How likely is Swanson to rebound?

[conalthomas]

Swanson is having a terrible sophomore season. Over 378 PAs he has produced an OPS+ of 58, meaning 58% of league average.

How likely is a 22/23 year old to rebound and have a nice career after such a bad stretch?

Here is a list For single seasons, From 1997 to 2017, Age between 22-23, From 1st season to 2nd season, For any choice in Season Totals, (requiring PA >= 350 and onbase_plus_slugging_vs_split <= 60), along with each player's career WAR and peak season WAR:

Brandon Phillips (31.4, 4.7)
Ronny Cedeno (negative)
Jack Wilson (23.5, 4.8)
Mike Caruso (0.8, 2.7)
Ramon Santiago (7.9, 2.7)
Cesar Izturis (6.3, 3.8)

Perhaps unsurprisingly, all of those players were SSs, except BP who was a terrific defender at 2B.

Swanson is young enough that there is some time for him to rebound and be a Phillips/Wilson level player. Of course, there is a significant chance he ends up being more similar to the other 4 guys.

He still has the floor of a contributor like Izturis or Santiago, but let's hope he ends up being the next Jack Wilson or Brandon Phillips instead.

I'm sure I don't spend anything like as much time looking analyzing baseball stats as you do, but I don't see how this type of analysis tells us much of anything. A baseball player's future production has to be an extremely complex system with a huge number of variables including things like psychological profile and injuries (which are inherently unpredictable). I wouldn't be surprised at all to learn that it is a chaotic system (that is, slight differences in initial conditions can have very large effects since response is non-linear), so I'm curious, how good do projections of baseball players turn out to be? In this case, you have a total sample size of six. Surely that can't possibly capture the complexity of the system.

[Enscheff]

I'm sure I don't spend anything like as much time looking analyzing baseball stats as you do, but I don't see how this type of analysis tells us much of anything. A baseball player's future production has to be an extremely complex system with a huge number of variables including things like psychological profile and injuries (which are inherently unpredictable). I wouldn't be surprised at all to learn that it is a chaotic system (that is, slight differences in initial conditions can have very large effects since response is non-linear), so I'm curious, how good do projections of baseball players turn out to be? In this case, you have a total sample size of six. Surely that can't possibly capture the complexity of the system.

You are more than welcome to come up with your own list of comps for Swanson, and draw conclusions as you see fit.

Then we can discuss your findings. Until then, contrarian posts like this bring nothing of value to the discussion.

[conalthomas]

You are more than welcome to come up with your own list of comps for Swanson, and draw conclusions as you see fit.

Then we can discuss your findings. Until then, contrarian posts like this bring nothing of value to the discussion.

Well, actually, my point is that comps aren't likely to work, so the last thing I would want to do is come up with some limited sample set of my own that does no better of a job of predicting future outcomes. And, I contend that there is value brought to the discussion; my argument is that predicting a career trajectory of a player like Swanson based on comparisons with the career trajectories of six other players that are similar with regard to some particular parameters has quite limited utility. I mean look at the comparisons you describe; you have a spread of career WAR from some negative value all the way up to 31.4 in just six samples. What would it look like if you had a hundred samples? Having looked at lots and lots of data sets (not to do with baseball), my very strong feeling is that the spread would be even greater with more data, and I know you really can't make any kind of conclusion about the shape of the distribution with just six samples. This data says to me; OK, Swanson could have a career WAR somewhere between some negative value and probably something higher than 31.4 (because what are the chances you have captured the true range of possibilities with 6 samples?), possibly quite a bit higher. And, with just six samples, we certainly can't conclude with this data set that a low career WAR like 2 or 6 or 8 is any more likely than a very different career WAR of 20-30 (or more).

What I was really hoping you'd say (in my perfect version of the world) is something like, 'I can understand your skepticism, but the method of using comps like this has proven to be quite successful in making future projections.', along with a link to an analysis that showed this. I am truly skeptical that the data set you provided has much in the way of predictive utility. I can absolutely promise you that in professional data analysis circles, predictive statements based on a data set like this would be laughed at. But, here's the thing, I don't spend near as much time looking at baseball stats as you do. I could do an analysis of the predictive utility of comps, and even do an analysis of variance that would let us weight the various parameters used in comparisons, but I don't want to do that. I'm hoping that someone has already done it. I see comps used a lot in baseball, and what I'm hoping is that you know of something that indicates that they actually mean anything.

[Enscheff]

Well, actually, my point is that comps aren't likely to work, so the last thing I would want to do is come up with some limited sample set of my own that does no better of a job of predicting future outcomes. And, I contend that there is value brought to the discussion; my argument is that predicting a career trajectory of a player like Swanson based on comparisons with the career trajectories of six other players that are similar with regard to some particular parameters has quite limited utility. I mean look at the comparisons you describe; you have a spread of career WAR from some negative value all the way up to 31.4 in just six samples. What would it look like if you had a hundred samples? Having looked at lots and lots of data sets (not to do with baseball), my very strong feeling is that the spread would be even greater with more data, and I know you really can't make any kind of conclusion about the shape of the distribution with just six samples. This data says to me; OK, Swanson could have a career WAR somewhere between some negative value and probably something higher than 31.4 (because what are the chances you have captured the true range of possibilities with 6 samples?), possibly quite a bit higher. And, with just six samples, we certainly can't conclude with this data set that a low career WAR like 2 or 6 or 8 is any more likely than a very different career WAR of 20-30 (or more).

What I was really hoping you'd say (in my perfect version of the world) is something like, 'I can understand your skepticism, but the method of using comps like this has proven to be quite successful in making future projections.', along with a link to an analysis that showed this. I am truly skeptical that the data set you provided has much in the way of predictive utility. I can absolutely promise you that in professional data analysis circles, predictive statements based on a data set like this would be laughed at. But, here's the thing, I don't spend near as much time looking at baseball stats as you do. I could do an analysis of the predictive utility of comps, and even do an analysis of variance that would let us weight the various parameters used in comparisons, but I don't want to do that. I'm hoping that someone has already done it. I see comps used a lot in baseball, and what I'm hoping is that you know of something that indicates that they actually mean anything.

Players comps are, by definition, designed to return a limited sample size. One could loosen the constraints to get a larger sample size, but then they cease to be interesting.

More general projections already exist for Swanson, and many updated ones will be available this offseason. Those projection systems have been scrutinized endlessly in countless articles, so you can go read up on them until you're content.

Further, I didn't draw any conclusions that require any sort of analytical certainty. Notice I didn't say anything like "he has a 33% chance of....". My conclusion was quoted by you, so you can reread it if you'd like.

Lastly, your suggestion that your post was nothing but an innocent request for information is pretty silly. In the amount of time it took you to write those 2 posts, you could have easily googled "predictive power of player comps", perused the contents of a few articles, and drawn your own conclusion.

Your post was meant to be contrarian, and nothing more. This thread was started with the hopes of starting a discussion where folks would say "here is another list of players that have done X by Y age", or "players with batted ball profile X, Y and Z tend to produce like this....".

Sadly, that never seems to happen. All we get are contrarian posts lacking any sort of data, and designed to argue. Unless they are glowing proclamations about how great the Braves are...then they are cheerfully agreed with....doubly so if they include a list of Braves prospects that highlights just how impressive it all is. Objectivity is the devil on these boards.

[conalthomas]

Yes, I know what it says:

"Swanson is young enough that there is some time for him to rebound and be a Phillips/Wilson level player. Of course, there is a significant chance he ends up being more similar to the other 4 guys."

Based on the data, there is absolutely no reason to choose between the two options. We don't know if he has a 60% chance of being like Phillips/Wilson, a 40% chance, a 20% chance...and the same could be said for him being like the other 4 guys. So, yes, I suppose there is "significant" chance that he'll end up like the other 4 guys, but that might be in reality like a 20% chance. The way it was phrased makes it sound like a good outcome was becoming increasingly unlikely and the sad outcome was the more probable of the two.

But I actually do appreciate the reply; I had been wondering about comps. When you have a very small sample size and you get a very big spread (as with the comps you cited for Swanson), the comparison is basically saying we have very little idea where Dansby is going to end up. Using them is probably more likely to mislead than otherwise, unless the specific point is - we have no idea, comps are worthless in this case. I suppose if your comps returned a spread of 25-55, or some negative value-10, you might think, "Oh, goody", in the first case, and, "Oh, bummer" in the second, but honestly, with samples sizes in the single digits, but I'm not sure that's justified. Your error rate is going to be high. I was just hoping someone had analyzed how high it tended to be.

Yes, you can say my original post was being merely contrarian, and therefore adding nothing of value, but I disagree. The point of my post was not to be merely contrarian, but to posit that comps are actually a waste of time unless the purpose is entertainment and idle speculation. I would say such a point of view was entirely relevant to the discussion. I could have googled, as you say, but I felt quite confident that you have spent many more hours reading about all of this than I have, and certainly more time that it takes to write a post, or two. Maybe there is an obscure and difficult-to-find, but highly valuable, reference out there. Possibly, you might share insight gained from a synthesis of the many sources I have no doubt you are familiar with. Plus, I found it hard to believe that comps are used so much if they truly have little value. I was thinking I was wrong and would be corrected, which would have been just fine with me.

[conalthomas]

Your post was meant to be contrarian, and nothing more. This thread was started with the hopes of starting a discussion where folks would say "here is another list of players that have done X by Y age", or "players with batted ball profile X, Y and Z tend to produce like this....".

Sadly, that never seems to happen. All we get are contrarian posts lacking any sort of data, and designed to argue. Unless they are glowing proclamations about how great the Braves are...then they are cheerfully agreed with....doubly so if they include a list of Braves prospects that highlights just how impressive it all is. Objectivity is the devil on these boards.

OK, this is just ridiculous. "Objectivity is the devil..." when what you are upset about is an argument citing reasons why your original contention is, if not incorrect, at least misleading. Just because something has data doesn't mean it has any value. A discussion of whether data is relevant and useful is essential to objectivity. How is it objective to cite data and conclusions, when the data don't justify the conclusions? My whole point is that it doesn't matter what you are using for comparisons if your sample sizes are 4 or 6 or 8 and you are dealing with a very complex system. Heck, if you just wanted to find out whether a coin was true, flipping just six times wouldn't be much help.

So, yes, I could have cited other comparisons, but it wouldn't have contributed, at all, to anything to do with objectivity.

[Enscheff]

Simple question...

What do you project for Swanson moving forward? Why?

If you are unwilling to contribute that to the discussion, you are just being contrarian. That's what "just being contrarian" means.

You know how you can spot a contrarian post from a mile away? It starts with, "I don't actually know the answer, but....", just like yours did.

[chop2chip]

There are four realistic things I'm looking to see out of Dansby next year:

(1) K% < 20%
(2) DRS > +5
(3) Baserunning value (Fangraphs) > +3
(4) ISO > .150 ... + 30 doubles

Those are the four areas where I have been truly disappointed in Dansby this year. (Seriously, how can a player whose defining quality is his baseball smarts with his above average speed have negative baserunning value?). If he can do those things, then he should reach 3 WAR next year assuming normal luck (and no significant regression in BB%, though some regression should be expected). I feel like all of those things are reasonable to expect and in line with the player we were promised before this season.

[conalthomas]

Simple question...

What do you project for Swanson moving forward? Why?

If you are unwilling to contribute that to the discussion, you are just being contrarian. That's what "just being contrarian" means.

You know how you can spot a contrarian post from a mile away? It starts with, "I don't actually know the answer, but....", just like yours did.

You keep using the word "contrarian", but all it means is to go against prevailing opinion, and it certainly isn't necessarily a bad thing. In fact, in enterprises like science, contrarian arguments are essential to forward progress. You, yourself, often take a contrarian position on this board, and are quite proud of it, and I say more power to you.

I do not have a projection for Swanson, and yet I was contributing in a meaningful way to the discussion, just not in a way that you liked. I questioned the validity of the conclusions of the original post, and I gave the reasons for my skepticism. In your responses you have not made a reasoned argument as to why those conclusions are, in fact, valid; you have mainly relied on argumentem ad hominem. I think if you look that phrase up you will discover that has been pretty much been entirely the basis of your argument. In all honesty, I had hoped for more.

And, no, the point of all of this is not to get into some sort of contest with you. I actually admire much of what you bring to the board, but you have to admit that if projections based on comps are, for want of a better word, garbage, that's pretty important to the discussion.

[Enscheff]

Contrarian opinions are fine when they are accompanied by another opinion supported by additional data. When someone is "just being contrarian" it means they are objecting without providing anything else. Or...exactly what you're doing.

Now we are arguing with me over terminology just like Gov used to do when he didn't have anything of value to contribute, but wanted to argue until he was "right" about something...anything...just to declare a victory over me, no matter how small. Exactly like the kid in class pointing out the professor's spelling mistakes on the whiteboard. Even more pointless than "just being contrarian".

I think we would have been better off had you remained silently lurking in the weeds. Anywho, that's what the ignore feature is for.

[smootness]

Don't worry about it, conal. His argument is flawed, just like the last time he did this.

[conalthomas]

Contrarian opinions are fine when they are accompanied by another opinion supported by additional data. When someone is "just being contrarian" it means they are objecting without providing anything else. Or...exactly what you're doing.

Now we are arguing over terminology...just like Gov used to do when he didn't have anything of value to contribute, but wanted to argue until he was "right" about something...anything...just to prove me wrong. More pointless than "just being contrarian".

I think we would have been better off had you remained silently lurking in the weeds. Anywho, that's what the ignore feature is for.

Yes, and the reason why we are arguing about terminology is because you are deliberately trying to bamboozle and obfuscate. OK, rather than get caught up in the back and forth, let me ask you a simple question:

If it is true that comps as used here are basically worthless in making projections, don't you think that is relevant to a discussion involving the use of comps to make projections?

I have given reasons (based, I believe, on valid statistical reasoning) why I think comps are not useful for projections. You have never, in any way, countered that argument based on any kind of rational, objective reasoning. Your arguments have been largely an attempt to demean, to confuse the issue, to try and create a false standard that must be used in a discussion. In one of your previous posts you said, "I didn't draw any conclusions that require any sort of analytical certainty. Notice I didn't say anything like "he has a 33% chance of....". OK, so are you making a projection, or what? What is the range of Swanson's possible career WAR? Is it truly from a negative value to 32+? And, what is the probability of that interval? You have no idea, you can't, not from that data. You have in a number of posts, elsewhere, made projections for Swanson based on various sorts of data, and those may well be valid, but based on everything you have said here, I have no reason to think that any projections based on comps are any more valid than projections based on perceptions of Swanson's "makeup", or whatever else it is that "stats users" like to criticize people for doing.

Let me turn what you are saying around on you; when I said that I had doubts about the validity of your projections, that could have been a basis of exploring what we can and cannot know about future projections based on comps. I would have preferred that. But instead, it became about winning for you, and your arguments were pretty much insults, etc., all the while decrying the death of reason and objectivity. I have to say that I am surprised that in a community that uses sample size concept so frequently that "small sample size" has an abbreviation all of its own, that no one seems concerned about the ridiculously small sample size of comps based on a sample size of six. Career arcs have to be as complex as at bats. Would you project someone's next 50 at bats, say, based on a sample of six at bats? (I keep wondering if I'm missing something, here.)

[conalthomas]

Don't worry about it, conal. His argument is flawed, just like the last time he did this.

Thank you.

I'm not worried, although I think I might be somewhat embarrassed to have gotten into something so "full of sound and fury, and signifying nothing".

[Super]

Your arguments have been largely an attempt to demean, to confuse the issue, to try and create a false standard that must be used in a discussion.

have you met my friend enscheff?
perfect way to describe how he operates. bravo.
you absolutely have a point, and that fact has made him upset.