was picked apart quite successfully by "alskor", so I'm not sure that it's conclusions are all that valid. Not much agreement with you here. In fact BP has gone strongly in the opposite direction. All these projections (other than BP) are assuming a big bounce back for Cano from .715. Not sure how much bounce back you expect.

Jack Cust .893

other projections - KW 856, HQ 854, ZIPS 858, CHONE 827, BP 831

- everybody agrees with you here. Looks like an unrealistic projection from THT. Although it should be noted that he posted a .912 OPS with the A's in 2007


Ryan Howard .973

other projections - KW 962, HQ 949, ZIPS 958, CHONE 955, BP 921

- everybody has him slightly lower than THT, and BP has him much lower. Not sure I understand your comment though. The only difference between a high-K guy and a low-K guy is that the high K guy will rate to have a lower batting average. Certainly "betting" on him in 2005 or 2006 or 2007 would have reaped huge dividends. And he did post a .999 OPS in the 2nd half of last season. That qualifies as "superstar level of performance" in my book. Not to mention .923 in 2005, 1.084 in 2006, and .976 in 2007. I'm not sure how the K's are a problem. We know he's gonna strike out a lot and we project that way. It actually makes predicting his BABIP less critical than for a guy who puts a lot of balls in play.

Ichiro Suzuki .738

other projections - KW 780, HQ 765, ZIPS 837, CHONE 745, BP 697

- wow, what a variation. ZIPS is totally with you, and BP thinks that another big drop is coming in 2009. I find it hard to fathom that he won't show significant improvement in 2009.

Evan Longoria .837

other projections - KW 870, HQ 868, ZIPS 831, CHONE 822, BP 824

- well, Shandler and I agree with you to a certain extent but nobody else is as confident of a 900 OPS as you are.

BJ Upton .775

other projections - KW 836, HQ 819, ZIPS 811, CHONE 827, BP 791

- I agree with you, with the highest of any of the projections. But I am not basing my projection on last year's postseason.

Chris Davis .884

other projections - KW 828, HQ 894, ZIPS 818, CHONE 808, BP 802

- everybody agrees with you except Shandler

Dan Haren 4.22

other projections - KW 3.31, HQ 3.46, ZIPS 3.41, CHONE 3.72, BP 3.53

- who knows what THT is smoking.....even if Haren puts up exactly a 4.22 ERA this is still a ridiculous projection

Derek Lowe 3.35

other projections - KW 3.60, HQ 3.50, ZIPS 3.73, CHONE 3.54, BP 3.70

- well nobody else thinks he's gonna post a 3.35 ERA, so you found another outlier

Clay Buchholz 4.68

other projections - KW 3.98, HQ 3.98, ZIPS 4.48, CHONE 4.27, BP 4.56

- I'm with you here. All prognosticators are more optimistic than THT but Shandler and BP are in the same neighbourhood.

Mark Buehrle 4.69

other projections - KW 4.16, HQ 4.16, ZIPS 4.10, CHONE 4.52, BP 4.58

- again we agree, but CHONE and BP aren't too sure

Edinson Volquez 4.16

other projections - KW 3.76, HQ 3.87, ZIPS 3.97, CHONE 3.79, BP 4.21

- I agree with you, as do the others to some extent

Fausto Carmona 4.06

other projections - KW 4.10, HQ 3.98, ZIPS 4.06, CHONE 4.36, BP 4.68

- BP is with you and to some extent CHONE

Jeremy Bonderman 4.14

other projections - KW 4.06, HQ 4.09, ZIPS 4.45, CHONE 4.06, BP 4.50

- ZIPS and BP are with you

Zack Greinke 4.44

other projections - KW 3.62, HQ 3.66, ZIPS 4.11, CHONE 4.20, BP 3.96

- I would say that THT is out to lunch....good catch

Francisco Liriano 4.11

other projections - KW 3.54, HQ 3.46, ZIPS 3.76, CHONE 3.97, BP 4.16

- well BP is smoking the same brand as THT, but you are bang on here.

C. C. Sabathia 3.23

other projections - KW 2.98, HQ 3.02, ZIPS 3.07, CHONE 3.41, BP 3.43

- not much support for you here

Justin Duchsherer 3.73

other projections - KW 3.70, HQ 3.60, ZIPS 3.30, CHONE 3.96, BP 4.02

- some support from CHONE and BP. He may not be pitching much this year.
He misses most of the season in 2007 with an injury and when he comes back in 2008, they decide he's a starter. How insane can you be. This frequently happens to healthy relievers when they are asked to become starters.


Dana Eveland 3.93

other projections - KW 4.17, HQ 4.36, ZIPS 4.36, CHONE 4.13, BP 4.51

- everybody is in agreement with you. Looks like THT is a little too optimistic.

Joe Blanton 4.26

other projections - KW 4.04, HQ 4.39, ZIPS 4.53, CHONE 3.88, BP 4.43

- THT seems to be in the middle of the pack


Best Regards.....Ken

I think that this is a fun exercise.

Me too. However, I'd be more interested in whether a Grizzled Old Scout or maybe Joe Morgan can beat the system for a select few players.

I'm not sure that cherry picking 29 players after the computer has done it's work, is exactly beating the system.

I see is just the opposite. In part that's because I expect the outcome will be that DSG can't beat the computer (I've spent too much time reading comments on Szymborski's blog that tried to do the same and failed), so it would be best for my argument's sake if we let the human picker choose the terms. And if he does succeed, the point is that somebody, a GM maybe, doesn't have to beat the computer on every pick to add value. He just has to beat the computer on guys he has on his roster (or could've been on the roster but aren't on it).

I'm not sure that cherry picking 29 players after the computer has done it's work, is exactly beating the system.

I see is just the opposite. In part that's because I expect the outcome will be that DSG can't beat the computer so it would be best for my argument's sake if we let the human picker choose the terms. And if he does succeed, the point is that somebody, a GM maybe, doesn't have to beat the computer on every pick to add value. He just has to beat the computer on guys he has on his roster (or could've been on the roster but aren't on it).

You could take any prediction system, find the outliers, pick against them and do very well.

For example using the THT projections if you simply took Justin Upton, Jack Cust, BJ Upton, Suzuki, BJ Upton, Chris Davis, Haren, Buchholz, Liriano, Grienke, & Eveland you would have a very good comparative record.

Off the top of my head you could do the same thing with ZIPS using Springer, Lowrie, McLouth, Nelson Cruz, Hochevar, Ichiro, etc.

With Shandler you have Furcal, Napoli.

CHONE.....Shoppach. BP.....Christian Guzman.

Best Regards.....Ken

I think picking 29 players does tell you something, since I didn't look at what other projection systems had to say. Rather, the computer looks at each player's past statistics and tells us what it expects him to do next season, and that's what I did, too. For 29 players, my results differed significantly from the computer's. The question is, Can I see something the computer cannot? We'll see at the end of the season.

This reminds me of this:

http://www.insidethebook.com/ee/index.php/site/comments/joe_sheehan_v_the_monkey/

I'd take the .853 OPS from Delgado.

For a team expected to be a top contender, that figure from your starting 1B isn't all that hot (but that's a different issue).