RUMOR: Cubs interested in Willits

[haltz]

note the consistently strong LD%, which is correlated with a high babip. add in a little bump for his speed and being quick out of the batter's box, and voila... high babip year in and year out.

He's fast and he hit 50% GB last season, but I might not put too much stock in LD rates. As a zero power hitter, if he's not a BABIP superstar, he's not going to have a lot of value as a slap-hitting speedster with a 20% strikeout rate. Even if his true talent is a .360 BABIP, it's not predictable that he posts an IsoD of .100 in the big leagues. That's what he was doing as an old-for-his-levels minor leaguer. Even if that walk-rate is real, a dip to a .330 BABIP makes him a sub-.700 OPS player last year.

 

PECOTA sees a .354/.354 line, which is just going to be a better answer than speculation.

[98navigator]

I kind of wish we just had gotten Josh Hamilton....then we could have used Pie to trade for a top pitcher.

 

The middle of the lineup could have been deadly. Who would it have taken? Gallagher? Marshall? Marshall is only one year older than Volquez with much more impressive numbers in the minors, plus more proven in the majors, where he has a season of 100+ innings with a sub-4 ERA and good control (above average BB/9 in 2007) and averaged 8.774 K/9 in the minor leagues.

 

I know this is meaningless now, but looking back on it, I wish it is what we had done.

 

Dream Lineup

 

1. Roberts S (2B)

2. Fukudome L (RF)

3. Lee R (1B)

4. Ramirez R (3B)

5. Soriano R (LF)

6. Hamilton L (CF)

7. Soto R ©

8. Cedeno R (SS)

 

A full lineup of above average to excellent hitters, no weak spots, all play above average defense.

 

Whatever, too late now.

 

My dream lineup

 

Soriano (2B)

Murton (LF)

Fukudome (RF)

Ramirez (3B)

Greene (SS)

Lee (1B)

Soto ©

Lofton/Fuld/Willits (CF)

 

Z, Hill, Burnett, Lilly, Lieber

 

The defense would suffer dearly with that situation. Soriano is LF and almost anyone else at 2B would work. Also, DLee is too far down in the lineup to fully utilize his high OBP. He'd lose too many ABs over the course of the year.

[Mephistopheles]

note the consistently strong LD%, which is correlated with a high babip. add in a little bump for his speed and being quick out of the batter's box, and voila... high babip year in and year out.

He's fast and he hit 50% GB last season, but I might not put too much stock in LD rates. As a zero power hitter, if he's not a BABIP superstar, he's not going to have a lot of value as a slap-hitting speedster with a 20% strikeout rate. Even if his true talent is a .360 BABIP, it's not predictable that he posts an IsoD of .100 in the big leagues. That's what he was doing as an old-for-his-levels minor leaguer. Even if that walk-rate is real, a dip to a .330 BABIP makes him a sub-.700 OPS player last year.

 

PECOTA sees a .354/.354 line, which is just going to be a better answer than speculation.

 

that asmodai guys pretty smart, eh?

[TruffleShuffle]

note the consistently strong LD%, which is correlated with a high babip. add in a little bump for his speed and being quick out of the batter's box, and voila... high babip year in and year out.

He's fast and he hit 50% GB last season, but I might not put too much stock in LD rates. As a zero power hitter, if he's not a BABIP superstar, he's not going to have a lot of value as a slap-hitting speedster with a 20% strikeout rate. Even if his true talent is a .360 BABIP, it's not predictable that he posts an IsoD of .100 in the big leagues. That's what he was doing as an old-for-his-levels minor leaguer. Even if that walk-rate is real, a dip to a .330 BABIP makes him a sub-.700 OPS player last year.

 

PECOTA sees a .354/.354 line, which is just going to be a better answer than speculation.

 

a guy who can get on base 39% of his PAs has value, regardless of how much power he has.

 

i also think that pecota is wrong sometimes.

[haltz]

a guy who can get on base 39% of his PAs has value, regardless of how much power he has.

No joke, but I didn't question whether a .390 OBP is valuable. The question is if Willits is that guy.

 

i also think that pecota is wrong sometimes.

In a given season, it may or may not be right, but that depends on if you view it as a crystal ball for the next season (which it is not) or if you view it as a true talent estimator based on the information we have at the time (which it's pretty good at).

[CubinNY]

In a given season, it may or may not be right, but that depends on if you view it as a crystal ball for the next season (which it is not) or if you view it as a true talent estimator based on the information we have at the time (which it's pretty good at).

The statement above is a lot of words that essentially means nothing, isn't it? I'm not exactly sure what "true talent" means though. What value does PECOTA if it cannot make accurate predictions? Or I guess I should say more accurate than other predictions.

[Mephistopheles]

True talent is a players intrinsic skill level. Players don't always produce at this level, but are expected to produce seasons with a normal distributions centered around this level.

 

Example. If a player's true level is a 800 OPS and there's a 30 pt standard deviation for the normal player in a given season. Then we would expect his OPS to be above 830 about 16 percent of the time. We would expect his OPS to be below 770 about 16 percent of the time.

[seanimal]

In a given season, it may or may not be right, but that depends on if you view it as a crystal ball for the next season (which it is not) or if you view it as a true talent estimator based on the information we have at the time (which it's pretty good at).

 

The statement above is a lot of words that essentially means nothing, isn't it? I'm not exactly sure what "true talent" means though. What value does PECOTA if it cannot make accurate predictions? Or I guess I should say more accurate than other predictions.

 

I'm convinced you didn't actually read the sentence. He was pretty clear. Thinking that PECOTA, or any prediction scheme, is a "crystal ball" is dumb. You then proceed to fault it because it's not a crystal ball. Chew on that for a while.

[CubinNY]

True talent is a players intrinsic skill level. Players don't always produce at this level, but are expected to produce seasons with a normal distributions centered around this level.

 

Example. If a player's true level is a 800 OPS and there's a 30 pt standard deviation for the normal player in a given season. Then we would expect his OPS to be above 830 about 16 percent of the time. We would expect his OPS to be below 770 about 16 percent of the time.

Intrinsic skill level?

[CubinNY]

In a given season, it may or may not be right, but that depends on if you view it as a crystal ball for the next season (which it is not) or if you view it as a true talent estimator based on the information we have at the time (which it's pretty good at).

 

The statement above is a lot of words that essentially means nothing, isn't it? I'm not exactly sure what "true talent" means though. What value does PECOTA if it cannot make accurate predictions? Or I guess I should say more accurate than other predictions.

 

I'm convinced you didn't actually read the sentence. He was pretty clear. Thinking that PECOTA, or any prediction scheme, is a "crystal ball" is dumb. You then proceed to fault it because it's not a crystal ball. Chew on that for a while.

It doesn't taste all that good so I spit it out.

 

I'm questing what value PECOTA has if not for making predictions, since that's what it appears to be used for. If it's not any more useful than a guess based on past performance why does one need all the complicated math.

 

I think it may be time to change the tampon. You're a little cranky.

[seanimal]

OH I GET IT! I'M A MENSTRUATING FEMALE!

[Rob]

That's a very sexist implication on his part.

[David]

OH I GET IT! I'M A MENSTRUATING FEMALE!

 

lulz i see what he did there!

[haltz]

I'm questing what value PECOTA has if not for making predictions, since that's what it appears to be used for. If it's not any more useful than a guess based on past performance why does one need all the complicated math.

It is better than guessing, which is what all the math in the black box is for, but it's statistically impossible for it to nail player seasons across the board. If god told you in OPS terms exactly how good Matt Murton is, that doesn't mean he'll hit at exactly that level next year; there's expected variance in a given season. Sample size works both ways.

 

Even on a team level, I can run 100 seasons with ZIPS on Diamond Mind, and the Cardinals might average 77 wins, but they'll also win the division outright 7 times. Does that mean that they are likely to win the division or that to do so is their true talent? No. Is it useless information? I don't think so.

 

Usually people will say something like "well, PECOTA didn't predict the Rockies to be in the World Series" or "the games aren't played on paper LOLOL" but all that means is that something that may have been unlikely happened, and as far as the latter goes, statheads will be paying attention too because they understand that it isn't a crystal ball.

[TruffleShuffle]

i think there is some skill in projecting that certain players will be better or worse than their pecota projections. common sense, scouting, etc.

 

i predict that ronny cedeno and felix pie will be worse than their pecota projections for 2008, and not by an insignificant amount.

[CubinNY]

I'm questing what value PECOTA has if not for making predictions, since that's what it appears to be used for. If it's not any more useful than a guess based on past performance why does one need all the complicated math.

It is better than guessing, which is what all the math in the black box is for, but it's statistically impossible for it to nail player seasons across the board. If god told you in OPS terms exactly how good Matt Murton is, that doesn't mean he'll hit at exactly that level next year; there's expected variance in a given season. Sample size works both ways.

I don't think anyone who halfway knows how projections work expects something like PECOTA to be exactly accurate. I think it would be much better to give ranges and confidence intervals. Something like, We predict Matt Murton's OPS will be between .800 and .850 with 95% confidence. We predict it will be between .825 and .850 with 70% confidence. We predict it will be between .700 and 1.000 with 100%.

 

Nonetheless, if God told me what Murton's OPS was going to be, the first thing I'd do is seek medical help. If God was then wrong, the second thing I'd do is question his deity credentials.

 

This is the problem I have with some sabermentricly informed opinions. If the model is wrong it it is the model that is at fault. If the model says such and such a player will perform at X level and the player underperforms or overperforms, it is the model at fault and not the player, unless there are special circumstances.

 

A players "true" performance is what how he actually performs, not what PECOTA says he will do. The value of PECTOA lies in its ability to approximate a player's "true" performance.

[frostwyrm]

Again, what use would Willits be except the short end of a CF platoon and PR? Maybe I'm out of step with the crowd here but I just don't believe in using a zero-power player as a pinch hitter unless bases are empty. If it's late in the game, you have men on base, and you need a run, then you better PH with somebody who hits the ball hard enough to move runners around the bases and hopefully get an RBI. What I would not do is put a slappy OBP guy out there in hopes that he'll get a walk.

[BigbadB]

Again, what use would Willits be except the short end of a CF platoon and PR? Maybe I'm out of step with the crowd here but I just don't believe in using a zero-power player as a pinch hitter unless bases are empty. If it's late in the game, you have men on base, and you need a run, then you better PH with somebody who hits the ball hard enough to move runners around the bases and hopefully get an RBI. What I would not do is put a slappy OBP guy out there in hopes that he'll get a walk.

 

And we have one of those guys already in Theriot when Cedeno beats him out of a job.

[CubColtPacer]

Again, what use would Willits be except the short end of a CF platoon and PR? Maybe I'm out of step with the crowd here but I just don't believe in using a zero-power player as a pinch hitter unless bases are empty. If it's late in the game, you have men on base, and you need a run, then you better PH with somebody who hits the ball hard enough to move runners around the bases and hopefully get an RBI. What I would not do is put a slappy OBP guy out there in hopes that he'll get a walk.

 

I would agree that Willits should only be used as a pinch hitter with nobody on. That's not exactly a detriment though. The Cubs could use a player like that on the bench. Ward and Murton are the two best to bring on with runners on. Cedeno and Blanco give you the occasional power, but neither are ideal with the bases empty. Willits becomes the guy who pinch-hits with the bases empty.

 

I'm not saying that we desperately need to trade for him to accomplish that, but I think he would be useful as both a platoon partner and a pinch hitter. That's still only worth so much.

[Rob]

I'm questing what value PECOTA has if not for making predictions, since that's what it appears to be used for. If it's not any more useful than a guess based on past performance why does one need all the complicated math.

It is better than guessing, which is what all the math in the black box is for, but it's statistically impossible for it to nail player seasons across the board. If god told you in OPS terms exactly how good Matt Murton is, that doesn't mean he'll hit at exactly that level next year; there's expected variance in a given season. Sample size works both ways.

I don't think anyone who halfway knows how projections work expects something like PECOTA to be exactly accurate. I think it would be much better to give ranges and confidence intervals. Something like, We predict Matt Murton's OPS will be between .800 and .850 with 95% confidence. We predict it will be between .825 and .850 with 70% confidence. We predict it will be between .700 and 1.000 with 100%.

 

Nonetheless, if God told me what Murton's OPS was going to be, the first thing I'd do is seek medical help. If God was then wrong, the second thing I'd do is question his deity credentials.

 

This is the problem I have with some sabermentricly informed opinions. If the model is wrong it it is the model that is at fault. If the model says such and such a player will perform at X level and the player underperforms or overperforms, it is the model at fault and not the player, unless there are special circumstances.

 

A players "true" performance is what how he actually performs, not what PECOTA says he will do. The value of PECTOA lies in its ability to approximate a player's "true" performance.

 

All you've probably ever seen from PECOTA is the weighted mean projection. In reality, they break it down much further than that. Look at Murton for example.

 

http://img407.imageshack.us/img407/176/murtonjo8.png

 

And that's merely a fraction of his PECOTA card.

[frostwyrm]

I would have been far more agreeable to having Willits last year, seeing as we had Cliff Floyd and Daryle Ward in the OF. Willits could have been a useful defensive sub.

[CubinNY]

I'm questing what value PECOTA has if not for making predictions, since that's what it appears to be used for. If it's not any more useful than a guess based on past performance why does one need all the complicated math.

It is better than guessing, which is what all the math in the black box is for, but it's statistically impossible for it to nail player seasons across the board. If god told you in OPS terms exactly how good Matt Murton is, that doesn't mean he'll hit at exactly that level next year; there's expected variance in a given season. Sample size works both ways.

I don't think anyone who halfway knows how projections work expects something like PECOTA to be exactly accurate. I think it would be much better to give ranges and confidence intervals. Something like, We predict Matt Murton's OPS will be between .800 and .850 with 95% confidence. We predict it will be between .825 and .850 with 70% confidence. We predict it will be between .700 and 1.000 with 100%.

 

Nonetheless, if God told me what Murton's OPS was going to be, the first thing I'd do is seek medical help. If God was then wrong, the second thing I'd do is question his deity credentials.

 

This is the problem I have with some sabermentricly informed opinions. If the model is wrong it it is the model that is at fault. If the model says such and such a player will perform at X level and the player underperforms or overperforms, it is the model at fault and not the player, unless there are special circumstances.

 

A players "true" performance is what how he actually performs, not what PECOTA says he will do. The value of PECTOA lies in its ability to approximate a player's "true" performance.

 

All you've probably ever seen from PECOTA is the weighted mean projection. In reality, they break it down much further than that. Look at Murton for example.

 

http://img407.imageshack.us/img407/176/murtonjo8.png

 

And that's merely a fraction of his PECOTA card.

That's not what I'm arguing for. In fact, it is entirely the opposite. But really, I'm not arguing against PECOTA at all. I think it is an extremely useful tool. The problem lies not in the numbers generated but in the interpretation of the numbers.

[Post Count Padder]

Bump for this:

 

General manager Tony Reagins said Wednesday that the Angels may send Reggie Willits back to Triple-A in order to get him regular at-bats.

 

The Angels' outfield is extremely crowded and Willits is the most likely odd man out, but he doesn't deserve a trip back to the minors after a very strong rookie season that saw him bat .293 with a .391 on-base percentage and 27 steals in 136 games. Complicating the matter is that Nathan Haynes is out of options, so the Angels could choose to keep him around over Willits in order to retain both players.

[TheDude]

In a given season, it may or may not be right, but that depends on if you view it as a crystal ball for the next season (which it is not) or if you view it as a true talent estimator based on the information we have at the time (which it's pretty good at).

 

The statement above is a lot of words that essentially means nothing, isn't it? I'm not exactly sure what "true talent" means though. What value does PECOTA if it cannot make accurate predictions? Or I guess I should say more accurate than other predictions.

 

I'm convinced you didn't actually read the sentence. He was pretty clear. Thinking that PECOTA, or any prediction scheme, is a "crystal ball" is dumb. You then proceed to fault it because it's not a crystal ball. Chew on that for a while.

 

I don't think it's a matter of him not reading your statement. I'm in agreement with CubinNY here (which might be a first); it's a semantics issue. What good a is forecast model that causes a person to be labeled "dumb" for using it to forecast a player's production? Don't call PECOTA a forecast tool if you don't want anyone to use it as a "crystal ball" to predict production. Based on what you and Meph are stating, it's a capability model, not a forecast model.

[TruffleShuffle]

a guy who had a .391 OBP is going back to the minors this year. great work, bill stoneman.