General Offseason Thread (with projections)

[BigRapidsJackass]

Aspire, you raise an interesting theoretical problem and an interesting real-world problem.

The theoretical problem: how can you do any kind of reliable projection (typically based on weighted performance over 3 years) on a guy who only played -- at any level -- for about 2/3 of one year? We'd all have to admit that our level of confidence in any projection for Hamilton has got to be a lot lower than for the typical player.

The real-world problem: given how well Hamilton performed last year, and given his obvious tools, something must have gone wrong. Remember, the Reds had a coach basically living with him and watching his every move to keep him out of trouble. My guess (and it's only a guess): after his initial success, Hamilton got tired of being treated like a kid at summer camp. He wanted a regular degree of freedom, and the Reds think he hadn't earned it yet. He's a nice success story, but maybe the Reds are onto something: the percentage of hard-core drug addicts (and that's what he was) who are able to remain clean over an extended period of time is not reassuring.

So I'm surprised they pulled the plug on him, and I think they could've done better than Volquez. But the Reds know more about him than we (or anyone else) does, so I'm not willing to call Krivsky an idiot based on this move alone. Now as for his other moves, that's a body of work that certainly doesn't inspire confidence.

[BigRapidsJackass]

Aspire, you raise an interesting theoretical problem and an interesting real-world problem.

The theoretical problem: how can you do any kind of reliable projection (typically based on weighted performance over 3 years) on a guy who only played -- at any level -- for about 2/3 of one year? We'd all have to admit that our level of confidence in any projection for Hamilton has got to be a lot lower than for the typical player.

The real-world problem: given how well Hamilton performed last year, and given his obvious tools, something must have gone wrong. Remember, the Reds had a coach basically living with him and watching his every move to keep him out of trouble. My guess (and it's only a guess): after his initial success, Hamilton got tired of being treated like a kid at summer camp. He wanted a regular degree of freedom, and the Reds think he hadn't earned it yet. He's a nice success story, but maybe the Reds are onto something: the percentage of hard-core drug addicts (and that's what he was) who are able to remain clean over an extended period of time is not reassuring.

So I'm surprised they pulled the plug on him, and I think they could've done better than Volquez. But the Reds know more about him than we (or anyone else) does, so I'm not willing to call Krivsky an idiot based on this move alone. Now as for his other moves, that's a body of work that certainly doesn't inspire confidence.

EDIT: from his Wikipedia entry

He stopped using drugs on October 6, 2005; in 2006, Hamilton said that he had been in drug rehabilitation eight times since 2003, and had attempted suicide at least three times.[12]

8 times in rehab since 2003; 3 suicide attempts; Reds watching him like a hawk last year ... HF, you tell me what the proper discount rate is for that kind of player. I'm guessing you get less than 50% of his abstract value.

Oh, before anyone else says it:

Steve Howe and Alan Wiggins.

[Heltonfan]

We'd all have to admit that our level of confidence in any projection for Hamilton has got to be a lot lower than for the typical player.

Sure. But as I said before, a low level of confidence just means that the error bars are wider, not that the mean is lower. Hell, there's no better illustration of that than the '07 performance of Hamilton himself.

Besides, there are no red flags in Hamilton's performance. .318 BABIP - that's above average, but not anywhere near the point where we would conclude that he was really lucky. His K/BB numbers are league-average; if he had a 1:1 ratio or something, I would suspect that it was a colossal fluke and that pitchers simply had no idea what to do with him, because no one as inexperienced as he was should be commanding the strike zone that well, but that's not the case.

His numbers are impeccable. He's loaded with tools, a former #1 overall pick. His troubled past is a concern, absolutely. But it's the only concern.

8 times in rehab since 2003; 3 suicide attempts; Reds watching him like a hawk last year ... HF, you tell me what the proper discount rate is for that kind of player. I'm guessing you get less than 50% of his abstract value.

You may very well be right. But I'll take 40% of Hamilton over 100% of Volquez. That's how ridiculous this trade is from a talent exchange standpoint.

To put it another way: Krivsky got less in exchange for five years of slave-wage Hamilton than Schuerholz did for one year of market-value Renteria (and the guy who Schuerholz was trading with, Dave Dombrowski, is no idiot). That's just mind-boggling. Even if you think there's a 50/50 chance of Hamilton offing himself within the next three years, it's a terrible trade.

[Heltonfan]

PECOTAs are out. They feature the usual nonsense - there are about 600 players projected to be above-average, which obviously isn't anywhere close to being possible.

But, on the team level, this sort of stuff should mostly balance out, so I put together a set of team projections using PECOTA. Here's an overview of where all 30 teams stand right now, looking at both my system and PECOTA and highlighting places where they differ.

AL East
1. New York (95 HF wins, 90 PECOTA wins) - PECOTA thinks that Chien-Ming Wang and Phil Hughes are barely above average. I have them projected at ERCs of 3.47 and 3.26, respectively. The PECOTA pessimism on Wang is expected - it's been projecting doom for him and Aaron Cook for years now - but Hughes is tougher to explain, since the PECOTA MLEs are kinder than mine. I honestly have no clue what's going on there; I suspect my Hughes projection is a little too optimistic, but there's absolutely no way that he should project as low as PECOTA has him (3.94 ERC).

2. Tampa Bay (86 HF wins, 88 PECOTA wins) - yep, PECOTA loves the Rays even more than I do (to be expected, of course, given the MLE issue). I have yet to see any remotely compelling argument that the projections are wrong here; the Rays are being dismissed because of their poor 2007 record, but when you go through player by player, it's impossible not to like their chances. This is going to be the #1 shocker of 2008.

3. Boston (85 HF wins, 87 PECOTA wins) - not much to say here. This was not an elite team even before Schilling went down.
4. Toronto (82 HF wins, 83 PECOTA wins)
5. Baltimore (72 HF wins, 74 PECOTA wins)

AL Central
1. Detroit (88 HF wins, 86 PECOTA wins)
2. Cleveland (88 HF wins, 85 PECOTA wins) - Jake Westbrook is another in the Cook/Wang category, underrated by a similar-players-based system and evaluated properly by a simpler system that treats him as a distinct individual rather than trying to force him into a Procrustean bed (this is my big theoretical objection to PECOTA).

3. Minnesota (76 HF wins, 76 PECOTA wins)
4. Chicago (74 HF wins, 80 PECOTA wins) - wow. PECOTA loves the White Sox. Didn't see that coming. In case you're curious, the main reasons appear to be that 1) Juan Uribe is projected to have his best season since 2004, 2) A.J. Pierzynski, despite being a catcher in his 30s, is being expected to beat his '05-'07 averages, 3) Gavin Floyd and John Danks are projected not to suck, 4) Linebrink is projected to have his best year since 2005, and 5) PECOTA predictably loves the young hitters (Quentin and Fields). Uh, yeah. PECOTA and I were the lone voices in the wilderness warning that the '07 White Sox were going to be terrible. We were right. And they haven't improved much. It's also rather amusing to note that, as much as PECOTA seems to be engaging in wild optimism here, it still can't get this team over .500.

5. Kansas City (70 HF wins, 74 PECOTA wins)

AL West
1. Los Angeles (91 HF wins, 87 PECOTA wins) - PECOTA hates Garland and Escobar. Garland's worst FIP in the past three years is 3.80 - he's given a 4.13 ERC projection. Escobar's worst is a 3.33 - he's given a 3.57 projection. No idea why.

2. Oakland (HF 83 wins, PECOTA 83 wins) - this is noteworthy only in that there appear to be quite a few idiots out there who see the A's in fire-sale mode, and, based on the fact that they only won 76 games last year, judge this a surefire last-place team. Not happening.

3. Texas (80 HF wins, 78 PECOTA wins)
4. Seattle (75 HF wins, 75 PECOTA wins) - horrible offense, poor defense, no bullpen after Putz. PECOTA has produced its usual laughably pessimistic Ichiro projection (see my Procrustean bed comment above) - this year, he's projected a hair worse than Timo Perez - but on balance, we agree. This is a bad, bad team, which now has no future whatsoever.

NL East
1. New York (89 HF wins, 91 PECOTA wins) - yeah, that Santana guy helps.
2. Atlanta (87 HF wins, 83 PECOTA wins) - this is an interesting one. When I look through PECOTA's individual projections for the Braves, they look quite reasonable. Which, counterintuitively, means that PECOTA doesn't like the Braves at all, because their projections for MLB as a whole are ridiculously optimistic.
3. Philadelphia (82 HF wins, 87 PECOTA wins) - PECOTA thinks that Moyer and Eaton are going to blow away their three-year averages. Think about that. Jamie Moyer, who turns 46 this year, is projected to blow away his three-year average. A selection of pitchers that PECOTA thinks are worse than Moyer: Tim Hudson, Brad Penny, Derek Lowe, Jeff Francis, Aaron Cook, Greg Maddux, Matt Cain, Tom Gorzelanny, Jake Westbrook, Mark Buehrle, Gil Meche, Chien-Ming Wang, Phil Hughes, Dustin McGowan. High comedy indeed.

4. Florida (73 HF wins, 73 PECOTA wins)
5. Washington (71 HF wins, 73 PECOTA wins)

NL Central
1. Chicago (91 HF wins, 86 PECOTA wins) - see Atlanta comment.
2. Milwaukee (85 HF wins, 84 PECOTA wins)
3. Cincinnati (78 HF wins, 81 PECOTA wins) - PECOTA loves Bruce, Votto, and Volquez, as expected. It hates Homer Bailey, though. PECOTA rates Bailey behind Victor Zambrano, Tim Redding, and something named Esmerling Vasquez.
4. Pittsburgh (75 HF wins, 74 PECOTA wins)
5. St. Louis (72 HF wins, 76 PECOTA wins) - PECOTA likes Pineiro and Anthony Reyes a lot more than I do. And it projects an unbelievable .280 ABR for D'Angelo Jimenez.
6. Houston (72 HF wins, 75 PECOTA wins)

NL West
1. San Diego (86 HF wins, 84 PECOTA wins)
2. Arizona (84 HF wins, 84 PECOTA wins)
3. Los Angeles (84 HF wins, 82 PECOTA wins)
4. Colorado (82 HF wins, 77 PECOTA wins) - I've already mentioned PECOTA's issues with Aaron Cook. But there's an even bigger discrepancy to address - PECOTA hates Matt Holliday. It projects him for a .299 ABR (about halfway between his '05 and '06 performance). Other players with a PECOTA-projected .299 ABR include Joe Dillon and Corey Hart. Nick Swisher, who has never come particularly close to outhitting Holliday, is projected at .305. Are we the division favorites? Absolutely not. But the idea that we're closer to the Giants than we are to SD/AZ/LA is ridiculous.
5. San Francisco (72 HF wins, 74 PECOTA wins)

I suspect that one could make a ton of money by betting on the Rays to finish ahead of the Mariners, if one were so inclined...

[bedir than average]

[cough] Bedard[/cough]

I'm surprised that PECOTA thinks so poorly of the Seattle Bullpen.

[Heltonfan]

Aside from Putz and Sherrill, there are four relievers who were significant contributors to the '07 Mariners: Sean Green, Brandon Morrow, Eric O'Flaherty, and Ryan Rowland-Smith. Out of those four:
- Three have below-average control. One couldn't find the strike zone with two hands and a flashlight.
- Two have good strikeout rates, the other two are average-ish.
- Only one has a good groundball rate.
Any projection system that thought highly of that group would be drunk out of its mind.

I've been following the happenings at USSM over the past week or so, concerning the Bedard trade and all, and I just keep coming back to the conclusion that, if the M's hadn't gotten so lucky last year, none of the projections would be the least bit controversial. They were a 77.5-win team by the component stats. They're not particularly young. Their bullpen is in line for substantial regression across the board. And they didn't make unusually large personnel improvements this winter.

Which brings me to another point that I've been meaning to get around to, concerning the whole idea of measuring offseason improvement: on the surface, one might say that the Bedard trade represents a two-win upgrade, and the Silva signing a 1.5-win upgrade. And one would be correct. But that doesn't mean that we should revise the Mariners' projected win total up by 3.5 games compared to where it was at the beginning of the offseason. Why? Because everyone makes upgrades like that.

There have been about 80 wins bought on the free agent market this winter. Add in another 10 wins for trades (just a guess), and we're up to 90 wins gained this winter in MLB... which, of course, averages out to 3 wins per team. So the Mariners haven't actually taken a step forward this winter; they've just held serve. Of course, this whole issue just serves as another illustration of how much cleaner the "from the ground up" analysis is than the much more common "change from last year" approach...

[John Cocktoston]

Looks as though Hawpe and Atkins got done, which is no great surprise. It also looks as though Fuentes is going to go to the mattresses, which is also no great surprise. That thing was going to a hearing all along...

[TheIncredibleRox]

Apologies if this leaves others grossed out at the nerdiness, but I'm putting together my post-'08 spreadsheets and trying to follow HF's method for automating the projections...

For grabbing minor league stats, assuming that you're working from the MiLB.com data, you have to create a second player ID, of the form Franklin MoralesCOL (the team indicator is necessary here because in the minors, there are so many players that you can easily run into the "two guys with the same name" problem). And apply the same principle as above, putting this ID column in both the projections sheet and the minor league sheet. Then, in your target data cells (i.e. the one in which you want to have Morales' combined 2007 IP total, between all three levels), you should sum the results of the major league stats lookup and the minor league stats lookup.

What makes the parent organization necessary? In other words, why not use the year? I'm just envisioning problems with a guy who changed organizations. It seems it would present problems not only for the season in question, but for example putting together an '09 projection for a guy who switched orgs between '07 and '08. I mean, I have "Chris RichardTB" for this season, but if you want to talk about past or future seasons, that's not going to work. And if he had changed orgs this season mid-year, lookup wouldn't find it.

[Heltonfan]

You're only putting your end-of-year sheets together now? Geez. Took you long enough. I had it done a week ago.

Quoting myself:

the team indicator is necessary here because in the minors, there are so many players that you can easily run into the "two guys with the same name" problem.

Off the top of my head, I know that there's an Aaron Cook in the Angels organization, and a Ramon Ramirez in the Reds organization. I'm sure there are more examples.

Now, obviously, there are far more examples of guys changing teams than there are of duplicate name problems. But the latter are harder to spot and more cumbersome to fix (you'd either have to change one of their names or go in and tweak the lookup formula itself in that cell).

As far as players changing organizations, I added a second team column (which can just be blank if there is no second team) to handle the mid-year trades problem. And with a little pivot table magic, you can quickly produce a list of all the guys in the minors who played for multiple organizations this year, and use that to correct the team fields in the projection sheet. And beyond that, I so rarely have to add anyone to the spreadsheet (I've got like 2700 players on it now) that I can just grab data from previous years manually, and if I had lookups for that, it would substantially increase the size of the file, and the thing calculates slowly enough as is.

Anyway, your way (year-based lookups rather than team-based) should work fine - I'm mostly just sticking with my system because I'm used to it - but you have to watch out for the duplicate name thing.