NFBC Message Boards

I like Jose Bautista. He messes with our minds. He is 3b eligible and hit over 50 home runs last year. Those two facts alone, would give first round status to most, but not Bautista.



It seems he didn't figure out Major League pitching till August of '09. Since then, he's the best power hitter in baseball. Most 'projections' have him hitting 25-35 home runs. No nads, really.

If you believe in Bautista, this season should be similar to last season. Even if he was 'a little' lucky last year, similar numbers with 40-45 home runs should be in order, right?



On the other end of the spectrum are the disbelievers. Why would Bautista hit 25-35 here?

He's never hit more than 18 before last year. If last year had you unbelieving, then he should fall back to 15-20 home run status.



The truth always lies in the middle.

At least for projectors.

Before last season, to project anything over 20 home runs for Bautista would bring one subject to ridicule.

Take a look back at Shandler and Joba a couple of years ago.

Hated the prediction, loved the cajones.



If I see a Bautista prediction for 25-35 home runs, I chuckle. In code, it means, 'I don't know what to do with this guy'. But, they can't put down 'I don't know' so, most run for middle ground. 'Malcolm in the Middle' has been replaced by 'Jose in the middle'.

He is the ultimate Missouri/Missouri guy.

Some will say show me twice and pass on him, some may overdraft him in the second round, believing in how much a 50 home run guy at 3b can help a fantasy team.



He is indeed, a wild card.

Last years other difference maker, Carlos Gonzalez, has first round status assured.

His skeletons in the closet were the injuries leading up to last year, both in the minors and majors. With the full season last year, all is forgotten on that front.



Projectionist are more ok predicting full years for injury prone players. They leave it up to us to reckon if there will be an injury. If injury occurs, it's not on them.

Ian Kinsler has stellar numbers on paper. But, after most seasons, his numbers don't compare with players that played with a full season of at bats.

Injury worries me more than whether a guy had a career year or not.

I have a bet with a friend this year that Bautista finishes higher on a certain player rater than CarGo. I know the odds are against me, but I'm going for the volume produced in Bautista's at bats over the overall talent and fragility of Gonzalez.

Hey, at the least, I'm not stuck in the middle.



[ February 14, 2011, 12:16 PM: Message edited by: DOUGHBOYS ]

Given where Sandoval and Reynolds were drafted (mid 2nd to mid 3rd) last year, Bautista looks like a steal in the late 3rd round in comparison.

Ned Pepper would call your bet 'mighty bold talk............'

However, I agree with you on Cargo and Bautista. They may both have spectacular seasons in 2011, but I'm not looking for spectacular at their ADP's. I reckon I'll be be settling for stability in their neighborhoods, "but it's enough that you know that I'll do what I have to do."

Trying to make some sense of it all,

But I can see that it makes no sense at all



excellent point about injury risk Dan, too often ppl (myself included) get caught up w/ the #'s we analyse & project for players. You absolutely better factor in LIKELIHOOD a player getting injured in to your draft evaluation. You have only yourself to blame if you end up w/ Chipper at 3B



there are ex's of taking a chance that have worked out...another 2B (like Kinsler who has hx of injury risk) Rickie Weeks, if you ignored his fragile past & took a flyer on him staying healthy it certainly paid off last yr.



what about back in (08?) when Pujols dropped to late 1st rd b/c of concerns about poss elbow injury/pending surgery. Those who got him didn't complain.



though yer injury risk argument must give those w/ designs on taking Cargo in 1st rd pause, esp if Braun & Crawford still avail.

The thing with Bautista's '10 year is that he wasn't the only Jay to have a power surge. The phenomenon was team wide for the most part. Can the common denominator be Cito & Dwayne Murphy's coaching?

Dough, from one projection guy to one non-projection guy, let me make a case.



Projections aren't made with the expectation of being right, they are made for 3 reasons:



(1) It forces the projector to delve into past results and other potentially relevant factors and to make presumptions about them which may be relevant to future results. So even if I didn't carry those projections with me to the draft table, they've given me something.



(2) A projection is designed to quantify a probability. If I believe there is a 50% chance Jose hits 45 HRs this season, and 50% he hits 25 HRs, and those odds are reasonable, Jose becomes a probable 35 HR hitter for me. Along with projecting his other stats, that allows me to put a value on Jose and compare him to alternatives and to build a balanced team based on probable outcomes - which, if the methodology is sound, should give my team a reasonable facsimile of what I am projecting aside from unexpected injury risk.



(3) And this is to me the critical part of doing projections ... If I develop projections for Jose I know what assumptions go into those projections. If, on the other hand, I am taking projections from Zola, Schandler or Rotoworld I don't know what assumptions are in there. At this point I not only lose insight behind the numbers, which can hugely affect the expected risk/variance of achieving those numbers (and who I want to draft for a particular team), I also lose the entire benefit of item #1 above.



While I can appreciate a seat-of-the-pants approach from an exceptionally well-read and close observer of baseball such as yourself, I can't imagine approaching a draft without doing what I do - which is making a lot of projections and knowing what has gone into those projections. We probably would agree on a couple things however; there is minimal value in relying on the projections of others, and if you draft based only on projections (even your own) you're not doing yourself any favors.



On another subject - the Schandler projection for Joba ... while you saw cajones, I saw a poorly-executed attempt at grandstanding. Ditto Mike Jacobs the following year. If you're going to grandstand, make damn sure you know what you're doing.



[ February 14, 2011, 07:02 PM: Message edited by: KJ Duke ]

Originally posted by KJ Duke:

Dough, from one projection guy to one non-projection guy, let me make a case.



Projections aren't made with the expectation of being right, they are made for 3 reasons:



(1) It forces the projector to delve into past results and other potentially relevant factors and to make presumptions about them which may be relevant to future results. So even if I didn't carry those projections with me to the draft table, they've given me something.



(2) A projection is designed to quantify a probability. If I believe there is a 50% chance Jose hits 45 HRs this season, and 50% he hits 25 HRs, and those odds are reasonable, Jose becomes a probable 35 HR hitter for me. Along with projecting his other stats, that allows me to put a value on Jose and compare him to alternatives and to build a balanced team based on probable outcomes - which, if the methodology is sound, should give my team a reasonable facsimile of what I am projecting aside from unexpected injury risk.



(3) And this is to me the critical part of doing projections ... If I develop projections for Jose I know what assumptions go into those projections. If, on the other hand, I am taking projections from Zola, Schandler or Rotoworld I don't know what assumptions are in there. At this point I not only lose insight behind the numbers, which can hugely affect the expected risk/variance of achieving those numbers (and who I want to draft for a particular team), I also lose the entire benefit of item #1 above.



While I can appreciate a seat-of-the-pants approach from an exceptionally well-read and close observer of baseball such as yourself, I can't imagine approaching a draft without doing what I do - which is making a lot of projections and knowing what has gone into those projections. We probably would agree on a couple things however; there is minimal value in relying on the projections of others, and if you draft based only on projections (even your own) you're not doing yourself any favors.

ding. ding

Fantastic post, KJ.

Originally posted by KJ Duke:

Dough, from one projection guy to one non-projection guy, let me make a case.



Projections aren't made with the expectation of being right, they are made for 3 reasons:



(1) It forces the projector to delve into past results and other potentially relevant factors and to make presumptions about them which may be relevant to future results. So even if I didn't carry those projections with me to the draft table, they've given me something.



(2) A projection is designed to quantify a probability. If I believe there is a 50% chance Jose hits 45 HRs this season, and 50% he hits 25 HRs, and those odds are reasonable, Jose becomes a probable 35 HR hitter for me. Along with projecting his other stats, that allows me to put a value on Jose and compare him to alternatives and to build a balanced team based on probable outcomes - which, if the methodology is sound, should give my team a reasonable facsimile of what I am projecting aside from unexpected injury risk.



(3) And this is to me the critical part of doing projections ... If I develop projections for Jose I know what assumptions go into those projections. If, on the other hand, I am taking projections from Zola, Schandler or Rotoworld I don't know what assumptions are in there. At this point I not only lose insight behind the numbers, which can hugely affect the expected risk/variance of achieving those numbers (and who I want to draft for a particular team), I also lose the entire benefit of item #1 above.



While I can appreciate a seat-of-the-pants approach from an exceptionally well-read and close observer of baseball such as yourself, I can't imagine approaching a draft without doing what I do - which is making a lot of projections and knowing what has gone into those projections. We probably would agree on a couple things however; there is minimal value in relying on the projections of others, and if you draft based only on projections (even your own) you're not doing yourself any favors.



On another subject - the Schandler projection for Joba ... while you saw cajones, I saw a poorly-executed attempt at grandstanding. Ditto Mike Jacobs the following year. If you're going to grandstand, make damn sure you know what you're doing. Yep.



Since all things have a tendency to regress to the mean, everything is pretty much stuck in the middle.

KJ, I understand that projections are very useful for some. For the most part, I am being flip at the folks who copy their 'findings' from publications.

I'll answer to your points anyway



1)I can understand how bringing projections to the table would be a good vehicle here. If we've done our homework we should no what each player is capable of whether projections are brought to the table or not. This gets us back to the 'to laptop, or not to laptop' argument, which I'll avoid here.



2)Bah! 50% one way and 50% the other means either 'I don't know or stuck in the middle with you. 35 home runs sounds reasonable to EVERYBODY. More or less, it is 'office cleanup' meaning little. In other words, Bautista has to be one of those players you're sold on or not. If yes, 40's. If no, 20's. I made a $20 bet after my post with a fella. He has any number in the 30's, I have the rest.



3) Agreed. Projections should be our own. If we copy from Zola, Shandler, or RotoWorld, they're certainly not going to refund our $1400. These projections are 'coveralls'. They cover trading leagues, 8 team leagues 10 team, 12 team, 15 team, you name it.



Projections made on our own are the result of study and are a viable vehicle for draft day.

Also, in your mind's eye, we should have an understanding of what each player is capable of, without projections.

Can we set up a league where no one has access to HQ, Rotoworld, Mastersball, Rotowire, James, CBS, ESPN etc projections but me? I'd pay an over the odds entry fee for that.



If the argument is that taking a cheat sheet from one of those services and blindly using it at the draft table is a weak strategy, then I agree.



However I strongly disagree that these projections aren't useful and should be ignored. My prep system is to take outside projections from a number of sources, break down some underlying stats from them (eg HR/PA, K/BB etc), combine them in a way that eliminates outliers, add my own adjustments for factors I've found to be important (eg health), then apply a valuation formula (which is baseed on SGP's).



If dozens of smart knowledgeble fantasy baseballers are spending time converting raw past data to current projections, why do I think I can do a better job? What I look for is which projections have done better in the past, then weight those more heavily.



This is a common statistical estimation technique that is used in many fields. (A company I was involved in used the same idea combining different weather forcasts and the combination produced forcasts superior to any single one).

Originally posted by jim.s:

Can we set up a league where no one has access to HQ, Rotoworld, Mastersball, Rotowire, James, CBS, ESPN etc projections but me? I'd pay an over the odds entry fee for that.



If the argument is that taking a cheat sheet from one of those services and blindly using it at the draft table is a weak strategy, then I agree.



However I strongly disagree that these projections aren't useful and should be ignored. My prep system is to take outside projections from a number of sources, break down some underlying stats from them (eg HR/PA, K/BB etc), combine them in a way that eliminates outliers, add my own adjustments for factors I've found to be important (eg health), then apply a valuation formula (which is baseed on SGP's).



If dozens of smart knowledgeble fantasy baseballers are spending time converting raw past data to current projections, why do I think I can do a better job? What I look for is which projections have done better in the past, then weight those more heavily.



This is a common statistical estimation technique that is used in many fields. (A company I was involved in used the same idea combining different weather forcasts and the combination produced forcasts superior to any single one). Jim, First, these guys are writers, not experts.

They're probably no more savvy than you when it comes to fantasy baseball. Before fantasy football season, these guys will be doing the same for that crowd, while I'm digesting the information and getting ready for the next fantasy baseball season.



They are paid to write, not win fantasy championships. They almost never color outside of the lines. For most players, their projections will be the same across the board. If most projections are the same, how can it possibly help you in your research?

By the way, Jim, I would love to be in a league without projections from these guys. In fact we had an e-mail draft that started as soon as the last out was made last season. I would have been quite comfortable drafting against the other 14 even if these writers were in their ear, giving them their projections for the coming year.



Now, let me propose something to you. Would you go into a paperless, computerless draft? These really test mettle. We had a local league that did it for a couple of years, but proved to tough to find replacements for.



Rules were, that no paper, computers, writing instruments, and hands were checked for 'crib notes'. The only one who could write was the director who wrote down names as they were called.

It was only a $20 league, but paid for itself in preparation. The opposite of a 'projections' league.



[ February 15, 2011, 09:18 AM: Message edited by: DOUGHBOYS ]

I think one point KJ makes that is dead-on is that we really have little idea as to what goes into projections you don't do yourself. Do they use a black box like PECOTA? Are they more subjective, like Shandler? What factors do they ignore? How do they weigh them? Who is to say they are being weighed appropriately?



Trevor Cahill seems like a great example of this to me. Cahill is a fairly extreme GB pitcher(56% GB/29% FB) who had a great year last year. 18-8, 2.97 ERA, 1.10 WHIP. He also doesn't strike anybody out (5.4 k/9) and has even worse swinging strike% (5.9%). And, despite being a groundball pitcher, he had a .234 BABIP last year, the lowest in MLB. His LOB rate was high (77%), but his HR/FB was, if anything, high as well (although there is more variance in HR/FB amoung extreme GB pitchers because of the smaller sample size).



So, is he fraud? Or is he a young guy getting better? Or both?



Obviously, my initial reaction was that a .234 BABIP was a complete fluke for an extreme GB pitcher. Generally, pitchers like Cahill will be above the league average since GB are about 70% more likely to be hits than fly balls. Even giving Cahill the benefit of the doubt, I would think he would project for a BABIP around .290, which would materially impact his performance.



Then I looked at some projections. Marcel projected him for a .257 BABIP. PECOTA had him projected as having the lowest BABIP among SP. Shandler I think around .270. What was I missing? Cahill also had a low BABIP the year before (.272), but 370 IP isn't enough to conclude he has some kind of magical GB BABIP skill.



Turns out that the A's defense had BY FAR the lowest GB BABIP in the majors last year: .205 as compared to an AL average of .231. Hmmm. Is that what these projections are relying upon to get to that projected BABIP? Brett Anderson has the same GB%, but he had a .294 BABIP. Is it different because he is LHP? Are Ellis/Barton THAT good as compared to Pennington/Kouz?



If they were, you'd expect to see some kind of year to year constancy, right? Well, in 2009, the A's BABIP on GBs was .255 - one of the worst in the AL. But in 2009, you say, Barton didn't play much - it was Cust/Giambi. OK, in 2008, Ellis/Barton played even more than in 2010; they had a .232 BABIP on GB - basically average.



The problem is I have no idea what any of these projection systems are doing to get to an on-its-face unsustainable .260 BABIP for Trevor Cahill. If it's the GB defense behind him, I'm not convinced, but, of course, I don't even know if that's it. If it's his magic ability to induce weak GB contact right at fielders, I'm really not convinced. So, for me, the projection for Cahill is worthless.



Which, as KJ said, is why I do this myself. Even if I'm wrong, I feel like I'm right.

Dan- don't know about the rest- but i could not disagree more as it pertains to Masterball. Todd, for one, is not not a writer. He writes but i believe he is a chemistry major. Some of the other people who write about stats on these boards are smart people who do the stats as a hobby of sorts. Still, i believe they are very good at it. Masterball does have stats people doing the projections. You admit you won't pay for any of this but to criticize what people like masterball are doing without paying for; and thus seeing what they do provide, is a shot taken in ignorance. Masterball explains in detail what they do and how they do it and has tried repeatedly to explain what projections are for and what they are "saying" to us. To no avail apparently. I do know this- jim s., gecko, quohogs and kj duke are 4 stat guys. Quahogs uses them- to what extent i do not know, but he uses them. Those 4 are really really good and i would stack them up against any "seat of the pants" drafter. In fact, in my opinion, kj duke is probably a good example of why being great at the #'s part just won't get you there because his results are MUCH better in auctions than snakes. I belive there are reasons for this but those reason are not projection related. You need to understand the #'s to be a good at this and then you need to go from there to be the best.

hells satans. cahill- that drives me crazy. guess i am in the middle on this because i do use someone elses projections as a start, but cahill is one of those players that you could look at and say "luck" until you see this years and see those babips. it is a head scratcher. i won't do the #'s like you do, i guess i will be lazy and say he WAS lucky last year. will say this- i read where keith law loved him last year and it came from observing him quite a bit. Lots of things going on in this game which makes it fun. BTW- if you want to go even more crazy look at gavin floyd and ground balls. i watched a lot of sox games and that was the defense for sure. the question is- can beckham improve and konerko bounce back or has he just gotten old. i also think your groundball data tells you all you need to know as to why adrian beltre was signed by the rangers. they are owned by a pitcher. we love you mike- but you can't play 3rd base.

Originally posted by Hells Satans:

I think one point KJ makes that is dead-on is that we really have little idea as to what goes into projections you don't do yourself. Do they use a black box like PECOTA? Are they more subjective, like Shandler? What factors do they ignore? How do they weigh them? Who is to say they are being weighed appropriately?



Trevor Cahill seems like a great example of this to me. Cahill is a fairly extreme GB pitcher(56% GB/29% FB) who had a great year last year. 18-8, 2.97 ERA, 1.10 WHIP. He also doesn't strike anybody out (5.4 k/9) and has even worse swinging strike% (5.9%). And, despite being a groundball pitcher, he had a .234 BABIP last year, the lowest in MLB. His LOB rate was high (77%), but his HR/FB was, if anything, high as well (although there is more variance in HR/FB amoung extreme GB pitchers because of the smaller sample size).



So, is he fraud? Or is he a young guy getting better? Or both?



Obviously, my initial reaction was that a .234 BABIP was a complete fluke for an extreme GB pitcher. Generally, pitchers like Cahill will be above the league average since GB are about 70% more likely to be hits than fly balls. Even giving Cahill the benefit of the doubt, I would think he would project for a BABIP around .290, which would materially impact his performance.



Then I looked at some projections. Marcel projected him for a .257 BABIP. PECOTA had him projected as having the lowest BABIP among SP. Shandler I think around .270. What was I missing? Cahill also had a low BABIP the year before (.272), but 370 IP isn't enough to conclude he has some kind of magical GB BABIP skill.



Turns out that the A's defense had BY FAR the lowest GB BABIP in the majors last year: .205 as compared to an AL average of .231. Hmmm. Is that what these projections are relying upon to get to that projected BABIP? Brett Anderson has the same GB%, but he had a .294 BABIP. Is it different because he is LHP? Are Ellis/Barton THAT good as compared to Pennington/Kouz?



If they were, you'd expect to see some kind of year to year constancy, right? Well, in 2009, the A's BABIP on GBs was .255 - one of the worst in the AL. But in 2009, you say, Barton didn't play much - it was Cust/Giambi. OK, in 2008, Ellis/Barton played even more than in 2010; they had a .232 BABIP on GB - basically average.



The problem is I have no idea what any of these projection systems are doing to get to an on-its-face unsustainable .260 BABIP for Trevor Cahill. If it's the GB defense behind him, I'm not convinced, but, of course, I don't even know if that's it. If it's his magic ability to induce weak GB contact right at fielders, I'm really not convinced. So, for me, the projection for Cahill is worthless.



Which, as KJ said, is why I do this myself. Even if I'm wrong, I feel like I'm right. All those numbers, and I forgot the most important one. Cahill had a .153 BABIP on ground balls, which is .52 points below the A's average, which is already .26 points below the AL average. By comparison, Roy Halladay had a .220 BABIP on GB.



I'm willing to accept that RHP will have lower BABIPs on GB because they will face more LHB and therefore will see more balls hit to 1b/2b, meaning shorter/no throws to 1B (although I've not seen any numbers to support that assumption and would imagine they have a higher BABIP with runners on base because the 1B is holding the runner on). But not to that extent.

Originally posted by headhunters:

hells satans. cahill- that drives me crazy. guess i am in the middle on this because i do use someone elses projections as a start, but cahill is one of those players that you could look at and say "luck" until you see this years and see those babips. it is a head scratcher. i won't do the #'s like you do, i guess i will be lazy and say he WAS lucky last year. will say this- i read where keith law loved him last year and it came from observing him quite a bit. Lots of things going on in this game which makes it fun. BTW- if you want to go even more crazy look at gavin floyd and ground balls. i watched a lot of sox games and that was the defense for sure. the question is- can beckham improve and konerko bounce back or has he just gotten old. i also think your groundball data tells you all you need to know as to why adrian beltre was signed by the rangers. they are owned by a pitcher. we love you mike- but you can't play 3rd base. Funny you should mention the White Sox. They were by far the worst team in the AL last year in turning GBs into outs- .260 BABIP. I'm sure some of that is bad defense, but I have to think some of it is bad luck too. Morel is supposed to be a stud at 3B.

Mike, none of comments were meant to criticize Todd or Mastersball. I enjoy Todd's postings and writings more than most.

What Todd and others do, from my perspective, should be labeled 'for entertainment purposes only'. That could be taken as a shot at Todd and other writers, but it is'nt, it's meant for those who DEPEND on his numbers to be acutely accurate.



I have no problem with using any source as a starting point for study. I do have a problem with using it as a end all, be all. For a time, I read Shandler TOO closely. At drafts, I'd remember those damn little snipets that he would write, and I'd pick Shandlers player not mine. By saying this, I'm not blaming Shandler, I'm blaming myself for being like a crow with a shiny metal in sight.



I know Todd uses numerish to arrive at his projections. He has my admiration. But, I feel I am no more, less prepared bypassing these figures for a draft, then delving into it and seeing how figures were conceived for each player.



In the end, no matter how these projections are conjured, they all seem to have a similarity about them.

It is why I love Jose Bautista this year. My joke with friends is that if you love him, project 45. If you're a non-believer, project 20. If you're a writer, project 35.

Originally posted by headhunters:

Dan- don't know about the rest- but i could not disagree more as it pertains to Masterball. Todd, for one, is not not a writer. He writes but i believe he is a chemistry major. Some of the other people who write about stats on these boards are smart people who do the stats as a hobby of sorts. Still, i believe they are very good at it. Masterball does have stats people doing the projections. You admit you won't pay for any of this but to criticize what people like masterball are doing without paying for; and thus seeing what they do provide, is a shot taken in ignorance. Masterball explains in detail what they do and how they do it and has tried repeatedly to explain what projections are for and what they are "saying" to us. To no avail apparently. I do know this- jim s., gecko, quohogs and kj duke are 4 stat guys. Quahogs uses them- to what extent i do not know, but he uses them. Those 4 are really really good and i would stack them up against any "seat of the pants" drafter. In fact, in my opinion, kj duke is probably a good example of why being great at the #'s part just won't get you there because his results are MUCH better in auctions than snakes. I belive there are reasons for this but those reason are not projection related. You need to understand the #'s to be a good at this and then you need to go from there to be the best. I second the idea that better projections have a method behind them. It is somewhere beyond useless for a guy to sit around saying, "Pujols, meh, 40 homers sounds good."

Originally posted by Hells Satans:

I think one point KJ makes that is dead-on is that we really have little idea as to what goes into projections you don't do yourself. Do they use a black box like PECOTA? Are they more subjective, like Shandler? What factors do they ignore? How do they weigh them? Who is to say they are being weighed appropriately?



Trevor Cahill seems like a great example of this to me. Cahill is a fairly extreme GB pitcher(56% GB/29% FB) who had a great year last year. 18-8, 2.97 ERA, 1.10 WHIP. He also doesn't strike anybody out (5.4 k/9) and has even worse swinging strike% (5.9%). And, despite being a groundball pitcher, he had a .234 BABIP last year, the lowest in MLB. His LOB rate was high (77%), but his HR/FB was, if anything, high as well (although there is more variance in HR/FB amoung extreme GB pitchers because of the smaller sample size).



So, is he fraud? Or is he a young guy getting better? Or both?



Obviously, my initial reaction was that a .234 BABIP was a complete fluke for an extreme GB pitcher. Generally, pitchers like Cahill will be above the league average since GB are about 70% more likely to be hits than fly balls. Even giving Cahill the benefit of the doubt, I would think he would project for a BABIP around .290, which would materially impact his performance.



Then I looked at some projections. Marcel projected him for a .257 BABIP. PECOTA had him projected as having the lowest BABIP among SP. Shandler I think around .270. What was I missing? Cahill also had a low BABIP the year before (.272), but 370 IP isn't enough to conclude he has some kind of magical GB BABIP skill.



Turns out that the A's defense had BY FAR the lowest GB BABIP in the majors last year: .205 as compared to an AL average of .231. Hmmm. Is that what these projections are relying upon to get to that projected BABIP? Brett Anderson has the same GB%, but he had a .294 BABIP. Is it different because he is LHP? Are Ellis/Barton THAT good as compared to Pennington/Kouz?



If they were, you'd expect to see some kind of year to year constancy, right? Well, in 2009, the A's BABIP on GBs was .255 - one of the worst in the AL. But in 2009, you say, Barton didn't play much - it was Cust/Giambi. OK, in 2008, Ellis/Barton played even more than in 2010; they had a .232 BABIP on GB - basically average.



The problem is I have no idea what any of these projection systems are doing to get to an on-its-face unsustainable .260 BABIP for Trevor Cahill. If it's the GB defense behind him, I'm not convinced, but, of course, I don't even know if that's it. If it's his magic ability to induce weak GB contact right at fielders, I'm really not convinced. So, for me, the projection for Cahill is worthless.



Which, as KJ said, is why I do this myself. Even if I'm wrong, I feel like I'm right. Yes, the A's had the best defense in baseball last year. You can look at Dewan's +/- numbers, team defensive efficiency, or just the plain old park effect on defense (the A's catch more pop-ups due to the widest foul territory in the majors) and they are on top everywhere. Daric Barton may not have won a gold glove (you kind of have to be a Yankee to win one) but he ran away with the fielding bible award. Ellis may have been the best defensive second baseman over the last ten years. So they shine on the right side and are above average on the left. In fact the only A's regular to have below average defense by UZR or +/- was Rajai Davis.



That said, yes, defense is really difficult to project and a ton of regression needs to be in the mix. And there is no reason why Cahill should get better defense than all his teammates. So I don't think you should think he'll have average defense, but I don't think you should assume he'll carry over last year's numbers either. I'm stuck in the middle with you.

Originally posted by KJ Duke:

Dough, from one projection guy to one non-projection guy, let me make a case.



Projections aren't made with the expectation of being right, they are made for 3 reasons:



(1) It forces the projector to delve into past results and other potentially relevant factors and to make presumptions about them which may be relevant to future results. So even if I didn't carry those projections with me to the draft table, they've given me something.



(2) A projection is designed to quantify a probability. If I believe there is a 50% chance Jose hits 45 HRs this season, and 50% he hits 25 HRs, and those odds are reasonable, Jose becomes a probable 35 HR hitter for me. Along with projecting his other stats, that allows me to put a value on Jose and compare him to alternatives and to build a balanced team based on probable outcomes - which, if the methodology is sound, should give my team a reasonable facsimile of what I am projecting aside from unexpected injury risk.



(3) And this is to me the critical part of doing projections ... If I develop projections for Jose I know what assumptions go into those projections. If, on the other hand, I am taking projections from Zola, Schandler or Rotoworld I don't know what assumptions are in there. At this point I not only lose insight behind the numbers, which can hugely affect the expected risk/variance of achieving those numbers (and who I want to draft for a particular team), I also lose the entire benefit of item #1 above.



While I can appreciate a seat-of-the-pants approach from an exceptionally well-read and close observer of baseball such as yourself, I can't imagine approaching a draft without doing what I do - which is making a lot of projections and knowing what has gone into those projections. We probably would agree on a couple things however; there is minimal value in relying on the projections of others, and if you draft based only on projections (even your own) you're not doing yourself any favors.



On another subject - the Schandler projection for Joba ... while you saw cajones, I saw a poorly-executed attempt at grandstanding. Ditto Mike Jacobs the following year. If you're going to grandstand, make damn sure you know what you're doing. I agree with others that this was stated really well and I'm glad I didn't try to do it first because I'd have messed it up, but the sentiments were identical to my own.



You know, it also gives you an exact number. Doughy is fond of saying he doesn't make projections, but of course he does; we all do. Otherwise we are just throwing darts at a board. It's just that his projection is 'Albert Pujols: Great power, great average, some speed,' while mine might be 40/.320/12, but they're both projections.



The benefit of having an exact number is that I have something I can add up as I am going through the draft. I don't know how to add up things like Great power, some power, and no power for three guys. I might project 40, 20, and 2 for those three guys and I don't expect that those are going to be perfectly accurate, but 62 is something I can work with. Dan can just sort of feel that out. That is an awesome talent. I just don't have it and as long as I get to use my crutch I'll use it. There, not stated nearly as eloquent as KJ.

2)Bah! 50% one way and 50% the other means either 'I don't know or stuck in the middle with you. 35 home runs sounds reasonable to EVERYBODY. More or less, it is 'office cleanup' meaning little. In other words, Bautista has to be one of those players you're sold on or not. If yes, 40's. If no, 20's. I made a $20 bet after my post with a fella. He has any number in the 30's, I have the rest.



I'm not sure you are getting his point so let me try another way. Yes! 50% one way and 50% the other means I don't know. In fact this is not specific to Bautista; it is true of every player. Granted he has a wider variety of outcomes than others, but the thing is when I'm drafting him I don't even care if he hits 25 or 45. Why? Because if I have good mean (average outcome) projections I will hit on some other players. It is not that you have to be right about everyone; it is that you draft 14 hitters and if you are right on 7 and wrong on 7, you still come out with the team you expected. It is about being stuck in the middle, but that's exactly where I want to be.



Back to Bautista, if I put him down for 25 I'll miss on drafting him. At 45 I will plan a draft that could be short on power. If I truly believe in both outcomes, splitting the difference is the right thing to do.



[ February 15, 2011, 11:45 AM: Message edited by: bjoak ]

One final point about making your own projections. You don't have to regress to the mean. A good projection engine treats everyone equally and bad projections are too subjective, but on your own you can treat special circumstances differently.



I don't have my notes with me and forget which pitcher I am thinking of but just yesterday I was working on someone who had declining K rates for four years. Normally I would expect him to get something back (though K rates do stabilize quickly), but I also noticed that he had declining velocity for those same four years. An engine expects a bump in that guy's rate, but I chose an even lower K rate for him.



If I had stronger evidence about Bautista, I'd adjust him accordingly.

Originally posted by DOUGHBOYS:

By the way, Jim, I would love to be in a league without projections from these guys. In fact we had an e-mail draft that started as soon as the last out was made last season. I would have been quite comfortable drafting against the other 14 even if these writers were in their ear, giving them their projections for the coming year.



Now, let me propose something to you. Would you go into a paperless, computerless draft? These really test mettle. We had a local league that did it for a couple of years, but proved to tough to find replacements for.



Rules were, that no paper, computers, writing instruments, and hands were checked for 'crib notes'. The only one who could write was the director who wrote down names as they were called.

It was only a $20 league, but paid for itself in preparation. The opposite of a 'projections' league. I would not do well in a paperless draft. My point was I'd like to be the only one in a draft allowed to use projections from HQ etc. For those who feel that those projections are worthless, it shouldn't give me any advantage being the only one to use them.