How much do you trust projections?

[Comx]

With the most popular projection systems being ZiPS and Steamer, I'm wondering just how reliable are these formulas? I've found with young and old players the projections can get a bit wonky. These formulas are a best guess using the intelligence and historical data available but nobody would admit them to be near perfect. Here's a three examples of projections that are looking pretty off:

 

Brad Miller - 24 y.o. Great milb stats and performed well in 1/2 of 2013, but has faltered and shown no real signs of improvement in 2014, BABIP is low but consistently.

Current 2014 wRC+: 34

Projected 2014 wRC+: 90 and 97 (ZiPS and Steamer RoS, respectively)

 

Alfonso Soriano - 38 y.o. BABIP and other peripheral stats pretty normal, just not performing as well as expected.

Current 2014 wRC+: 78

Projected 2014 wRC+: 105 and 90 (ZiPS and Steamer RoS, respectively)

 

Joe Nathan - 39 y.o. K's are down, BB's are up, BABIP is lower than average, velocity down about 1 mph

Current 2014 FIP: 4.82

Projected 2014 FIP: 3.42 and 3.19 (ZiPS and Steamer RoS, respectively)

[BigBounceyBlueBalls]

Projections are advanced hype! That's why we play the games! Any team in any league is only 3 or 4 players or injuries away from first to worst any given season or year. Unless your the Evil Empire and just buy em! Lol

[NorthOf49]

They're great if you're realistic about what they represent. They summarize what a player's past performance means with regard to his future. I look at a lot of the peripheral stats that projections take into account; they're a good way to check what those peripherals mean at face-value.

 

Obviously they tend to be conservative too. Sometimes a player changes his true talent drastically in a short span of time; projections will be late on picking that up. Most of the time that's a good thing though, it's important to be skeptical of players deviating from their established level.

[Comx]

Sometimes a player changes his true talent drastically in a short span of time; projections will be late on picking that up. Most of the time that's a good thing though, it's important to be skeptical of players deviating from their established level.

 

This is basically what I'm referring to with young and old players. Projection systems are late/slow to pick-up on rookies struggling in the majors, and late/slow to pick-up on that age-related decline/drop-off... in both cases the projections are too high.

[TheHurl]

This is basically what I'm referring to with young and old players. Projection systems are late/slow to pick-up on rookies struggling in the majors, and late/slow to pick-up on that age-related decline/drop-off... in both cases the projections are too high.

 

here's the crazy thing about projections though...they are projected for the whole season. I'm pretty sure that Soriano will improve, Miller if given the chance will improve. You can't expect projections to be exact...and some won't be close in some cases. I mean there isn't a true breakout or bust season that is full predicted. Projection systems do help seeing that a breakout or bust is more likely. Essentially if you take projections as the law you are in for disappointment...but don't dismiss them fully either.

[Laika]

here's the crazy thing about projections though...they are projected for the whole season. I'm pretty sure that Soriano will improve, Miller if given the chance will improve. You can't expect projections to be exact...and some won't be close in some cases. I mean there isn't a true breakout or bust season that is full predicted. Projection systems do help seeing that a breakout or bust is more likely. Essentially if you take projections as the law you are in for disappointment...but don't dismiss them fully either.

 

Well said. They are powerful tools for player evaluation, but they aren't the end all be all, nor do they purport to be. Trying to predict/project what players will do is only their purpose in the most shallow and superficial sense. The real utility lies not in the conclusions of projection systems, but in how and why they got there, taken in conjunction with information like scouting reports.

 

There is A LOT of information and data that projection systems can parse much better than humans, but at the same time, there is also A LOT of information that humans can interpret more effectively than projection machines. Knowing which is which separates the Friedman's from the Anthopouli.

[jaysblue]

Yeah like some have said, they aren't the be-all end-all. It was making me sick how some posters in the offseason were one-dimensional in their arguments/assessments by just saying "Steamer projects this player to do this and that" when having discussions on certain players.

[GD]

Steamer RoS is for the rest of the season, by the way.

 

They're very good for their intended uses. Too bad people pretend like they're gospel.

[Comx]

here's the crazy thing about projections though...they are projected for the whole season.

 

If you read my post I'm referring to RoS (rest of season) projections, not whole seasons as you said.

 

I'm pretty sure that Soriano will improve, Miller if given the chance will improve. You can't expect projections to be exact...and some won't be close in some cases. I mean there isn't a true breakout or bust season that is full predicted.

 

I'm not expecting projections to be exact, I'm saying with young and old players often have seasons that are defying projection systems altogether, you can basically throw them out completely in some cases. Mike Trout's rookie season is an example, projections would of been way off. And then again his next season would of been way off.

 

Essentially if you take projections as the law you are in for disappointment...but don't dismiss them fully either.

 

Agree on this point.

 

But I'm not sure you're getting my argument: projection systems are better suited for players with a few full seasons in the majors under their belt and 35-37 years old and under, outside of this they are significantly less predictable.

[Nox]

Projections are estimates of true talent. Best guesses. They can and will be wrong. That said, more often than not they're going to be more accurate than what your intuition tells you.

 

Regarding Trout: Sure his pre rookie season ZIPs would have underestimated his true talent but I don't think you'd find a scout working in professional baseball who would have predicted he'd be the best player in baseball from day 1 either.

[Jimcanuck]

The potential floor and ceiling of any player's yearly performance is quite wide. Some are yoyo's like Rios was for the longest time, good year, bad year.

 

I've been watching baseball long enough to take projections with a grain of salt. At best, they can reasonably predict the average result from a population of players (a team), but project individual results? BS

[Jimcanuck]

Projections are estimates of true talent. Best guesses. They can and will be wrong. That said, more often than not they're going to be more accurate than what your intuition tells you.

 

Intuition tells me EE is going to end up approximately where he ended up last year. I don't need to spend hours doing advanced stats to come up with that answer.

[Nox]

I don't need to spend hours doing advanced stats to come up with that answer.

 

Took me less than 5 seconds to look up Edwin's ROS projections. Slow connection Jim? Bad with the Google?

[Jimcanuck]

Took me less than 5 seconds to look up Edwin's ROS projections. Slow connection Jim? Bad with the Google?

 

Ha, nice deflect

 

Projections are rather pointless. Why don't you look up the projection for Morrow while you are at it? And Fielder?

 

Injuries, normal year to year variance, getting dumped by the g/f (Romero), scenery change (Aaron Hill), and what have you means spending 5 seconds of your life looking up a projection is a waste of time.

[John_Havok]

Projections take emotion out player evaluation and let the numbers start the conversation rather than s*** like "he's just a guy that brings his lunch pail to work, plays hard, and gets his uniform dirty..." This is also known as the Kevin Towers scouting model.

 

It's isn't about "trusting" projections, they're just a starting point of a conversation that asks questions that might not otherwise get asked.

[Nox]

Projections are rather pointless. Why don't you look up the projection for Morrow while you are at it? And Fielder?

 

No, that's a pretty pointless statement. What did your projection for Fielder look like before the season? I'm guessing it would have been: "Intuition tells me Fielder is going to end up approximately where he ended up last year."

Guess that would have been wrong too.

[Jimcanuck]

No, that's a pretty pointless statement. What did your projection for Fielder look like before the season? I'm guessing it would have been: "Intuition tells me Fielder is going to end up approximately where he ended up last year."

Guess that would have been wrong too.

 

Yep, and no stats required

[TheHurl]

Make Grant sit and watch every at bat of Pedroia's (I'm not sure why this concept entertains me so much) and he'll still project him to bat .140 and die a horrible death. Look at the two Gose threads right now, how many people have said things like Gose is the type that will be better in the majors than in the minors. I also heard a lot of Toronto media saying the same thing. There is a very small percentage of players that are (especially right away) but Gose is suddenly expected to...mostly because most fans, scouts, media, experts want him to be the guy they thought he would be.

[Nox]

Yep, and no stats required

 

Alright, so a projection system appears to be your equivalent in it's worst case scenario. What about the rest of the players in the baseball universe who don't have their spine explode? Sure you might be "close" from eyeballing some basic numbers but you'll do worse on the whole.

 

The whole "it takes too much time to look up their projections" narrative is ridiculous when it can be found an inch below where their past performance is (something you'd look up if someone asked you how you thought Edwin would do this year).

[GD]

Alright, so a projection system appears to be your equivalent in it's worst case scenario. What about the rest of the players in the baseball universe who don't have their spine explode? Sure you might be "close" from eyeballing some basic numbers but you'll do worse on the whole.

 

The whole "it takes too much time to look up their projections" narrative is ridiculous when it can be found an inch below where their past performance is (something you'd look up if someone asked you how you thought Edwin would do this year).

 

Please, you just know he looks up stats on Baseball Reference.

[Nox]

Please, you just know he looks up stats on Baseball Reference.

 

This seems more likely:

 

http://www.tsn.ca/mlb/statistics/

[jaysblue]

This seems more likely:

 

http://www.tsn.ca/mlb/statistics/

 

The hockey minds at TSN know their stuff.

[Nox]

The hockey minds at TSN know their stuff.

 

Without the rights to show hockey next year Bobby Mac might have to become a big crossover star. Lookout Tabby, Big Bob has you in his sights.

[GD]

Without the rights to show hockey next year Bobby Mac might have to become a big crossover star. Lookout Tabby, Big Bob has you in his sights.

 

And hopefully Dreger will just get canned.

[Nox]

And hopefully Dreger will just get canned.

 

Seems like an ideal candidate to be slowly molded into Vic Router's replacement for the 2031 curling season.

[connorp]

With the most popular projection systems being ZiPS and Steamer, I'm wondering just how reliable are these formulas? I've found with young and old players the projections can get a bit wonky. These formulas are a best guess using the intelligence and historical data available but nobody would admit them to be near perfect. Here's a three examples of projections that are looking pretty off:

 

Brad Miller - 24 y.o. Great milb stats and performed well in 1/2 of 2013, but has faltered and shown no real signs of improvement in 2014, BABIP is low but consistently.

Current 2014 wRC+: 34

Projected 2014 wRC+: 90 and 97 (ZiPS and Steamer RoS, respectively)

 

Alfonso Soriano - 38 y.o. BABIP and other peripheral stats pretty normal, just not performing as well as expected.

Current 2014 wRC+: 78

Projected 2014 wRC+: 105 and 90 (ZiPS and Steamer RoS, respectively)

 

Joe Nathan - 39 y.o. K's are down, BB's are up, BABIP is lower than average, velocity down about 1 mph

Current 2014 FIP: 4.82

Projected 2014 FIP: 3.42 and 3.19 (ZiPS and Steamer RoS, respectively)

 

Idk.. Feels like your taking ax extremely, extremely small sample size and making a generalization on it. With young players especially, just in common sense.. I'd probably trust jamming the numbers of thousands of players into a machine and cranking out a prediction over a scouts opinion. How bad do scouts miss on tons of guys.. Lawrie, Snider, Wallace, Ackley, Miller.. You could go on forever obviously.. So it's clearly better to take some sort of average factoring in all the guys that failed.. Even so, obviously it'll be wrong all the time but I don't know how you could think the scouts consensus would be any better..

 

I think it's safer to question projections when we have plenty of data to look at. I think I made a thread of this sort in off-season. One prime example was Clayton Kershaw. Think he avgd 6.5 WAR the last 3 years and was projected at 3.6 or something. I mean in times like this I question the system. I also understand right now the computer is looking good in Clayton's case.. But really.. How many ppl would take the under with him on 3.6 pre-season.

 

I guess you have to account for injury of course..

[TheHurl]

My projection for Fielder was that he was a fat phuck who was likely to break down sooner rather than later, and that, when his performance declined, it would be fairly dramatic and persistent

 

Steamer's projection for Fielder was "5" or some other number around his 2013 and 2012 numbers.

 

yes but much like me, you probably predicted this every year from 2010 on.

[GD]

They are meant as a way of comparing players. So why wouldn't you use them for that purpose?

 

stats r scary1!!!!

[eastcoastjaysfan]

Please, you just know he looks up stats on Baseball Reference.

 

You make it sound like B-R is an absolute joke of a site. Is it as good as FanGraphs? Of course not, but it's still a fantastic website.

[GD]

You make it sound like B-R is an absolute joke of a site. Is it as good as FanGraphs? Of course not, but it's still a fantastic website.

 

Eh. I don't like it. It has its uses, though. Just not a fan. I don't hate it for the statistics, I hate it for what a pain in the ass it can be to pull data.