As far as BABIP, players with lots of power and/or speed often carry higher BABIP's than the normal hitter correct? Howard's career BABIP is .333, Pujols' is .321, Cust's is .331, Ramirez's is .344...I'm sure there are more examples and plenty that disprove it but hitters with lots of power and/or speed can sustain high BABIPs.


I said high BABIPs, not higher than normal! Snider's BABIPs have been in the .390-.420 range, which is completely unmaintainable. There are a few hitters that can maintain BABIPs in the .340-.350 range and as you go higher, the number of players that can maintain drops off like the actuarial table of 110-year-olds.

I need to ask Sean or Nate if I'm the only one that can't seem to communicate this after years and years of doing this or if it's just because my projections tend to be tethered to a more discussion-based area.

It's the forum, and the participants. We're an argumentative group of bastards, and a good chunk of us are so eager to argue that we often don't finish reading that to which we're planning to respond.

BABIP is always higher in the minors though. What are you estimating Snyder's BABIP to be with that projection? I don't think it's too unrealistic to think he could easily keep a .320-.330 BABIP.

As far as projecting young players, is that rate any higher with high profile players? Heyward is right up there with Bruce/Upton in expectations, he wasn't quite on their level entering the draft but his performance in the minors have raised expectations. Not all rookies are the same and I don't think you should treat them the same...but that argument leads us down the road we have traveled too often so I'll leave it with that I guess.

In Lind's case, ZiPS has Lind retaining about 60% of his gain. Young players that take big step forwards without taking steps back are the exceptions, not the rule, and if you ignore this fact, then overrating every single young player having a good season and underrating every single young player having a bad season will be a built-in feature of your projection system.

BABIP is always higher in the minors though. What are you estimating Snyder's BABIP to be with that projection? I don't think it's too unrealistic to think he could easily keep a .320-.330 BABIP.

Using the numbers available, Snider is expected to get 78 non-HR (99 - 21) hits in 264 (429 AB - 144 K - 21 HR) Ball-in-play ABs, for a BABIP of about 0.295.

A .295 seems shockingly low if that is accurate. It's below average even, that just doesn't make much sense to me.

Again I ask, is the rate different for high profile prospects? Lind/Snider were both highly rated prospects. I think it's safer to project them (like other top prospects) to have less regression than normal.

A .295 seems shockingly low.

If you bump it to 0.330, then he picks up an extra 9 hits, of which about 2/3 will be singles. Doing a quick estimate, that would bump his line up to a 0.252/0.337/0.464, which would probably be about a 110-115 OPS+.

I don't think there's a subjective element in ZIPs for Dan to tell it if a prospect is high-profile. It just sees the performance record. Which strikes me as reasonable, actually. High-profile prospects are still humans, and would be a part of the performance pattern Dan's describing.

Actually, I better check Snider now - .295 does seem lower than it should be.

I don't think there's a subjective element in ZIPs for Dan to tell it if a prospect is high-profile. It just sees the performance record. Which strikes me as reasonable, actually. High-profile prospects are still humans, and would be a part of the performance pattern Dan's describing.

Sure they are humans but people whose job it is to assess the likelihood of players' success think the odds are greater for them. I don't see how it is reasonable to treat all rookies the same. Jason Heyward has a far FAR FAR greater chance of being successful out of the gate than say...Jeremy Slayden. If both of them debut this year and have very good seasons, how is it reasonable to regress Heyward just as much as Slayden?

. I don't see how it is reasonable to treat all rookies the same.

But ZIPs doesn't do that. It sees minor league performance as well -- it's not comparing Jason Heyward to Jeremy Slayden, it's comparing him to other 20-year-old outfielders who ripped through the high minors as teenagers. Even star prospects regress (or progress) towards a mean. When Dan ran those numbers for you, the stars are included in it -- in fact, there are plenty of stars who broke out, fell back, and broke out further. The point is that even great prospects aren't usually like Albert Pujols -- they're usually more like Brian McCann (143 OPS+ collapsed by 40 points before he built back to previous levels), Matt Kemp (127 OPS+ followed by 110 before he returned to the mid-120s), or Joe Mauer (OPS+ at 23: 144; following year: 118).

Sure the stars are included, my point is players with far less ability are also included which skews the numbers. I'm saying top prospects are probably less likely to regress 65% so they shouldn't be treated the same as someone who was thought to be roster filler. Nowhere am I suggesting they shouldn't regress (even though I think expecting all young players to regress is a poor way of handling it), I'm saying they shouldn't regress as much.

Edit: Changed 68% to 65%, misread above but not much of a change. Also Dan says under 25 so he isn't comparing Heyward to just 20 year olds who reached the majors. Unless I'm reading it wrong, he lumped everyone who was 25 or younger in the same group.

I'm saying top prospects are probably less likely to regress 68% so they shouldn't be treated the same as someone who was thought to be roster filler.

Why are they less likely to regress? They're probably just as likely to regress, but the stars are regressing from a much higher starting point - for true young stars, even if they do regress and lose about 32% of their gains (and not 68%) from the previous year, they're still valuable performers. For those non-stars, however, that loss of the 32% can turn them from contributors to non-contributors.

Edit: Changed 68% to 65%, misread above but not much of a change. Also Dan says under 25 so he isn't comparing Heyward to just 20 year olds who reached the majors. Unless I'm reading it wrong, he lumped everyone who was 25 or younger in the same group.

No, ZiPS has a detailed model of regression, but illustration is frequently a more useful technique for explaining things than describing a fairly rigorous model.

There was a Snider problem. Not sure how it happened (I re-checked all the Blue Jay hitters and everyone else was fine). Rogers is a poor park for $H and players like Snider are the worst bets to improve, but .295 was too low.

You don't think top prospects are less likely to regress as much as players believed to be roster filler or organizational depth? Their skills are thought to be superior, that is why.

Definitely misread above, 34 and 65 is a big difference but it doesn't change my point. To expect say Adam Lind or Travis Snider to regress 34% just because he had a good year last year and is 25 or under seems wrong. They should only be compared to players that fit their profile, their performance, their pedigree.

As usual, this is a losing battle. I continue to be amazed the community here genuinely feels using MLB trends including all players can be applied the same to every player instead of trying to use similar players/groups. Take Justin Upton, he isn't just your average under 25 year old OF'er so why treat him so?

No, ZiPS has a detailed model of regression, but illustration is frequently a more useful technique for explaining things than describing a fairly rigorous model.

So assume (for arguments sake) Heyward debuts this year and puts up a 125 OPS+. Next year when ZiPS goes to put out a projection for Hewyard, what will be the factors? Heyward's minor league performance, rookie season performance, park/league adjustments, and only 20 year old OF'ers?

They should only be compared to players that fit their profile, their performance, their pedigree.

Which, in fact, is the case. Profile and performance are indicated. The 65% is simply an illustration about baseball history, not something that's contained in a model.

As for pedigree, I haven't found much in the way of worth regarding regression (I checked using variables for both BA prospect position and draft pick # as inputs for estimating regression and I found them to be worthless).

As for pedigree, I haven't found much in the way of worth regarding regression (I checked using variables for both BA prospect position and draft pick # as inputs for estimating regression and I found them to be worthless).

So ZiPS thinks a top 10 selection is likely to regress the same amount as a player taking somewhere between the 40th-50th rounds?

So ZiPS thinks a top 10 selection is likely to regress the same amount as a player taking in the 40th-50th rounds?

If the two players play exactly the same and have the exact same performance history, yes.

Don't like it? Your complaint is essentially that reality is inconsistent with reality.

#117 Here's the thing. It's "seems" wrong to you. Dan on the other hand has checked. Why not run your own systematic study?

EDIT: To be clear, there's a very real chance that the projections for Snyder (or any given young stud) will be wrong. And the spectacular misses rate to be low projections.

As for Jason Heyward, if he puts up Nate McLouth's 2008 season (a 125 OPS+), his age and profile are strong enough that he then would project to a 127 OPS+ for 2011.

What would your projection be for a 26 year-old who hasn't played baseball since the rec league travel squad (so age 14, let's say)? Keep in mind, back then I was an OBP machine. Like .300/.450/.450.

I will go to fangraphs to ask for a playing time projection, in keeping with Tango's latest findings.

Probably because a top 10 selection is likely to regress.

Just because ZiPS thinks it doesn't mean it is reality. How many picks taking between the 40th and 50th rounds even reach the majors? How many have had 1 above average season? How many have had 5 above averages seasons? How many have had 10 above average seasons? You are telling me that in reality, there is absolutely no distinction between the success rate of a the 1st pick and the last pick of the draft?

Well it's probably going to sound selfish but I don't have the time to start building a database of every player that reached the majors this afternoon and get it done in time to provide evidence within the next 20 years. I'm assuming Dan has this stuff readily available if ZiPS does in fact know the difference between the success rate and rate of regression for top 10 picks and extremely late draft selections.

Probably because a top 10 selection is likely to regress.

Sure but not likely to regress as much as someone with a much lower profile.

Anyway, I've spent too much time reading this thread as it is so better go accomplish something this afternoon. I appreciate what answers I did get but it still doesn't sound as realistic as you all keep saying. I know top 10 picks have a high attrition rate but it can't be the same as someone taking between the 20th-29th rounds, 30th-39th rounds, and nowhere remotely close to someone taking between the 40th-49th rounds.

But how is a statement like that at all useful when talking about a player as young and inexperienced (at the MLB level) as Travis Snider?
It goes without saying that almost every MLB player out there had to improve significantly from their 22 year old talent level to become good MLB players. It certainly isn't very useful as far as projecting anything. You won't project him to be a good player until he becomes a good player?

Depends. I have been suggesting for some years now that high-K prospects tend not to develop all that well. More specifically, I have been "under-estimating" high-K prospects who have had "breakout years" based on unsustainable on-contact numbers. It's not rocket science, it's simply a matter of pointing out how unlikely it is that a player hits like Babe Ruth.

The numbers above project Snider to K 1 per 3 AB. It is very, very difficult to be a quality ML hitter while doing that. The numbers above project Snider to hit about 350/650 on-contact. That seems reasonable although maybe, after checking, Dan will upgrade that. But, for example, those are much better on-contact numbers than Gary Sheffield's career numbers, they are roughly equivalent to Alfonso Soriano's career numbers. Nothing embarrassing about that.

But that's the rub. Alfonso Soriano has a 113 career OPS+ while hitting 350/650 on-contact ... and K'ing about 1 per 4.5 AB. If Snider maintains his on-contact performance, he has to cut his K-rate from 1 per 3 to 1 per 4.5 -- not impossible but that's a big change. And that's to turn out as an average-hitting corner OF. Alternatively, as I pointed out before, he could maintain his 1 per 3 K-rate but greatly improve his on-contact (which he might be doing as we speak as Dan re-runs the numbers) to, say, 380/720 and he'd be an average-hitting corner player. To become "good" or a "star" he needs to do both of those things. I put forward the claim that it is unlikely that any young player will greatly improve both his K-rate and his on-contact production on a long-term basis. Dan or someone else can test this hypothesis (as much grief as I give Dan, ZiPS and I tend to be in agreement on such players.)

Now, some will. But none of them should be _expected_ to until they have actually done so. (With, yes, the caveat that scouting reports or o-swing %s or hiring hitting gurus or draft position or whatever may have some predictive value over and above past performance ... something I also won't believe until someone shows that they do.) It's also possible that Snider could develop the awesome power of Howard or Pena but I can't see any reason to _expect_ him to do that either.

That said, on closer inspection (which, fair enough, I should have done earlier), Snider's projection seems a little low. ZiPS is projecting improvement in HR-rate (from about 1 per 29 to 1 per 20) but essentially constant on-contact BA -- essentially turning singles and doubles into HRs and killing the BABIP. Maybe that makes sense but it seems a little odd. And yeah, if you add 20 points of BA (and thereby SLG), you get Lyle Overbay with an extra 40 points of SLG or, roughly, a 110 OPS+.

Just because ZiPS thinks it doesn't mean it is reality. How many picks taking between the 40th and 50th rounds even reach the majors? How many have had 1 above average season? How many have had 5 above averages seasons? How many have had 10 above average seasons? You are telling me that in reality, there is absolutely no distinction between the success rate of a the 1st pick and the last pick of the draft?

Not if they put up the same stats. Why should there be?

If two players are drafted #1 and #1000, and proceed to put up exactly the same stats. Say, 150 OPS+ in Rookie, A, AA, and AAA, and they both debut in MLB at the same age, and put up a 100 OPS+, why would you project them differently based on how they were evealuated 4 or 5 years ago?

Just because ZiPS thinks it doesn't mean it is reality.

No, it's reality is why ZiPS thinks it's reality. To this date, there is no relationship between regression and draft position. There is a relationship between quality and draft position, but quality and regression are very different things.

How many picks taking between the 40th and 50th rounds even reach the majors?

- Not many, but there are a lot of players between the 5th round and the 40th-to-50th rounds.

- Regression isn't something that manifests itself solely at the major league level.

Since you're paying attention Dan, are you leaving BTF?

The fixed Snider is up.

Since you're paying attention Dan, are you leaving BTF?

No, why would I?

Also, I strongly urge META discussions to be taken to the lounge or elsewhere.

You are telling me that in reality, there is absolutely no distinction between the success rate of a the 1st pick and the last pick of the draft?

Man, oh man. You really have no concept what you are talking about. You simply have no idea what these models are doing. You don't grasp the basic concepts. You might as well go onto physics boards and argue against string theory.

Do 1st round picks have a better chance of making the majors than 25th round picks? Of course. What's your point?

What Dan is saying is that if you take the population of 1st round picks projected to put up a 110 OPS+ at age 22 and the population of 25th round picks projected to put up a 110 OPS+ at age 22, you will see the same distribution of performance.

You do understand that the above statement in no way suggests that the probability of a 25th round pick being projected to put up a 110 OPS+ at age 22 is the same (or even remotely similar to) the probability that a 1st round pick will be projected to put up a 110 OPS+ at age 22.

Furthermore, Dan is saying that young players retain most but not all of their "improvement" from the previous year (similarly they retain only some of their poor performance) and that this holds regardless of, say, draft position. All that means is that if a young player improves by 20 points of OPS+, he can expect to retain 14 points of it. If a Justin Upton jumps from a 110 to a 130 OPS+, he might project to a 124. If a borderline player jumps from an 80 OPS+ to a 100 OPS+, he might project to a 94. I don't see any reason to think that's strange. I don't see why that's inconsistent with scouting reports or draft order position or BA rankings or whatever.

You seem to somehow confuse this with the notion that Justin Upton and the scrub will get the same projection but that's obviously not the case. ZiPS doesn't treat all young players the same. It treats all 22-year-olds the same after controlling for past performance (and body type and some other things). Dan is telling you that adding draft order or BA rankings or whatever to the model doesn't improve the prediction -- i.e. AFTER CONTROLLING FOR PAST PERFORMANCE, ETC. draft order has no independent effect. Why is that hard to believe?

Perhaps you'll understand it better if put in these terms. On the one hand, you would scoff if Dan had projected Pujols at age 22 to put up a 125 OPS+. On the other hand, you would scoff if Dan projected a 13th round pick (Pujols) who had a great season at age 21 to regress less than a 1st round pick who did so. In short, you want ZiPS both to project Pujols to develop much better than BJ Upton and much worse than BJ Upton. You can't have it both ways.

You also clearly don't understand how models work. There is great power in having a large sample. Reducing comparisons to only "very similar players" is going to give you tiny sample sizes and blow your variance up astronomically. Even with the large samples, the variance on a prediction is already extremely high, increasing it would make any such projection useless.

Models take advantage of both the broad population trends and the specific group trends as best they can. They do so by controlling for individual characteristics (primarily weighted past performance and age) which (informally speaking) define the sub-populations of interest while also using the power of the large sample of baseball history to improve the precision of the estimates and to avoid "wild" over/under-estimates.

And what Dan is saying is that after you have used the individual-level variables such as age and past performance to (informally speaking) define a group of similar players, then sub-dividing that group by draft order does not help you predict future performance (well, next year's performance). I don't find that difficult to believe whatsoever.

Does that make scouts wrong? Does that mean draft order doesn't matter? Of course not. The group of 22-year-olds with a projection similar to Heyward is surely largely a group of high draft picks (and international signings which create problems with using draft order in the model). All Dan is saying is that if a player has developed into a guy who projects to a 110 OPS+ at age 22 then it doesn't matter when he was drafted and he has emprical evidence this is the case. If a team hit the jackpot with Pujols, a team hit the jackpot with Pujols. That doesn't mean the expected outcome of a 13th round pick is Pujols.

In fact, ZiPS seems to be saying pretty much just what you want it to say -- "if there's a player who's torn up the minors and torn up the majors at a young age, you'd better pay attention."

And I'm sorry jfish but you seem incapable (or unwilling) to understand what's being claimed and what's not being claimed. Until you can get that straight, it's difficult to have a discussion with you.

There was a Snider problem.

I tried to tell you.....

The fixed Snider is up.

This makes much more sense. And if I am right in my speculations in #89, he might still beat the revised projection, assuming he can get that K Rate under 30%. I am far less inclined to take the over now that the projection has been corrected.

As for pedigree, I haven't found much in the way of worth regarding regression (I checked using variables for both BA prospect position and draft pick # as inputs for estimating regression and I found them to be worthless).

This is really surprising to me, as it essentially says that BA's scouting output is worthless. It's especially surprising for pitchers, given how much is made about junkballers with great stats in the low minors having little chance once they move up.

Are your findings also true for players with very limited pro experience?

The fixed Snider is up.

That looks a lot more realistic. Thanks.

For Dan and Walt, continuing on the discussion from upthread, I'd be curious to see if a decent walk rate (somewhat) mitigates concerns about a high strikeout hitter. Walt brought up Soriano, who has a much lower walk rate. With hitter like Upton and Snider, who have decent-but-not-great walk rates, it seems like they're striking out sometimes as a result of at least attempting to work the count, rather than swinging at anything in their zipcode.

In terms of projecting growth based on reputation WTF happened to Rios last year? Was anyone expecting him to completely crater like this?

You might as well go onto physics boards and argue against string theory.

From my understanding, he'd actually get a lot more support if he went on a physics board and argued against string theory.

This is really surprising to me, as it essentially says that BA's scouting output is worthless. It's especially surprising for pitchers, given how much is made about junkballers with great stats in the low minors having little chance once they move up.


Again, we're talking regression here, not ability.

BA does an extremely good job at predicting low minor performance of amateurs - I've attempted to translate college stats and draft pick position is far more useful.

Also, I'm mostly talking about MLB/AAA/AA/A+ performance. I rarely project mid-A and lower, so have not checked for regression characteristics at those levels.

as much grief as I give Dan, ZiPS and I tend to be in agreement on such players.)

I'm very disappointed your BABIP radar didn't pick up on Snider being way too low!

This is really surprising to me, as it essentially says that BA's scouting output is worthless.

No it doesn't. It says that great young studs regress just like lousy little weiner boys do.

Fist Powers, 1B, Manhattan Rich Giants, drafted round 1 pick 1, Eastern League God among Men, OPS+(1) = 130; projected OPS+(2) = 123

Limpy Laxative, Beaverton Sad Cats, OPS+1, drafted round 50 supplemental, pick 30, crappy minor league records, OPS+(1) = 80; projected OPS+(2) = 70

The studs just start from a higher point.

From my understanding, he'd actually get a lot more support if he went on a physics board and argued against string theory.

I thought that too, but I think the suggestion is that he wouldn't understand it, regardless of its own veracity.

For Dan and Walt, continuing on the discussion from upthread, I'd be curious to see if a decent walk rate (somewhat) mitigates concerns about a high strikeout hitter.

In general, not really. Bill James's observation of old and young player skills actually holds up quite well to rigorous examination in general.

I thought that too, but I think the suggestion is that he wouldn't understand it, regardless of its own veracity.

"You don't understand string theory!" is an extremely weak insult unless you're in a room of physicists.

Again, we're talking regression here, not ability.

BA does an extremely good job at predicting low minor performance of amateurs - I've attempted to translate college stats and draft pick position is far more useful.

Also, I'm mostly talking about MLB/AAA/AA/A+ performance. I rarely project mid-A and lower, so have not checked for regression characteristics at those levels.

I understand we're talking about regression. I'm surprised that knowing Ben Hornbeck's (lack of) prospect status is worthless in projecting him going forward. If he put up those same numbers with a mid-90s fastball and a top 50 BA prospect ranking, most observers would be a lot more optimistic about his future than they are now.

I'm not saying your findings are wrong, of course. I'm just saying I'm surprised by them.

Or am I misunderstanding something?

String theory.

Danny, I think part of what's going on is that we've gotten better at discerning what in a statistical profile suggests future success. Those junkballers might not have the super high k/9 for one, etc... so Dan is already accounting for some of the same attributes (albeit in a proxied way) as do the scouts.
edit: not talking about hornbeck specifically, obviously.

Bocock related news

While there are always some junkballers that put up good numbers in the minors, using multiple years tends to weed out a lot of them as their big years tend to be spottier and their $H more erratic than average (I check out a lot of things).

That's awesome! I'll go post that right nwo.

While there are always some junkballers that put up good numbers in the minors, using multiple years tends to weed out a lot of them as their big years tend to be spottier and their $H more erratic than average (I check out a lot of things).

So is there a big difference in the ability of major league caliber pitchers in the minors to prevent H on BIP vs. minor league caliber guys?

So is there a big difference in the ability of major league caliber pitchers in the minors to prevent H on BIP vs. minor league caliber guys?

$H regresses a mega-shitload in the majors and a mere shitload in the minors.

Oops, I have a new word to nanny.

Oops, I have a new word to nanny.

You'll also want to block off shitty while you're at it.

And Francoeur.

I would think ability and regression should have some negative correlation even if it is weak. Just like I would think ability and success would have a positive correlation. The more ability you have the better odds of success and improvement to a point until you level off and eventually begin to decline or get injured. That just makes sense when you say it, especially for young players who are improving their craft. Every player that gets to the majors has amazing ability but there is still a range and MLB clubs spend millions upon millions to discern the difference between players that can and can't be successful at the MLB level. Every player that reaches the majors will have varying levels of success but the ones with the most ability should produce more and regress less than players of lesser ability. Prospecting/scouting isn't a science but I continually get the feeling this community doesn't think it matters at all. Not to mention coaches sometimes have players working on specific skills like using their CB or change more, focusing on hitting the other way, etc. We have already went over changes in hitting/pitching mechanics, no need to get into that again though.

Even if a 1st round pick and a 40th round pick have similar minor league stats, similar park/league factors, and are the same age, yes I would still have more confidence in the 1st round pick. Even if they reach the majors at the same time and put up similar stats, I'd have more confidence in the 1st rounder to improve moving forward and to regress less than the 40th round pick. With every jump in level going all the way back from tee-ball to little league, a certain percentage of players can't adjust to the new talent level. Unless of course you feel the talent level in MLB is no different than the minor leagues? Many of the players that can't keep up at the next level were very successful until that point. You see this often with pitching, a great FB can get you through AA, sometimes AAA and perhaps a season or two in the majors but the superior talent around him should eventually prevail. All of those AAAA players run into that many of which were pretty successful until they reached the majors.

Perhaps you are right Walt. Perhaps I don't understand what a model is or what ZiPS is trying to do. To me, it doesn't seem to be trying to project future performance because I don't think numbers are the only thing worth considering especially when talking about players that are this young. Maybe I'm wrong though. Perhaps ability doesn't matter, only results, regardless of how you get them. Unfortunately, I have a problem believing that so maybe this model isn't for me. Every team that is released has numbers that make no sense to me and I find it odd that you've said before you disagree with ZiPS often but seem to think it does its job...which I thought was to accurately project future performance? I'll never understand how you can dismiss important factors when projecting players future performance but I'll just stop posting so you don't have to put up with my stupidity.

Again, I appreciate the answers and explanation, I'm just trying to figure out exactly how ZiPS gets the numbers it does. I see some comments from Dan that he has or is trying to quantify things that previously haven't been so perhaps in the future I'll feel better about the ZiPS model. Obviously no projection system will be perfect but the vocal minority of this community seem to share the belief that ZiPS couldn't do anything to improve.

Why not just accept that, if a guy is drafted in the 40th round but puts up numbers as good as the best first rounder, the logical conclusion is that the scouts got the 40th round guy wrong?

Does the email function (from user's profile) work on here...does Dan's?

Obviously no projection system will be perfect but the vocal minority of this community seem to share the belief that ZiPS couldn't do anything to improve.

We don't think that it can't be improved. What we do think is that the sort of things that you've suggested (stuff like limited scouting opinions from a very limited subset of scouts on a limited subset of players) won't actually act to improve it across a sufficient number of players, since the depth and uniformity of the type of information you're suggesting isn't available.

Does the email function (from user's profile) work on here...does Dan's?

Yes to both questions.

Even if a 1st round pick and a 40th round pick have similar minor league stats, similar park/league factors, and are the same age, yes I would still have more confidence in the 1st round pick.

With High school (and even the college guys I suppose) picks still growing and changing, most with nothing more than tools that need to develop...I would think if they go toe-toe in performance for multiple years, they would have to be treated the same eventually...especially through multiple levels. Not that those tools and pedigree don't matter, but until they show themselves to be more than the 'lesser' prospect, why treat them differently, and how would you even go about doing it?

But I do understand having more confidence in that player to continue a solid performance over long period of time, although to randomly project that jump or the other player's regression at a larger rate would feel unsupported for now.

The revised ZiPS projection for Snider is entirely reasonable. Dan has him with an IsoP over .200. If Snider turns into an offensive star, it'll probably be along the lines of .270, 40 homers, 90 walks, 160 strikeouts. The age 22 line is perfectly consistent with that.

I'll take the over on the Aaron Hill projection. What ZiPS does not know is that Hill's 2008 comes with a big asterisk. He played just the first quarter of the season under the tutelage of Gary Denbo. The whole offence tanked with Denbo, but unlike the rest of the players, Hill did not get a chance to redeem himself the rest of the season due to a David Eckstein elbow. Subjectively, I would use the mid-point of the 2007/2009 numbers to describe where he is, but I expect that pitchers are going to get cute after his unusual home run outing last year, and that Hill will adjust successfully. Hence, higher BA, more walks, lower IsoP.

Hill is 28 with a career OBP of .337. A projected OBP 14 points below career for a 28 year old does seem unusual.

I can't see what Snider's old proj looked like - it's gone from 99 OPS+ to 107?
Will the ODDIBE be updated? What are his chances (re post#92) that he'll post a 100 OPS+, now over 50% chance? Chances of 120+?
Thanks for the revise.

Oh, and porting Brandon Morrow's proj 107 ERA+ to the top of that pitching staff helps. Jumping on Dana Eveland this week (costing a roster spot too) shows how thin they are.

I would think ability and regression should have some negative correlation even if it is weak. Just like I would think ability and success would have a positive correlation.

While it'd be a lot of fun (ok, maybe an abnormal definition of fun) to regress based on smaller and smaller groupings, at some point sooner rather than later, I'd think you'd get diminishing returns, especially if you shrink the groups further. As an example, right handed pitchers drafted from high school in the first round in the past 20 years include Todd Van Poppel, Josh Beckett, and Phil Hughes.

As an example, right handed pitchers drafted from high school in the first round in the past 20 years

Roy Halladay & Matt ####### Bush

Just like I would think ability and success would have a positive correlation.

Why not just accept that, if a guy is drafted in the 40th round but puts up numbers as good as the best first rounder, the logical conclusion is that the scouts got the 40th round guy wrong?

What T&B;said.

Why wouldn't you think that if the the 40th round guy has put up similar performance through 3 or 4 levels of mLB and then in MLB that he has the same ability as the 1st round guy?

If this "ability advantage" the scouts saw doesn't translate into anything in professional baseball performance, isn't it far more likely that the gap wasn't there, than that it's hibernating somewhere ready to impact the age 24 seasons?

in 1994 would you still have been downgrading Mike Piazza for being the 2000th pick?

Why wouldn't you think that if the the 40th round guy has put up similar performance through 3 or 4 levels of mLB and then in MLB that he has the same ability as the 1st round guy?

Because Mike Piazza never existed.

Because Mike Piazza never existed.

Or Keith Hernandez, or Don Mattingly.

Of course, Matt ####### Bush was a shortstop at the time.

*****************************************************

Dear jfish...

Prospecting/scouting isn't a science but I continually get the feeling this community doesn't think it matters at all.
Not remotely true. Many of the names in this thread I see at Sickels' site or commenting at BA. Articles on stuff like PitchFx are (imo) very well received here, by and large. Dan mentioned that he looked into integrating BA rankings here - only that it wasn't significant. Heck, one of the site's contributors left to become a MLB scout.
I'm not sure why your equating "doesn't matter for this purpose" with "doesn't matter". Very, very different.

Sidenote: Even if I accepted your premise, I'd be careful with the 1st round / 40th round stuff. You'd be better off, I think, using bonus as a proxy for the talent/ceiling (where available).

every jump in level going all the way back from tee-ball to little league, a certain percentage of players can't adjust to the new talent level.
It really isn't that bi-modal at high levels (thinking AA on) - it's really more whether you're clearly good enough or whether you get the right fit. There's no evidence, for example, of a true AAAA hitter - the people that we put that label on have either been at the wrong place at the wrong time or, more often, have some other issue (bad d, for example) that keeps them from success at the next level. [That statement is complicated by how you react to the data issues associated with cup of coffee / failed prospect types. Also, I'm not as unequivocal with regard to pitching, largely because of the starter/reliever split, but I digress.]

Unless of course you feel the talent level in MLB is no different than the minor leagues?
This is, of course, simply stupid.

Obviously no projection system will be perfect but the vocal minority of this community seem to share the belief that ZiPS couldn't do anything to improve.
Imo, ZiPS has changed fairly dramatically since I was first exposed to it (enough that I discontinued my own model making as his clearly surpassed my work). I presume it will continue to do so. Furthermore - Dan has made generic versions available in the past. If you want to test alterations or add components - please do so! But the burden of proof that your ideas matter will fall on you - if you hadn't noticed, Dan is believed to be pretty good at this sort of thing - and, right now, your arguments aren't that persuasive.

Playing devil's (jfish's) advocate,

Could BA rank / draft position / bonus still affect the "optimistic" or upper-percentile projections? I guess once you have an adequate sample of data, it probably can't make much more than a marginal difference anyway, but "ceiling" is perhaps the one area it could affect. A 40th rounder, with the exact same age and pro stats as Snider, might be perceived as having a lower ceiling than Snider -- "more 1st rounders make the hall of fame than 40th rounders." But then again, that's probably just a bunk perception -- not many 1st rounders make the hall of fame anyway -- and one of the valuable things about ZiPS is that is can sort through the numbers and point those things out to us (particularly when league/park/age factors muddy the waters).

I think jfish's primary confusion is over what a projection system like ZiPS is supposed to do. It's not supposed to identify breakout/collapse candidates in the traditional sense -- it's supposed to detect patterns, particularly patterns below the surface of the numbers, and tell us likely outcomes based on those patterns. That may not seem very useful or sexy to the casual fan, but it is really the only way to systematically project a large group of players with any degree of accuracy. That's not to say that observation and scouting aren't valuable too, and maybe with the right pair of eyes you can better identify breakout/collapse candidates, but it's just not practical to apply that to a large group and expect accurate results over a long period of time.

Hate to post this in the Blue Jays one, but doesnt look like Dan goes way back to the RedSox. Was wondering if Ryan Kalish was going to get a projection in the final disk. Also wondering if there was any chance that Casey Kelly or Jose Iglesias get one.

I go through all requests after the main chunk of teams are done. I don't imagine, in this case, that Iglesias or Kelly are good candidates for projections. I imagine Iglesias will receive a projection next year and Kelly if he has a better season above the Sally.

Thanks for responding to that Dan. I fully understand about Kelly and Westmoreland, just figured that Iglesias will definitely see time in AA and maybe AAA by then end of the year and would get a projection. No biggy.
Looking forward to the last two teams.
Keep up the good work.

Ordinarily, Iglesias would be high enough to get him a projection, but we have limited data on Cuban play and projections aren't really good in situations where there's not much to go on. If Iglesias was a sure-thing to be in the majors, I'd have to bite the bullet, but since he's not, I don't have to yet (like I had to with Alexei Ramirez a few years ago). I may have to already with Chapman.

So, the only Jays projected above a 100 OPS+ are either injured or currently blowing goats. The pitching staff also isn't at all 'thin.' And this team's kinda not at all boring. Oh random and fickle fates, how dearly we loved ye.

So I'm wondering what the ODDIBE was of Bautista hitting 40 bombs...

Those Hill and Lind projections would be nice.
Meanwhile Overbay's done a good job of working his way back up to surpass his ZiPs!

I'm guessing we do this thread all over again in February, with Jose Bautista playing the role of Ben Zobrist?
(Who by the way is sporting a nice .252/.358/.354 OPS+ 94)

Considering the level of competition this might be the most surprising team in baseball to me. I thought they had a good chance to be the worst team in baseball. Is there any talk about who Cito's replacement will be? It looks like he is inheriting a much better situation than was expected.

Considering the level of competition this might be the most surprising team in baseball to me. I thought they had a good chance to be the worst team in baseball.

It has been a wonderful season to follow, since I also expected them to be among the worst teams in baseball. Instead, they've managed to put together a solid pitching staff, and an incredibly entertaining all-or-nothing offense.

Is there any talk about who Cito's replacement will be? It looks like he is inheriting a much better situation than was expected.

There was an article in the Toronto Star a while ago about possible replacements, but it was basically a list of a dozen of the standard candidates - every high profile former coach looking for work, mixed with the bench/3B coach of every succesful team, and a couple guys doing well as AAA managers.

I suspect Zobrist may be playing hurt; he was having a fine year until he tweaked his back at the end of July. Since then he's been drawing walks, but his BA is .178, and his slugging is .250.

Considering the level of competition this might be the most surprising team in baseball to me.

Me too. If you told me ahead of time that none of Hill, Lind, or Snider would be all that good and they'd still be above .500, I would say something with a hint of snark.

I suspect Zobrist may be playing hurt; he was having a fine year until he tweaked his back at the end of July. Since then he's been drawing walks, but his BA is .178, and his slugging is .250.

Though his 1st half numbers, while solid, don't match even his ZiPs projection.

EDIT: Perhaps that's too strong, his OBP was quite good, but with the lack of power his OPS was lower than the projection.

I just reread this whole thread and I have to tell you that it's a rare and strange feeling to be the guy who was closest to having predicted something correctly.

In re-reading it I was confused by the 1st comment, made in February of 2010

You're missing Yunel Escobar's offensive projection.

Am I missing something? Or is Teal and Black a time traveler?

That's a funny coincidence.

As I recall, Dan did the Projections in reverse order by Team name, which meant "Blue Jays" followed "Braves." Dan did indeed miss Yunel in his Braves projections, and T&B tried to get his attention by posting in the next ZiPS thread.

And now it looks alarmingly prescient.

So seven guys on this team managed to get to the 20-HR mark, and Sea Bass had 17 before being traded in June. Remarkable.

So next year, will someone be calling Dan a moron when Bautista is projected to hit "only" 27 bombs or something?

(Note: that person will not be yours truly)

I'm thinking 29 (will be the projection)

I think an over/under on that would be tough. Not sure.

I'm thinking 29 (will be the projection)

I think an over/under on that would be tough. Not sure.

29 would be a sensible number for a projection system to spit out. But I could see him getting to 35; he hasn't had the long HR droughts that would make you suspect his success isn't repeatable. He's tailed off a little this month and still hit 11.

So seven guys on this team managed to get to the 20-HR mark, and Sea Bass had 17 before being traded in June. Remarkable.

And Encarnacion got to 19 before being released (360 PA) and Snider is at 14 in 319 PA.

And Encarnacion got to 19 before being released (360 PA)

Encarnacion wasn't released. He hit his 20th earlier today.

Back when I was a kid, I remember reading that the 1961 Yankees were the only team in baseball history to have 6 different players hit at least 20 home runs.

I probably read that in...1996 or so.

Has another team topped that? Because if not, these 2010 Blue Jays just got their 7th (Bautista, Wells, Hill, Lind, Buck, Overbay, and now Encarnacion).

Has another team topped that? Because if not, these 2010 Blue Jays just got their 7th (Bautista, Wells, Hill, Lind, Buck, Overbay, and now Encarnacion).

The 2000 Blue Jays had 7. (My "methodology": clicking on a homerun-happy team in the power era).

I probably read that in...1996 or so.

Perhaps in an article about the '96 Orioles, who had 7 players hit 20+ with Baltimore and acquired 2 more who wound up with 20+ on the season.

They mentioned on today's broadcast that 7 ties the Major League record, set by those 2000 Jays.

It's a shame they didn't keep Gonzalez, just for that trivia....although Gonzalez+Escobar is 21.

Last year's Yankee team also had seven 20-homer hitters. On the Toronto broadcast, they named about five teams who predated this Jays team.

Beaverton Sad Cats

Mascot.