Stupid question: why do people think the Angels offense will be terrible this year?

Best reason is designated outmaker Shea Hillenbrand. Garret Anderson is a pretty weak excuse for a #4 hitter. Seattle has hitters as good or better (Guillen, Johjima) hitting 6th and 7th. Right now there is no reason to ever pitch to Vlad. Mike Napoli could help the offense if he can hit .220, but he's got to prove he can make enough contact to hit .220, and he hasn't shown power since last July - its tough to hit for power when you can't even meet the ball.

If I were in charge we would have signed JD Drew instead of Matthews, he'd be hitting behind Vlad and playing centerfield. That would make Maicer the leadoff hitter.

If I were in charge we would have signed JD Drew instead of Matthews

Wow - you are smart. I would never have thought of that...

Why the sarcasm? Bill Stoneman didn't think of it. Or if he did, wasn't willing to go very far with it. They offered Soriano a good bit of money, not quite what the Cubs gave him, but a lot more than Boston gave Drew. I would have offered Drew the same contract Soriano turned down.

Second time the Angels passed on Drew to sign an inferior player - after 2004 it was Steve Finley.

Why the sarcasm?

I'm joking. I think signing Matthews was really not smart. I mean that's not a slap at you, but rather the idea that signing Matthews was a good idea (which clearly the Angels thought).

Angels apparently thought it was a good idea, but most of the fans did not.

Matthews is probably a better option than doing nothing and filling CF with Figgins, Willits, Terry Evans, Aybar, whoever, but is not a cost effective upgrade. Angels needed another big bat, did not get it. Now Vlad gets hit in the hand. If he's out for any length of time, we can forget about 2007.

If that's the case I hope Stoneman at least has the sense to trade Colon (impending free agency) as soon as (and assuming) el Barto comes off the DL and has a few decent starts.

I find it particularly amusing that mgl was arguing the same point regarding the same player back in 2004-2005 (although from the opposite direction), and was standing virtually alone.

When the Mariners signed Beltre there were a lot of comments on how he had turned the corner in 2004. mgl pointed out the same things he is in this thread about regression, not ignoring data points, etc. There were a lot of posters here that wanted to ignore or discount a few of the years before due to the appendectomy. mgl basically just shrugged (well, with mgl it's never as quiet as a shrug) and said that you had to take every season into account.

Now some people want to throw out 2004 (because it's obviously an outlier) and again mgl says to take every season into account.

Greg, you have a good memory! It's not really my opinion. The research is easy enough to do. As I always say, take all players in history who have any particular pattern you want to investigate. Given a reasonably large sample and you get your answer (always with some uncertainty of course). Ignoring any season, good or bad, or doing anything but a basic Marcel gives you the WRONG answer. Again, that is without knowing anything pertinent about a player's health, etc.

That does not always mean that a good projection turns out the way you project (duh!). A projection is always just a best estimate of a player's true talent. There are three ways a projection can turn out wrong (or some combination of the three). One, the estimate of the true talent is wrong because the historical data represents the player having gotten lucky or unlucky. Two, the future performance (against which the projection is compared) is lucky or unlucky. And three, the player's true talent changes, above and beyond the typical aging curve.

Someday maybe we'll stop presenting a projection as one number. It is really a distribution of numbers with any paricular number or interval of numbers having a certain probability of occurring (the single-number projection being the mean or median of that distribution).

Regardless of how a projection is presented, what constitutes being right or wrong? If I say that Beltre has a 25% probability of being between .780 and .810, 15% prob of being .770 to .780, 15% between .810 and .820, etc, and he ends up at .812, was I right or wrong?

Even if G-d comes down and tells us a player's exact projection, his performance after that, say in one year, is still going to fluctuate randomly. As I have also said many times, given a series of perfect projections (where G-d tells you the real number), a certain percentage of performances will be very far from that perfect projection. This whole idea of mentioning how someone was wrong (incrediby wrong sometimes) or right about a particular projection, as if this, in and of itself, is an indictment of that person's projection methodology, is incredibly silly.

If the greatest forecaster in the world were not "wrong" much of the time, and incredibly wrong some of the time, either he got incredibly lucky (depending how many projections we are judging), or there is something very, very freaky about that person...

Matthews is probably a better option than doing nothing and filling CF with Figgins, Willits, Terry Evans, Aybar, whoever, but is not a cost effective upgrade.

It's a particularly poor use of money. He's only going to be marginally better than that collection, and there is probably some FAT out there which makes it worse, IMO.

Ignoring any season, good or bad, or doing anything but a basic Marcel gives you the WRONG answer.

No, it doesn't. And you go on to say so below.

"Regardless of how a projection is presented, what constitutes being right or wrong?"

Scroll back in time and tell me what your projection for Beltre was for 2005. a 90 OPS+? What were the odds of that? Even with a weighted Marcel.

It's a particularly poor use of money. He's only going to be marginally better than that collection, and there is probably some FAT out there which makes it worse, IMO.

I won't argue its poor use of money. Marginally better indeed. The kind of thing that might make an 89 win team into a 90 win team. I disagree on the FAT, there might be a centerfielder freely available who can give you the same or better production, but the odds are against you finding one with that route. More likely you'd waste 150 AB here, 200 AB there, get poor production from them, and sooner or later your season is over and you played it with a big hole in CF.

I wish they had spent the money needed to get a true upgrade, and I'll be angry when they opt not to sign Andruw Jones because they already have 40 million tied up in center field, but having Matthews start the season in CF made the Angels slightly more likely to win the 2007 AL West than not having Matthews.

Of course, that's keeping his preseason projections in mind, as one should when its only the middle of April. If he continues hitting .220 and dropping flyballs I'm going to be very angry about him as this year progresses.

Ignoring any season, good or bad, or doing anything but a basic Marcel gives you the WRONG answer.

No, it doesn't. And you go on to say so below.

In hindsight, dropping Beltre's 2004 season would have greatly improved his 2005 projection. But who knew that beforehand? You didn't - I'll quote:

Here’s what I see: There’s almost no chance that Beltre will fall to a 109 OPS+.

And even if that was wrong and Beltre fell much further than that, your reasoning was sound.

Forecasters know that if we try and subjectively pick out certain data points, we are going to be wrong more often than we are right.

But who knew that beforehand? You didn't

Well, sort of. I *did*, but after I went through the research I convinced myself that my initial expectation almost couldn't be right. turned out it was.

What's that SAT -taking tip? Go with your first instinct?

Yes, it could be that we are wrong with respect to Young or anyone else. What is the point of quoting how someone did last year, last month, or two years ago, when we are talking about our estimate of true talent and about future projections?

I guess it depends whether we are rating Young prospectively, or retrospectively.

Since every year is a sample of a player's true talent, it is simply his ecact true talent for that year plus or minus random sample error.

The existence of 'true talent,' of which each PA is a sample, is an assumption. It may be a plausible one, but it remains an assumption.