2016 Fantasy Baseball Rankings (w/ Steamer Proj’s)
It’s Holiday time. Here is my Holiday present to you: 2016 Fantasy Baseball Rankings and $ Values using Steamer Projections and NFBC’s Format: (2)c,1b,2b,3b,ss,ci,mi,(5)of,u,(9)p and (17)bench spots.
For review, I’ll discuss the approach after the rankings, but you can check out Zach’s FVARz series of posts as well. Zach and I do a few things differently, but I use this template over others, such as SGP because I like how hands-on it allows you to be when ranking your projections. It allows for both pragmatism (drafting a certain number of players within a specific position for your league format) and objectivity (focusing on the Z-Sum/5×5 value of each player). I will describe below:
There are three tabs: Hitters, Pitchers and the posted “$” for dollar values and combined Hitter-Pitcher rankings.
The Approach:
I’m going to attend only to hitters, but the same approach is used for pitchers (separating SP and RP from each other though…feel free to comment with questions).
I exported each position within Steamer’s projections, removed duplicates and players that likely won’t reach the majors/will retire. I also associated players to their most scarce positions. DH’s, such as David Ortiz are with 1B’s. For your convenience, in the hitters tab, you will find the 1B’s highlighted in Red if they DH only.
Finding Means and Standard Deviations for Z-Scores:
I combined all hitters and sorted by Plate Appearances. For 40 roster spots * 15 teams, I determined that in general, 360 hitters will be drafted. I therefore sort by PA and pull the mean and standard deviation for each 5×5 stat (HR,R,RBI,SB,AVG) for the first 360 hitters (hitters with 237+ PA).
The 5×5 Z-Sums are in the 10th column (in the Hitters tab).
Position Adjustments:
Now comes the pragmatic objectivity. Stick to the Hitters tab. It’s currently sorted in the way that will make the most sense to you. For the means and standard deviations, we used the universe of hitters that will be drafted (360). Here, we are looking at the number of hitters that will be on active rosters (210) = 14 hitting roster spots * 15 teams. There are therefore 210 hitters “Above Replacement Value.” We have to find the replacement value/player within each position; take their Z-Sum and adjust all other players in that same position by that Z-Sum.
For example, in this format, approximately 30 1B will be drafted for 1B, CI and U slots. The replacement player is therefore the 31st 1B (Justin Bour). Bour’s Z-Sum is -0.12. We therefore add .12 to him and all other 1B. His “PosAdj” (column 11) value becomes 0.00, which will turn into $1.00 when we associated a dollar value.
The objectivity is adhering to the Z-sums. You don’t want to put another few 1B’s above replacement value, because the next 1B will have a Z-Sum that approaches our 2B/SS replacement value, and we know that middle infielders are more scarce than 1B. Scroll to the right in the hitter tab for the replacement players and associated Z-Sums used for the position adjustments:
C: JR Murphy = -5.04
1B: Justin Bour = +.12
2B: Johnny Giavotella = -0.58
3B: Jake Lamb = -0.63
SS: Brad Miller = -0.86
OF: Rusney Castillo = -0.61
SP: Jeremy Hellickson = -1.56
RP: Jason Grilli = -0.98
Now you should know exactly what you’re looking at in the Hitters (and Pitchers) tab.
$-Valuation:
The Auction conversion is right from Zach’s series: [(260-(1*23))/23]*(FVARz / average FVARz for above-replacement players) + 1.
260 = Budget
23 = # of active players per team
FVARz = Position Adjusted Z-Sum
Average FVARz for above-replacement players = the average Position Adjusted Z-Sum for all hitters and pitchers “above the replacement value” i.e. our active roster players. Our avgFVARz = 3.27.
For example, Mike Trout had a 5×5 Z-Sum of 12.70. It was adjusted by Rusney Castillo’s -0.61 to jump up to 13.31. This outputs $42.95 = [(260-(1*23))/23]*(13.31/3.27)+1.
Results:
The embedded file was updated on 12/22 at 10:32 EST.
Miguel Sano was moved from 3B to 1B (highlighted in red in the Hitters tab since he is currently Util-eligible only).
Keep in mind that Steamer has known closers at 65 IP and 28 Saves. In reality, Dellin Betances would sky rocket in value with more IP and more saves if the Yankees trade Andrew Miller.
Matt Harvey, Jacob deGrom and Noah Syndergaard are 18 spots apart at #26 to #43. That’s awesome. Let’s go Mets!
Daniel Schwartz contributes for RotoGraphs when he's not selling industry leading thermal packaging. You can follow him on twitter @RotoBanter
Troy Tulowitzki will go undrafted? Can I join your league?
What do you mean? He’s above replacement. He’s the 8th SS according to these steamer projections.
Weird. I searched for him like 3x through the list and did not see him. Apologies.
Many thanks DS. This is great and will give me a lot to think about during this slow week of work before a holiday break.
One question, this seems to project Capps as the closer in Miami. Has there been any indication that Ramos will be supplanted or tried at SP?
Good ?. I would ask the bullpen guys once they start with their bullpen reports. I’d think it’s still Ramos. Same with Miller and betances – I think yanks might actually trade Miller which will skyrocket betances’ value in saves-only leagues.
For now id think it’s still Ramos unless he’s traded b/c he’s arbitration eligible.
do one for OBP since only antiquated leagues use AVG still
Closer to the season I’ll do one for Obp, sure. I’ll do rankings for other formats too like ESPN and maybe yahoo.
Please, and you mean like in the next few weeks “closer to the season,” right? I really wish I could find more OBP information out there, but most of it is pay site only. Most of the time I have to tinker with stats and guess. More OBP and keeper please.
Some of the positional eligibility are incorrect (Sano should only be DH-eligible), which will have an influence on replacement level.
yeah at this point He should only be until eligible according to nfbc. Unless plouffe goes he might eventually gain OF eligibility but at this point who knows. I’ll do another closer to the season.
Instead his zsum should be adjusted by .42 instead .57 which wouldn’t kill his value – might only drop him a few in the rankings
Sano will be 3B eligible in Y! leagues.
I hope this doesn’t come across as snarky, but it’s an honest question: I know you mentioned Donaldson was too low, but was his ranking a function of some quirk in the ranking system or is the top-20 field just that competitive?
I had to ask because I noticed that 7 sb’s seems to be the one and only difference in projections between donaldson and bryant… is that enough to place donaldson 22nd instead of 13th?
Also: I realize position scarcity plays a factor, but that still seemed really strange to me.
Just think the projection is low on Donaldson. I’m assuming all counting stats will be closer to last year.
Fair enough. I decided this offseason I’d try and learn the inner workings of some of the projection models, so that’s why I was curious.
Could you explain the process for separating SP and RP? Finding averages etc.
Thanks!
Sure. Between the 9 active P’s and 7 extra bench spots filled with pitchers, we are finding The means and SD’s for 240 pitchers. I’m on my phone but I think it was 11 starters and 5 relievers for each team. So 11*15=165 starters and 5×15=75 relievers. Take the means and Sd’s of the top 165sp and 75rp innings pitched totals and find the zscores for w, era, whip, k and sv’s. Zsum and Then see the pitching tab for the position adjustments.
Thanks!
Position scarcity only matters for catchers in 2-catcher leagues.
I would agree to a degree. 5 OF leagues * 15 teams make OF well scarcer than 3 OF leagues with 12 teams. Especially if those leagues also have mi and ci. In those leagues I won’t focus on OF after the elites are gone.
In a 15 team league when you rank 210 hitters, most of the bottom 10 or so will be catchers – that is because they’re worth less than a buck. For 12 team leagues the adjustment is even more defined. The reason you adjust for catchers is because the alternative (alternative being you get a guy worth < $1) is an "overbuy" – you pay a buck for a guy worth less than that. There a few – you can count on one hand few – other position players that are worth less than a dollar. For that reason you don't need to adjust the other positions. Notice how those other adjustments are very close to each other (.38 to .65) whereas the catcher is massively different (5.57)?
To add to that, I can’t see the spreadsheet right now but catcher adjustments can be highly sensitive to the catcher projections at the bottom of the catcher list (i.e. the Iannetta’s of the world). This is why Posey and Schwarber are so high in your adjusted numbers.
They are sensitive to the poor options at the bottom of the catcher list, yes.
My 2 cents… historical averages are better for finding means/stdevs IMO. I’ve been using the FG depth charts for SV projections because they actually account for quality of team (as opposed to the steamer 28-across-the-board).
The C adjustment is so crazy for 2-C leagues. It kind of makes me want to try using ‘average value of rostered players’ to position adjust instead of ‘value of the final player rostered’.
Matt Duffy is listed twice on 3B’s…both above Pablo Sandoval and above Alex Guerrero.
There are two Matt Duffy’s now in the majors.
Quick question: Is there a reason that Lindor’s AVG is so low? Seems to me like his profile would allow him to maintain high BABIPs (good speed, sprays balls to all fields, and a good amount of LD’s and GB’s) even if a .348 isn’t sustainable number, his MiLB numbers seem to indicate that .310+ is likely and that we should probably expect a batting average almost 10 points higher than projected. Combine that with a solid K%, and it seems to me like Steamer might be a bit lower on him than it should be. Everything else in his line looks legit though.
If he were to hit ~.280, how much would his value go up?
I’m just using Steamer projections – I’d agree it could be higher but I think consensus is he won’t be as special as he was: http://www.fangraphs.com/blogs/what-can-we-make-of-francisco-lindor/
Would anyone justify keeping these 6 in a 14 team league (must keep 1 closer).
Harper, Dee Gordon, Correa, Bumgarner, Sale, Melancon/Jansen and letting Braun go back into the pool?
I can justify it easily.
If the rest of your team consists of Billy Butlers, those keepers look fantastic.
Thank you.
Not sure if it was done, but the statistician in me always asks: in the z-score did you weigh the rate statistics by opportunities? (the appropriate choice of at-bats, plate appearances or innings pitched)
Logically a 1.000 hitter with one at bat only contributes one extra hit in one at bat. Their z-score should be equal to the difference between a player with 179 hits in 599 at-bats (0.2988) versus a player with 180 hits in 600 at-bats (0.300).
Without the correction the better players who don’t play as frequently (i.e good catchers) get rated too highly.
The corrected version is actually:
Z = (batting average – league batting average)/stdev of b.a. * at-bats / league average at-bats
The steamer projections on this website do not match the projections on this spreadsheet. Look at Mookie Betts as an example.
Probably just an update b/c I pulled from link that is in the post early last week
Not that I don’t understand the projections will change over time, but I’d think we’d want them to match this post for at least a little while…
In the case of Betts his new projections give him +3 HR +17 R and +14 RBI over the above. That’s considerable. Looks like the AB/PA for him changed dramatically in the update, perhaps Steamer updated their AB/PA projections.
Lucroy…wow. I was thinking he’ll go in the 8th round this year, I guess he’s a steal then.
I’ve used SGP (or most likely a dumb version of it) myself. I’ve gotten similar results that play up speed, catchers and SP.
I know you’re using Steamer so I am not questioning the underlying projections .. but just looking at the top two SS and struggling. We got Correa with +10 homers and +30 RBI compared to Reyes, while giving Reyes a few steals and about 20-25 points of batting average? Yet Reyes is $6 better in this valuation. That’s .. interesting.
Yeah I’m going to pull steamer updates at some point this week and re-embed. For now I guess Colorado and that batting average over that overly high number of plate appearances does the trick. Baffling!
I remember in previous years weighting the zSUMs for the rate stats (BA, ERA, WHIP) with the PA or IP. Thoughts on this? Is that already done here?
Yes of course – you can download the file and unhide the columns to see that the raw initial z’s were adjusted by IP/PA for BA, era and whip
updating this in 10 minutes ….stupid zsum incorporated the wrong column! i.e. i incorporated the wrong column in the zsum calculation. will do the trick.
Embedded file updated at 10:32 CST this morning, 12/22!
Are you going to have auction values for nfbc?
They’re in the embedded file/sheet shown in the last, column “$”
Where is Joey Votto? I can’t access the document at work, so I’m trying to view it on my phone and can’t seem to find him.
60th overall
Zach Greinke didn’t have a 3.07 ERA in 2015! Are you using career averages? If not, maybe 2014 STATS!?!?!
These are projections from the future. That future is next year. Your future is now.
Do your position adjustments account for the utility spot for hitters? How do you do this?
There’s so few util-only’s that I stick them in 1b. They remain highlighted in red ( their “1b” is in red noting that they’re until only).