Saturday, December 3, 2011

Hot Stove Update: Projecting the Giants

When the Hot Stove is burning low and the winter chill starts to creep into the room, sometimes there isn't anything for a poor baseball fan to do than just fantasize about what their favorite players are going to do in the upcoming season. In this formerly blissful Field of Dreams, everybody hits .300(or if not, at least .280) and hits at least 20 HR's. In recent years, this reverie has been harshly interrupted by a phenomenon known as "projections". Projections are an attempt to use statistical analysis to inject some reality into the fantasies about what players will do in the upcoming season.

Some of these projection systems are quite simple, such as simply taking the player's average for the past 3 seasons and projecting that as their performance for next season. Primitive, but surprisingly accurate. Others are more sophisticated and take not only the player's recent performances, but also career trajectory and secondary stats into account. One of the more advanced projection metrics is one known by the acronym ZIPS. It was invented by a dude named Dan Szymborski, who publishes them at Baseball Think Factory and also on Fangraphs. ZIPS tends to be more conservative than some of the other systems out there. Szymborski recently completed his projections for the Giants. I thought it might be interesting to see what a possible Giants lineup might do statistically using ZIPS projections. I vaguely recall doing something like this last offseason and concluding that the Giants were a lock to repeat as World Series Champions, but hey, let's try it again!

CF Melky Cabrera .284/.330/.435, 36 2B, 5 3B, 13 HR, 15 SB.
2B Freddy Sanchez .273/.313/.373, 20 2B, 5 HR. (370 AB's).
3B Pablo Sandoval .299/.347/.497, 36 2B, 4 3B, 23 HR
C Buster Posey .287/.365/.452, 25 2B, 15 HR.
1B Aubrey Huff .261/.329/.422, 28 2B, 16 HR.
RF Nate Schierholtz .267/.316/.420, 21 2B, 8 HR, 6 SB(333 AB).
LF Brandon Belt .268/.365/.452, 28 2B, 5 3B, 17 HR, 12 SB.
SS Brandon Crawford .225/.291/.336, 19 2B, 4 3B, 7 HR, 6 SB.

Possible FA Signing:

Carlos Beltran .282/.363/.474, 26 2B, 4 3B, 14 HR, 7 SB.(397 AB).
Cody Ross .254/.317/.422, 27 2B, 16 HR. (472 AB).

Possible Reserves:

Mike Fontenot .245/.308/367, 16 2B, 3 3B, 4 HR. (278 AB)
Jeff Keppinger .278/.321/.372, 24 2B, 5 HR. (435 AB).
Chris Stewart .239/.309/.331, 14 2B, 3 HR. (272 AB).
Hector Sanchez .245/.288/.365, 26 2B, 9 HR. (479 AB).
Brett Pill .267/.298/.412, 35 2B, 16 HR. (600 AB).
Justin Christian .247/.298/.355, 24 2B, 4 3B, 6 HR, 25 SB. (462 AB).
Andres Torres .235/.311/.394, 24 2B, 5 3B, 9 HR, 16 SB. (383 AB).

Starting Pitchers:

Tim Lincecum 16-9, 2.92, 212.7 IP, 75 BB, 225 K.
Matt Cain 14-8, 3.04, 213 IP, 62 BB, 171 K.
Madison Bumgarner 13-10, 3.45, 198.3 IP, 48 BB, 164 K.
Ryan Vogelsong 8-10, 4.18, 153 IP, 60 BB, 119 K.
Barry Zito 7-8, 4.28, 124 IP, 52 BB, 92 K.

Eric Surkamp 7-7, 3.85, 142.7 IP, 59 BB, 116 K.

Relief Pitchers:

Brian Wilson 5-3, 3.02, 62.7 IP, 28 BB, 69 K.
Sergio Romo 4-1, 2.20, 49 IP, 11 BB, 62 K.
Santiago Casilla 3-2, 3.40, 53 IP, 26 BB, 48 K.
Ramon Ramirez 4-3, 3.32, 65 IP, 27 BB, 58 K.
Jeremy Affeldt 3-2, 3.52, 53.7 IP, 26 BB, 48 K.
Javier Lopez 3-3, 3.70, 48.7 IP, 22 BB, 33 K.
Guillermo Mota 2-2, 4.05, 60 IP, 24 BB, 51 K.

Dan Runzler 3-3, 3.80, 64 IP, 38 BB, 63 K.

Notes:

All projection systems rely heavily on regression to career and league averages. They do not predict breakouts or collapses very well. I wish they would either project the stats over a full season's worth of AB's or else try to be more realistic about how many AB's the players will get in reality.

Most of these systems really think Belt can hold his own and more through a full season of starting. That's a huge factor in how the Giants plan for the future. So much is riding on whether the Giants believe in him as much as the projection systems and whether he actually plays up to that.

Signing Beltran makes the lineup better, but only marginally so if the opportunity cost is Belt or even Aubrey Huff. Obviously the projection is skeptical about Beltran's ability to stay healthy. If he's only good for <400 AB's, then he would be a bad investment. A Michael Cuddyer or Josh Willingham might stay healthier, but the marginal benefit may be even less than with Beltran.

For the difference in price and the better versatility, Fonty looks like a better utility IF over Keppinger.

For the difference in price, Justin Christian looks like a better option than Andres Torres.

I would probably "eyeball" Vogey to be better and Zito to be worse than the projection. If Zito pitches that well, he will almost certainly pitch more innings. The system is very optimistic about Surkamp. Nice to know he might be a good fallback option if somebody gets hurt or if Zito is done.

If you believe the projection, the Giants should trade one of the either Lopez or Affeldt and go with Runzler as the second lefty reliever.

Just for interest's sake, here's some selected projections for prospects. Remember this is how the system thinks they would produce if they played at the MLB level this year:

Gary Brown .275/.335/.398, 30 2B, 11 3B, 8 HR, 36 SB. (621 AB).
Conor Gillaspie .261/.324/.376, 26 2B, 7 3B, 6 HR. (556 AB).
Francisco Peguero .274/.295/.375, 16 2B, 9 3B, 5 HR, 19 SB. (485 AB)
Tyler Graham .246/.293/.308, 16 2B, 2 3B, 2 HR, 39 SB. (426 AB).
Tommy Joseph .226/.266/.368, 30 2B, 17 HR. (598 AB).
Roger Kieschnick .226/.269/.367, 21 2B, 5 3B, 12 HR. (477 AB).
Ehire Adrianza .234/.295/.332, 28 2B, 5 3B, 4 HR, 15 SB. (509 AB)
Jarrett Parker .212/.300/.319, 24 2B, 3 3B, 11 HR, 16 SB. (589 AB).
Chris Dominguez .222/.257/.354, 30 2B, 4 3B, 15 HR. (630 AB).

Heath Hembree 1-1, 3.63, 52 IP, 32 BB, 58 K.
Dan Otero 3-3, 3.86, 49 IP, 12 BB, 36 K.
Hector Correa 4-5, 4.45, 62 IP, 29 BB, 47 K.

11 comments:

  1. DrB....
    Thanks for posting this. Much more realistic than our own fantastical projections, indeed.

    I'd like to think Posey and Nate can both hit a few more homeruns next season. With Posey perhaps a bit rusty, I suppose I wouldn't necessarily bet on it. However, Nate hit 9 HRs in the same amount of his projected ABs. And that was his first as a regular...and even that was somewhat sporadic. I'd like to think he'll get more than 330 ABs, and if so, I wouldn't be surprised to get 15 out of him.

    Unfortunately, if Aubrey hits that line, I'd be pleasantly surprised. And I agree, if Belt can attain those projections, perhaps acquiring another OF is not worth it. In fact, I'd say it isn't if we knew in advance he'd do that well. No way of knowing, though.

    And finally, I may be coming around to the thought of Christian being our backup CF. With Belt and Nate being lefties, and Melky a switch hitter, JC would be a nice righty off the bench with speed. Melky can move over to the corners on some days, with JC getting the occasional start in CF.

    Lucky

    ReplyDelete
  2. I was going to write on this on my blog and still might but I love projections and I hate projections.

    I love projections because it gives me an objective view as to the players MLE so that I get a guesstimate on what we might get from a prospect or player. I can then plug that into the lineup calculator to get an idea how that lineup might do, from a quantitative basis, instead of my qualitative ones I used to do. That analysis is invaluable in trying to assess how the team might do in the next season.

    I hate projections because 99.999% of the people misuse the projections in discussions. You cannot just trot the projection out and say, "see? End of story." Any projection is just the average of a range of possibilities for a player. The risk of error is minimized by, say, putting it into a lineup calculator and the ups will on average balance out the downs.

    For example, Brandon Belt. It is correct to say, like DrB says, that he has the potential for that. That weighs heavily in how Belt is ranked by Baseball America and Klaw.

    However, their rankings is NOT a guarantee that he will ever reach that potential nor that he will do it in 2012, it really only means that when he does put it all together, that batting line is what is expected based on past production. As with any prospect, when he develops is the big question mark, it could be 2012, it could be 2013. See how Matt Williams developed for an example of what I mean, he struggled for a few years before reaching his potential.

    If we put him in and he struggles again, our offense would be screwed. And our season would be wasted, which many fans complain that Sabean is doing yet do not realize that installing Belt in without a good backup in place is doing the same thing. That is why I like Sabean getting vets, people do not realize it but they are insurance policies for our upcoming season, sometimes they pay off, sometimes they do not, but if you do not have them, you are screwed. It is risk mitigation, I have been writing on this for years now.

    ReplyDelete
  3. Projections that rely entirely on statistical evidence run the risk of falling flat when confronted by reality. If the statistician looked up from his spreadsheets for a moment and actually watched Runzler and Affeldt pitch, he wouldn't favor Runzler over Affeldt. Runzler can't get a breaking ball over the plate when it counts to save his life.

    And if the projections are anything close to correct, then Gary Brown plays every inning in center field and there's no talk of Christian or Torres even on the team.

    Gary Brown hasn't played an inning above high A level so how that projection was manufactured seems to have a lot more smoke and mirrors involved than actual relevant data.

    If the Giants had a .398 OBP and 36 SB from their center fielder last year, we'd be talking about what a great season the Giants had. And if they get it this year from Gary Brown, the entire pitching staff will pitch in to buy him as many Mercedes as he can fit in his garage.

    ReplyDelete
  4. Walter,

    Projections are projections and there's a reason why they play the games. One thing you have to understand about these is that the published lines are actually means out of a range of possible outcomes. The more data the projection is based on the narrower the Range and Standard Deviation. You would thus expect that projections for a 30 year old vet with 6 or 7 years of experience might be more precise than ones for a rookie or somone with very limited playing time.

    As I pointed out in the original post, projections are terrible at predicting breakouts and collapses. By their very nature, they are conservative while predicting breakouts and collapses is risky.

    Also, Gary Brown's OBP projection is 335, not .398.

    ReplyDelete
  5. Seeing the average age of the lineup right now cracks me up - NTG the old man at 28, Melky at 27 and Belt 24 opening day. That would really screw with the whole Greybeards don't play the kids theory eh? Or the 5/8 being homegrown guys, shh don't tell anybody Sabean actually isn't all about vets.

    Ty Cobb is a complex character, most likely the most natural hitter to play the game, going hard every play and yes, most likely a racist and tortured human being. His quote about baseball sticks with me. There is a reason they play the game, and its between the lines that matter. The other quote is Mark Twain, who is often misquoted, but it involves lies and statistics. That gets lost way too often in this crazy age where everything is at your fingertips.

    All signs lead to Belt being a big key to the season. If he succeeds fans will yell "I told you so Bochy, all he needed was more ABs". If he doesn't there needs to be a plan B (as fans stumble around and grumble). And that is the hard part about being a GM. Beltran provides consistency, at a certain price and risk the Giants may not be able to match. If it was easy everybody would do it.

    The best marginal moves to make are to re-sign Ross and sign Hairston Jr. 2 right hand batters who can hold their own at the plate, play good defense and baserunning and provide coverage all over, the 4th OF/utility IF role and caddy the 2 Brandons.

    Liked the projections on Pill except the 600 ABs. If he can help out at 1B/OF and provide some sock off the bench he can be pretty useful. Getting 30 points of OBP above his BA might be tough for him though.

    ReplyDelete
  6. Broadly, I think the Vet projections are too conservative and the rookie & reserve projections really, really optimistic.

    It's gotta be highly unlikely that Brown would hit .275 out of the gate at MLB level in '12. Pill (who I love, and am glad he's on the minds of our front office guys) is good for 35 doubles and 16 HRs? C'mon.

    I think I understand how the system is averaging out past performance for the vets. But what is the algorithm for projecting performance for kids with no track record? Seems off.

    ReplyDelete
  7. Good points, everybody. I'm wondering if the minor league numbers are corrected for level but not for league? If that is the case, then the hitter friendly leagues like the PCL and the Cal League would inflate the projections for players coming out of them.

    I wouldn't totally scoff at Brown's projection though. Will Clark broke in straight out of the Cal League, I believe. Sandoval and Posey essentially did. Heck, for all practical purposes, Posey broke in straight out of college! It would be a risk, but I think there is a decent chance Brown could put up close to those numbers if he was the starting CF next season. He played for a high level college program in a league with a lot of really good pitching.

    ReplyDelete
  8. i project that if that is the starting lineup for 2012, i will have more enjoyment watching games...even if they lose

    ReplyDelete
  9. I understand how and why you feel that way, Bacci, but you understand that a lot of the Sabean detractors don't feel that way, right?

    ReplyDelete
  10. BCB has an interesting chart up with the various usual suspects at short: Reyes/Rollins/Furcal/Hairston/Gonzalez/Betancourt. Ranks them by 4 metrics. Rollins looks like the safe option but the analysis also makes a pretty good case for Hairston as the best fit for the Giants, because he's the only one who has played 2nd regularly and would be willing and experienced as well as cheap in years and dollars. I was surprised at how low Reyes ranked in fielding, which makes me want him even less.

    I also have checked out some phillies blogs about Rollins. Its interesting because they are watching what the Giants are doing/not doing very closely and definitely view us as the main competition. There are so many teams in the ss fa market they are all watching each other and trying not to blink. Seems like MIL is doing exactly what the Giants are doing, but they may dive into the big guys. I'm not a big Furcal fan although a 1 year prove it deal would be OK. He's going to get a lot more I think. This maybe a time where one in the hand is worth 2 in the bush. If money became available Rollins would be my preference by a longshot.

    In a market where every MI has doubled their salary and received 2 years, I have to imagine that Hairston is worth 2/8MM. Why hasn't it happened yet? Agent playing MIL/SF off each other for more $? He's not getting more years I don't think. The market might get some more guys from us with the Hobbit/Kepp and the Cards with Theriot/Schumacher in a few days. Haven't heard anything about Nick Punto. Hairston's speed and ability to play the OF is appealing, although I bet his signing would make Torres return much less likely.

    Baggs opened up more Coco Crisp/Willingham speculation for the Giants. Annoying to me. Crisp is a walking injury, very similar to Torres and whats the point in grabbing Melky if not to avoid a 2-3 year deal? Willingham annoys me because you put up a couple more million a year and you have Beltran, a much better hitter, fielder and baserunner. As one year deals they wouldn't be that bad, but they are both getting multi-years. Ross makes much more sense in that light.

    Things should get a little wild this week, after the dead hot stove. And yeah DrB, the detractors want the kids played, but not during the offseason, and free agents signed but not the bad ones who turn into big busts and cramp our payroll. Agent mouthpiece Heyman claims that Timmy wants 8 years and the G's offered 4. Sounds about right to me. This will go on for a while.

    ReplyDelete
  11. Right now, I'd say re-upping Fonty and re-signing Ross and DeRosa just might be the best options out there.

    Bench would be CStew, Pill, Fonty, Ross and DeRosa. All the positions are covered with a backup. You've got a lefthanded leaning starting lineup with a mostly RH bench for PH or platooning.

    The detractors who like to call the owners cheapskates would have a fit, but I'm with Bacci on this. I'd love to watch it play out, win or lose.

    There's lots of guys who are going to be in Fresno who can step up midseason if needed.

    ReplyDelete