2015 Projections

Discuss all things Cardinals Baseball
Post Reply
jim
Red Lobster for the seafood lover in you
Posts: 50608
Joined: May 1 06, 2:41 pm

Re: 2015 Projections

Post by jim »

Does anyone take Win Shares seriously any longer?

phins
Sobbing quietly during Fox programming
Posts: 10247
Joined: June 9 06, 3:51 pm

Re: 2015 Projections

Post by phins »

jim wrote:Does anyone take Win Shares seriously any longer?
I sure hope not.

User avatar
33anda3rd
Replies Authoritatively
Posts: 8418
Joined: April 7 13, 9:45 am
Location: Chicago, IL

Re: 2015 Projections

Post by 33anda3rd »

Famous Mortimer wrote:what I want is a Dave Cameron or a Szymborski to go through and pick out the top 20 guys by WAR and see how close they got.
I'd rather know which system on the whole is the most accurate. Problem is things swing around so much that no projection system is the best year-in/year-out.

If you're picking 20 good/20 bad, the only way it's really useful is if you're able to detect something that improves your system going forward. Like "ZiPS is too bullish on hitting prospects" or "Marcel does not have a long enough memory." With 20 good/20 bad the sample size is too low. The truth, and the fun, to me, is in the whole population.

User avatar
Famous Mortimer
Perennial All-Star
Posts: 3636
Joined: November 14 14, 5:23 am
Location: Cherokee

Re: 2015 Projections

Post by Famous Mortimer »

That's true, of course, but we rarely use any of the projection systems to look at the league as a whole. In terms of sites like this, Fangraphs, etc, projections are used for single players, to evaluate trades, yes? And it's seen as the be-all-and-end-all of discussion - I've lost count of the number of times I've read "well, Steamer has him as a 3.2 WAR player next year, so that's a stupid trade suggestion" (or words very similar).

If they're consistently getting the projections wrong for the guys who are earning the major $$$ by a lot, then I'd have thought that would indicate there's some problem with their system somewhere. But it might just be me who wants this. It's not like I lose sleep over it or anything.

User avatar
33anda3rd
Replies Authoritatively
Posts: 8418
Joined: April 7 13, 9:45 am
Location: Chicago, IL

Re: 2015 Projections

Post by 33anda3rd »

Famous Mortimer wrote:And it's seen as the be-all-and-end-all of discussion - I've lost count of the number of times I've read "well, Steamer has him as a 3.2 WAR player next year, so that's a stupid trade suggestion" (or words very similar).
There are a couple populations who think this way.
1. People who don't understand how the projections work and use them as iron-clad predeterminations of the coming year.
2. People who don't understand how the projections work and hate them because they think they're meant to be iron-clad predeterminations of the coming year.

Since they're not meant to be iron-clad anything, anyone who's using them that way is not someone whose opinion you want to waste your time with.

That's not to say the projections are not useful. When we get more projections out on FanGraphs and from PECOTA and CAIRO and Clay Davenport and we look at them all together we should start seeing not a Google Maps-perfect route for a player but a general "He's headed SSW in general" kinda thing. Here are a couple examples.

If you're a GM talking to your owner right now about whether you should sign James Shields, you're not talking about if he's going to be worth 3.7 WAR this year and 3.4 WAR next year and 3.1 WAR the year after, you're talking about if he's worth spending ~$20MM/year on for the next 3-5 years. And if you think he's in the ballpark of 2.5-4 wins/year on average the next 4-5 years, you're ok making the deal. Can you project to that level of accuracy often enough to get ahead? Yup. Doesn't matter if Shields is 3.9 or 2.9 WAR for the next 4 years for $20MM, because you're in a range there where the money is either fine or a small discount. You need not know to one decimal where he's landing, you just need to know that your quants have him landing in the right range.

Say you're a GM wondering if you should trade Josh Donaldson for Brett Lawrie and three prospects. Where do you project Donaldson in his age 29-32 years? For what cost in Arb1-4? Ok. Do you project yourself to be winning in each of the next four years? Are you paying him for 4 wins/year when that only makes you a 78-win team in the first couple years? Can you flip him for four players who will ALL have value in what would be Donaldson's Arb 3-4 years at far less money than Donaldson? Where do you project the prospects--what do the scouts and quants say? Where do you project the one MLBer coming back? What if the math is:
Year 1: Donaldson 7 WAR, Lawrie 3 WAR, Prospects 0 WAR, you think you will win 78 games without the trade, 74 with
Year 2: Donaldson 6 WAR, Lawrie 4 WAR, Prospects 1 WAR, you think you will win 82 games without the trade, 81 with
Year 3: Donaldson 5 WAR, Lawrie 4 WAR, Prospects 5 WAR, you think you will win 88 games without the trade, 92 with
Year 4: Donaldson 3 WAR, Lawrie 0 WAR, Prospects 9 WAR, you think you will win 87 games without the trade, 93 with
How do you not make that trade? You can't. Keeping Donaldson during years you think you're rebuilding and paying him to help you not suck is a bad decision. Getting two back-end SP who can get 2.5 wins/year in the future plus a SS who could be a 5-win player in your next cycle of competitiveness is the move. Depending on how you project.

User avatar
pioneer98
Hall Of Famer
Posts: 22250
Joined: July 15 08, 8:24 pm
Location: High A Minors

Re: 2015 Projections

Post by pioneer98 »

If you don't need it to be that exact, then probably any of these systems are good enough...and the thing that will make our break that Shields contract will be if he stays healthy or not.

User avatar
33anda3rd
Replies Authoritatively
Posts: 8418
Joined: April 7 13, 9:45 am
Location: Chicago, IL

Re: 2015 Projections

Post by 33anda3rd »

Predicting pitcher injuries, if you can do that, there are 30 seven-figure jobs for you. That's part of the projection not in ZiPS but in team discussions. I'm certain of that. The public systems don't add likelihood of injury, they just weigh numbers that include past injury. They need not factor in future injury because they're not spending the money. But I'm sure teams take that into account and it would be fascinating to read what they've found.

User avatar
Jocephus
99% conan clips
Posts: 63658
Joined: April 18 06, 5:14 pm

Re: 2015 Projections

Post by Jocephus »


User avatar
Jocephus
99% conan clips
Posts: 63658
Joined: April 18 06, 5:14 pm

Re: 2015 Projections

Post by Jocephus »

Image

http://cdnmo.coveritlive.com/media/imag ... uzzle2.png

open up the direct link, big image...but if you can solve it dan will release the winners team of choice for zips

User avatar
Jocephus
99% conan clips
Posts: 63658
Joined: April 18 06, 5:14 pm

Re: 2015 Projections

Post by Jocephus »

Team Projections and Last Season’s Statistics

by Jeff Sullivan - December 17, 2014

The other day, I wrote the Chicago White Sox still don’t seem very good. Naturally, that drew something of a negative response, and that’s fine — people ought to be excited, and it was kind of a buzzkill headline. The White Sox have been active, and general manager Rick Hahn has succeeded in turning nothing into something, at least as far as 2015 is concerned. We all recognize there’s a reason they play the games. Who knows what might happen? Who knows what else Hahn might eventually do? Yet, within the comments, something caught my eye. Some people think Steamer projects too much regression with the White Sox. Which got some gears whirring: How do our current projections compare to roster projections using only last year’s stats?

Obviously, if you’re trying to predict Year X + 1, you need to look at information from more than just Year X. Different people will recommend looking at different windows, but as a rule of thumb, you want to consider at least three or four years, if the data’s there. Plus, there are still other things to take into account. But while Year X isn’t the only thing that’s important, Year X is also the freshest data set in memory. So when a projection differs from what literally just finished happening, people might be prone to thinking that something’s amiss.

The way I figure, we have projected information, and we also have perceived projected information. The former, right now, comes from Steamer. The latter, meanwhile, is influenced very heavily by what happened in 2014. So I wanted to try to find the biggest differences. There’s no right way to do this, but I decided to use our author-maintained, team-by-team depth charts. Those depth charts provide something like expected playing time. Then I opted to simply plug in 2014 WAR figures for each player, adjusting for plate appearances or innings. I set, as 2014 minimums, 100 plate appearances or 25 innings. I used 0.0 WAR for everyone else. For most of those guys, expectations will be low, since they didn’t do anything in the majors last summer. It doesn’t work for everyone — Matt Harvey, for example – but it works in most cases.

Before getting to the table, let me note this was written Tuesday night, and the Jimmy Rollins and Matt Kemp trades still aren’t official. Also, there’s talk of a Wil Myers thing? In case that’s happened by morning, it wasn’t not happening as I was writing this. So. The table. There’s Steamer projected WAR, and the league rank. Then there’s “projected” WAR — using projected playing time and last year’s statistics — and the league rank. Finally, there’s the second WAR, subtracted from the first WAR. Here, we might see differences between objective projections and expectations influenced by recency bias.

The sums don’t match up perfectly — there’s more total WAR using last year’s numbers. It doesn’t matter.
http://www.fangraphs.com/blogs/team-pro ... tatistics/

Post Reply