Sox sign Gordon Beckham, designate Viciedo for Assignment

[Reddy]

I guess I just don't understand why anyone takes these projections very seriously. They were off an average of 6.7 games per team last season. IMO, that's not very accurate, and not something I would use to determine anything. Not that anything really needs to be determined at this point.. But by all means, consider them accurate. If there are Oriole fans that think like you do, they probably planned a long summer vacation last year and missed some great baseball.

 

It's more accurate than any other system that's been created to this point, and again, the point is not exact accuracy. The point is a baseline from which standard deviations and luck affect the outcome. I think I've written that about fourteen times in this thread.

[Reddy]

neither do you, when you are proven wrong, you disappear. anywho, this does not involve you. no need to take pot shots. just to try to make you look good.

 

Are you referencing something in particular? A lot of the time I'll admit when I've hyperbolized.

 

[witesoxfan]

You think my posts are stupid and pointless, yet you keep responding. If you read my first post in this thread, I would have been done. But for some reason, people kept responding.

 

I don't think your posts are stupid and pointless, but you brought up a legitimate question as to why you felt they should be taken seriously, you've been given a multitude of responses as to why we feel they should be taken seriously, and you continue to ignore those points and say "they were off by 6.7 games." As I said, if they were predictions, which would be attempting to predict the future and give us accurate numbers for what will come up, I'd agree, these shouldn't be taken seriously. These are not predictions. They are a series of numbers that are determined by previous numbers to try and determine talent level of the 30 teams in the majors taking context out of the equation. Context and what happens is a huge, huge part of the equation, which is why we do the whole "playing the game thing." Without that, these numbers wouldn't exist in the first place.

 

If you can get past the difference between predictions and projections, you may see that these aren't designed to predict the future but to project a likely outcome based on previous information. It's a learning tool. You may also continue to ignore them, and that's fine, but if you are going to continually s*** on them, it may be worth more to you to leave the discussion. This isn't a threat or anything, and you can continue to say how you feel, but just know that by doing so, you're being a Debbie downer and really killing the vibe.

[Dick Allen]

I don't think your posts are stupid and pointless, but you brought up a legitimate question as to why you felt they should be taken seriously, you've been given a multitude of responses as to why we feel they should be taken seriously, and you continue to ignore those points and say "they were off by 6.7 games." As I said, if they were predictions, which would be attempting to predict the future and give us accurate numbers for what will come up, I'd agree, these shouldn't be taken seriously. These are not predictions. They are a series of numbers that are determined by previous numbers to try and determine talent level of the 30 teams in the majors taking context out of the equation. Context and what happens is a huge, huge part of the equation, which is why we do the whole "playing the game thing." Without that, these numbers wouldn't exist in the first place.

 

If you can get past the difference between predictions and projections, you may see that these aren't designed to predict the future but to project a likely outcome based on previous information. It's a learning tool. You may also continue to ignore them, and that's fine, but if you are going to continually s*** on them, it may be worth more to you to leave the discussion. This isn't a threat or anything, and you can continue to say how you feel, but just know that by doing so, you're being a Debbie downer and really killing the vibe.

Killing the vibe? I have been the one accused of ignoring the projection just thinking the Sox will be great. I think they will beat this projection and be a good team. Certainly capable of making the playoffs. Debbie Downers are the ones taking this projection and thinking the team isn't very good, with little chance at the playoffs.

Edited January 30, 2015 by Dick Allen

[witesoxfan]

Killing the vibe? I have been the one accused of ignoring the projection just thinking the Sox will be great. I think they will beat this projection and be a good team. Certainly capable of making the playoffs. Debbie Downers are the ones taking this projection and thinking the team isn't very good, with little chance at the playoffs.

 

It's whatever. We know how you feel about projections. I think they will beat them too. I hope we're right.

[Reddy]

Wite brought up a great distinction:

 

Predictions are what you think WILL happen

 

Projections are what you think SHOULD happen based on past performance.

 

PECOTA is a projection system.

[Reddy]

It's whatever. We know how you feel about projections. I think they will beat them too. I hope we're right.

I also think they'll beat them, by about 7 games. And if things go RIGHT (read Avi being a 2 war player, our bullpen being lights out, and danks not completely sucking) we could win 90 games.

[Dick Allen]

Wite brought up a great distinction:

 

Predictions are what you think WILL happen

 

Projections are what you think SHOULD happen based on past performance.

 

PECOTA is a projection system.

Then I will rephrase, what SHOULD have happened, and what a guy like Harold Reynolds who obviously doesn't have a big fanbase thinks WAS going to happen, wasn't so different accuracy-wise.

 

I also think most if pressed to be accurate with predictions, will use what SHOULD happen vs. what they hope or think will happen most times.

[shago]

Pls rename this the PECOTA thread....

 

"I don't know. But whatever it is, I hate it. " - Bart (Blazing Saddles)

 

 

[witesoxfan]

Then I will rephrase, what SHOULD have happened, and what a guy like Harold Reynolds who obviously doesn't have a big fanbase thinks WAS going to happen, wasn't so different accuracy-wise.

 

I also think most if pressed to be accurate with predictions, will use what SHOULD happen vs. what they hope or think will happen most times.

 

Right, but there is a distinction. If you asked Jeff Sullivan or Dave Cameron what will happen, they will say "I don't know."

[Dick Allen]

Right, but there is a distinction. If you asked Jeff Sullivan or Dave Cameron what will happen, they will say "I don't know."

But even what they are saying should happen is open to debate. I like BP. I get it every year. In fact, it was delivered yesterday. But I don't know what makes their projection something we should hold up on a pedestal. At the end of the day, if they project the White Sox to win 60 or 95 ,it doesn't matter, and that has been my point all along.

[witesoxfan]

OK. You win. I'm still going to pay attention to projections and cite them in the future, but OK.

[Chisoxfn]

I'd subscribe because more information and more inputs are always better than less, especially if you can properly contextualize the information.

 

Also, I'm not basing my excitement for the White Sox on a 78 win projection. I'm basing my excitement on the White Sox because I think they are going to have a really good and fun team. On top of thinking Garcia is going to have a good year, I also think Micah Johnson is going to win the 2B job and be a 1.5-2 WAR player, I honestly do not mind Gordon Beckham being on the roster as long as he's not the starter (or if he is, hopefully it's because he's finally hitting well), I'm really excited for what should be a much improved bullpen, and I'm really hoping Rodon can come in and be a shot in the arm and great back of the rotation starter.

More and more inputs are not always better. More good inputs are better. I think PECOTA is great for plans and is a data point but the real question is what do the top MLB exec's and front office people think of them. How flawed are they? It would appear given the overall volatility in production, etc, that you'd have a hard time using PECOTA as a benchmark or a good data set. It is more an interesting thing for fans to look at that is based upon assumptions, some of which are pretty significant.

 

Star players will consistently have their values under rated due to overall regression components while young players will in average, probably be accurate, but specific players can be all over the map in terms of progression / regression and thus again you could have a pretty wide variant on any particular team. Part of which is driven by level of talent of those players as the "talent" is not an assumption that can be applied. Basically put, I have no idea why people have gotten so we are a 79 win team because of X. I personally think these are just something to talk about, but certainly not something to write home about or get overly excited / depressed about.

 

 

[Chisoxfn]

But even what they are saying should happen is open to debate. I like BP. I get it every year. In fact, it was delivered yesterday. But I don't know what makes their projection something we should hold up on a pedestal. At the end of the day, if they project the White Sox to win 60 or 95 ,it doesn't matter, and that has been my point all along.

Your best bet is to combine projections and analyzst expectations, etc, and you'd probably get to something relatively accurate. Or you could analyze the projections done by BP, point out the flaws, make adjustments, re-run and see where we fall out.

[Eminor3rd]

Projections feed discussion. You just don't like opinion that is different than yours. Sorry. If you want to ignore the past and think this is a really accurate "tool" as you like to say which lets you know where every team is at, fine. I think you are wrong.

 

That's a ridiculous thing to say. You have asked, like seventy times, "what is the POINT of projections?" and we have answered it all seventy times. I'm not telling you to like them or agree with them, I'm just telling you why they're useful. It has nothing to do with my OPINION of them. You haven't even ASKED my opinion of them, and I haven't even offered it.

 

I HATE chapstick. But you can tell me why it's useful, and I believe you. But its benefits don't interest me and I don't like using it. But I'm not trying to s*** on everyone who DOES use it. My opinion doesn't change the fact that it has uses. It doesn't objectively suck simply because it doesn't permanently cure chapped lips.

Edited January 30, 2015 by Eminor3rd

[Eminor3rd]

But even what they are saying should happen is open to debate. I like BP. I get it every year. In fact, it was delivered yesterday. But I don't know what makes their projection something we should hold up on a pedestal. At the end of the day, if they project the White Sox to win 60 or 95 ,it doesn't matter, and that has been my point all along.

 

It IS open for debate. That's what we're doing. No one is holding anything up to a pedestal! In your issue of BP, does it say anything at all about how these projections are so accurate we don't even need to play the games? Or, instead, does it say something about how they provide a useful and interesting frame of reference and then go onto cite its shortcomings?

 

They DO matter. The number they land on doesn't matter, but that number in context of the other tell us things about our team. Every week, some random sap wins the freaking lottery, even though his odds were one in 2,000,000. He didn't actually buy 2,000,000 tickets to ensure a victory. But the odds were the odds.

Edited January 30, 2015 by Eminor3rd

[Dick Allen]

It IS open for debate. That's what we're doing. No one is holding anything up to a pedestal! In your issue of BP, does it say anything at all about how these projections are so accurate we don't even need to play the games? Or, instead, does it say something about how they provide a useful and interesting frame of reference and then go onto cite its shortcomings?

 

They DO matter. The number they land on doesn't matter, but that number in context of the other tell us things about our team. Every week, some random sap wins the freaking lottery, even though his odds were one in 2,000,000. He didn't actually buy 2,000,000 tickets to ensure a victory. But the odds were the odds.

The White Sox number is 78. What does that tell you about the team you didn't already know before this number was revealed? How did this number effect the way you feel about the current team?

 

Last year, Baltimore and Chicago each had a number of 75. One team won 96 games, the other 73. Washington had a number of 88, they won 96. Boston had an 89, they won 71.

 

The number doesn't matter. They didn't take Baltimore's division title away because PECOTA gave them 75 wins before the season started.

Edited January 30, 2015 by Dick Allen

[Eminor3rd]

The White Sox number is 78. What does that tell you about the team you didn't already know before this number was revealed?

 

Last year, Baltimore and Chicago each had a number of 75. One team won 96 games, the other 73. Washington had a number of 88, they won 96. Boston had an 89, they won 71.

 

The number doesn't matter. They didn't take Baltimore's division title away because PECOTA gave them 75 wins before the season started.

 

Here's what I thought when I saw the projections:

 

"Whoa, shouldn't that be higher? It seems like they added enough talent to be projected at at least .500.

 

 

Christ, I keep forgetting about Danks and Noesi. The projections don't like them.

 

 

Yeah, it sure makes sense why they don't like them. Those guys were both worse than they seemed last year, and could easily be even worse this year. I guess that's more of an issue than I originally thought.

 

Also, this projects Sale and Quintana and Abreu all to take major steps back. I guess i can;t really EXPECT a Cy/MVP season out of those guys. Sure they're capable, but I wouldn't BANK on it.

 

 

Why are these guys so high? Ah, they get a ton of value from their depth. When I look at the White Sox, I don;t see a lot of capable backups. i suppose it makes sense that when people get hurt, Leury garcia has to play, and that's going to make a big difference."

[Dick Allen]

Here's what I thought when I saw the projections:

 

"Whoa, shouldn't that be higher? It seems like they added enough talent to be projected at at least .500.

 

 

Christ, I keep forgetting about Danks and Noesi. The projections don't like them.

 

 

Yeah, it sure makes sense why they don't like them. Those guys were both worse than they seemed last year, and could easily be even worse this year. I guess that's more of an issue than I originally thought.

 

Also, this projects Sale and Quintana and Abreu all to take major steps back. I guess i can;t really EXPECT a Cy/MVP season out of those guys. Sure they're capable, but I wouldn't BANK on it.

 

 

Why are these guys so high? Ah, they get a ton of value from their depth. When I look at the White Sox, I don;t see a lot of capable backups. i suppose it makes sense that when people get hurt, Leury garcia has to play, and that's going to make a big difference."

Leury Garcia isn't going to play, although they did give him a better chance of improvement over Matt Davidson. BP projects Abreu to hit .295 with 23 homers 74 rbi and 36 walks. They project Avi Garcia to hit 7 homers. In their write up, they mentioned Q as one of the best lefties in the league and projected him to have a better chance of improving over just about every other White Sox. Then he made the top 10 list in highest projected WAR decrease. David Robertson is actually projected to have fewer saves than Mariano Rivera. It was pretty inconsistent this year.

Edited January 30, 2015 by Dick Allen

[ptatc]

Then I will rephrase, what SHOULD have happened, and what a guy like Harold Reynolds who obviously doesn't have a big fanbase thinks WAS going to happen, wasn't so different accuracy-wise.

 

I also think most if pressed to be accurate with predictions, will use what SHOULD happen vs. what they hope or think will happen most times.

Another view would be that people look at these projections (which are biased as I've said) and pay for this company to exist. It's kind of like weather personalities. They know they are most likely wrong but people still like them.

 

If I can be 7 games off and still have a company that gets paid to do it, I'm going to say that the teams who are going to finish in the bottom half will win 75, that gives me 68-82 as a range. Teams that look good I'm going to say 82, that gives me 75-89 range. I will be right most often and be correct as PECOTA and weather personalities, although I don't look as good as Cheryl Scott and I'm not as accurate as Tom Skilling.

[ptatc]

It IS open for debate. That's what we're doing. No one is holding anything up to a pedestal! In your issue of BP, does it say anything at all about how these projections are so accurate we don't even need to play the games? Or, instead, does it say something about how they provide a useful and interesting frame of reference and then go onto cite its shortcomings?

 

They DO matter. The number they land on doesn't matter, but that number in context of the other tell us things about our team. Every week, some random sap wins the freaking lottery, even though his odds were one in 2,000,000. He didn't actually buy 2,000,000 tickets to ensure a victory. But the odds were the odds.

I would disagree with this based on the concept that the algorithm is biased. Their results over the past few years show how inaccurate they are. They only thing they really matter for is to create discussion, which they've done.

[Dick Allen]

Another view would be that people look at these projections (which are biased as I've said) and pay for this company to exist. It's kind of like weather personalities. They know they are most likely wrong but people still like them.

 

If I can be 7 games off and still have a company that gets paid to do it, I'm going to say that the teams who are going to finish in the bottom half will win 75, that gives me 68-82 as a range. Teams that look good I'm going to say 82, that gives me 75-89 range. I will be right most often and be correct as PECOTA and weather personalities, although I don't look as good as Cheryl Scott and I'm not as accurate as Tom Skilling.

Yes. If you just said everyone would be .500, you would have been off an average of 8.16 a team. The White Sox number is 78 this year. Last year, there were 2 teams with a 78. One team won 88, the other 68. They averaged the 78 they were projected, but both projections obviously weren't accurate.

[chitownsportsfan]

PECOTA seems heavily regressed to the mean, and while that might make statistical sense, what we are interested in as fans isn't the mean case, but the outlier case(s). What is interesting is the breakouts, both at the team level and player level. I haven't seen where PECOTA predicts outliers any better than ZIPS, or even the Diamond Mind sim predictions.

 

Plus, the Sox have a lot of guys without a lot of data on them, or bad data due to injury. Garcia and Eaton (his bad 2013) fall into the injury related cases, and Jose Abreu is a 28 year old that had the 3rd highest wRC+ in baseball last year. Projections are going to put him down for a regression, but many scouts (and us fans) might be putting him down to actually IMPROVE, based on the fact that now he knows the grind of the season, many of the pitchers, how they plan to get him out, etc.

 

I wouldn't guarantee it, but with Jose, there just isn't a lot of data for the projections to look at, in this instance scouting might be better informed.

 

And of course playing time projections are often inaccurate.

 

I'm not completely arguing to "throw out the projections, including PECOTA" but it's been proven time and time again that their predictive value is mixed, at best. Yea I'd rather the Sox were projected for 100 wins, but it's not the end of the world to see 78 either. I don't think Hahn or us fans think this is a 78 win team. I think it's closer to 85 but that's just my opinion.

[Chisoxfn]

I would disagree with this based on the concept that the algorithm is biased. Their results over the past few years show how inaccurate they are. They only thing they really matter for is to create discussion, which they've done.

Their are inherent biases in the algorithm which will tend to favor certain teams, thus making where you fall, relatively biased based upon the team you have. I'd tell you right now...in my line of work, which is heavily focused on plan & analysis, I'd be pretty flipping pissed off if I consistently saw the plan vs. actual results that these models have historically predicted for our Sox and to be frank, I'd probably be without a job.

 

Note: I'm not saying they are biased against the Sox, rather, biased against certain types of rosters, etc (and I'd argue the Sox have a lot of guys who historically fall into some of these scales which drive this). The fact that your metrics can consistently be so wrong on specific teams is pretty alarming as well. Hard to confirm overall validity when you have such wide disparities and almost consistently. The statistic significance of the data becomes far more limited when you look at the deviations from the predictive model and realize you have some pretty extensive deviations (on both sides of the equation; Both in # (Total Win Variations) and in count (e.g., # of teams who have significant deviations).

 

 

[chitownsportsfan]

Their are inherent biases in the algorithm which will tend to favor certain teams, thus making where you fall, relatively biased based upon the team you have. I'd tell you right now...in my line of work, which is heavily focused on plan & analysis, I'd be pretty flipping pissed off if I consistently saw the plan vs. actual results that these models have historically predicted for our Sox and to be frank, I'd probably be without a job.

 

Note: I'm not saying they are biased against the Sox, rather, biased against certain types of rosters, etc (and I'd argue the Sox have a lot of guys who historically fall into some of these scales which drive this). The fact that your metrics can consistently be so wrong on specific teams is pretty alarming as well. Hard to confirm overall validity when you have such wide disparities and almost consistently. The statistic significance of the data becomes far more limited when you look at the deviations from the predictive model and realize you have some pretty extensive deviations (on both sides of the equation; Both in # (Total Win Variations) and in count (e.g., # of teams who have significant deviations).

 

Look at these r2 values:

http://www.hardballtimes.com/evaluating-th...ection-systems/

 

ZIPS 0.330

Steamer 0.316

PECOTA 0.304

Oliver 0.307

Marcel 0.240

Average 0.318

 

Nothing to write home about, that's for sure. Basically (in my limited stats 101 knowledge) the r2 is simply "the variance explained in the model".

Edited January 30, 2015 by chitownsportsfan