For my draft board, I'm using a composite of a few projections. I thought it would be interesting to take a look at how the different projection systems viewed our keepers.
First, a few assumptions. Many of these systems had back end keepers listed as negative values. In order to prevent very strange things from happening to the overall valuations, I've simply listed players as being worth $0 if the projection has the below $0. Second, these values are calculated assuming 5 SP, 2 RP, 1 P - which matches the way we tend to fill out our rosters. For whatever reason, a lot of these projection systems seem to think you can go all middle relievers and be competitive in pitching, despite what common sense would tell you. I think it may have something to do with many of them being very conservative about wins. While wins are inherently hard to predict, I think the conservative projections minimize the difference in reliever wins and starter wins.
To start out, Luke is #1 in two of the three projections. Unsurprisingly, Mike Trout leads the way. Depressingly, the value of Mike Trout surpasses the total value of keepers for 5 of the 7 other teams, just edging out Andrew and Mark, who come in 4th and 5th, respectively.
Taylor comes in at #2, thanks to his impressive array of value keepers, all at $11 or lower. ESPN's projections are unfavorable for de Aza and Escobar, in particular, but Taylor's overall value is fairly consistent across all three projections. Across the board, all three systems rate Cespedes as Taylor's best keeper.
As far as my keepers go, it all depends on who you believe. ESPN projections adore my keepers, in part because those values seem to highly value elite players, and have me at 2nd overall in terms of value. Baseball Prospectus, on the other hand, is not a fan of Beltre, Butler, or Santana, and pegs me at having the second WORST value overall. BP actually has those three players as providing negative value, with their salaries outstripping expected returns by almost $10. The total difference in valuation is an astounding $50. All three agree that Edwin Encarnacion is by far my best keeper.
In fourth we have Andrew, Mr. Consistency, despite his former team name. Amazingly, there is a $4 difference between his most favorable projection (ESPN) and his least favorable (BP). In fact, all of his players are remarkably consistent, falling in a fairly tight window of values. BP departs from the other two by rating Ellsbury as a positive and is very down on Willingham (but still projects value), but otherwise there is a great deal of consensus. Alex Gordon comes in at #1 in all three projections.
In fifth, we have Mark's keepers, led by Austin Jackson at $6. None of the other players provide a substantial amount of value individually, but he also has no real holes in his keepers. Mark also is in good shape to fill difficult positions. Both outfield and firstbase were decimated by large numbers of kept players, plus relief pitcher and shortstop are always a disaster. To top it off, Zobrist's ability to play SS, 2B, and OF gives him spectacular flexibility in the draft to pick up values as he sees them, without worrying as much about positional eligibility.
In sixth, Caleb stockpiles elite players, but also pays a good deal for them. While that does limit his upside, he has an enviable core of an elite first baseman, starting pitcher, and catcher to build around. Prince Fielder is rated as his top keeper, followed closely by Price and Milone.
Kate comes in at 7th. In her case, her overall valuation is hurt by a very low rating from Baseball Prospectus, how does not see good things coming for Scherzer (negative value overall) or Viciedo. She can take solace in the fact that the other two projections have her keepers much more highly rated, with solid returns for Peavy and Victor Martinez, her best keeper overall. She also had the second highest amount of variability in her keeper values. As interesting note, Ian Kinsler is her most consistently valued player, something that shocked me considering how up and down his career has been. However, all three projections have him in a $3 window. His actual production over the past several years has not been nearly so consistent, though, and Kinsler could provide anything between $15 worth of value and a $5 loss without me being too surprised.
Finally, we come to Spencer. Spencer has some nice value, particularly in Desmond Jennings. Unfortunately for Spencer, Steamer projections really don't like his keepers - rating them as only providing a meager $2. They are very low on Anderson ($3 value) and don't rank Ogando.
11 comments:
Not Last? That's what's UP!
There are a lot of players I've seen a big spread on, particularly Alcides Escobar and Doug Fister. That's one of the reasons I traded them. With Ellsbury and Lawrie already in the fold, more uncertain keepers seemed to be a bad idea.
Always nice to remember the blog exits, check in, and see a McDowell post with my team last...
Player Rater does the same thing with middle relievers.
Don't blame me! I didn't create the projections, I just compiled them!
Luke - not sure what you're getting at with middle relievers. Are you saying player rater overrates middle relievers? Granted, when it is all said and done, players like Jake McGee end up ranked higher than they would in any draft, but that doesn't mean they weren't major contributors. It just means that you can't predict which reliever will have an ERA of 1.50.
I guess I'm comparing apples and oranges, since player rater scores known performance, and you're talking about projection systems.
I'm not sure whether or not player rater overrates middle relievers, I just know that it gives the successful ones a much higher score than I would expect. There are a ton of middle relievers in the top 192 last year. I don't know exactly what the algorithm is for Player Rater, but I think it is based purely on the final stat line, with no correction for how many weeks it took to accumulate that line. McGee's 55 innings with 73 Ks and 5 Ws would be an amazing stat line to see from one of your starting pitchers halfway through May, but when you spread that same stat line over a full season it's less valuable.
Ah, I found an example to make a direct comparison. Joe Smith (reliever) and Andy Pettite (starter) finished with very similar player rater scores (1.60 and 1.78), and very similar final stat lines. However, I think Pettite was much more valuable, since he put up his numbers over 2 months, instead of over the whole season. You could've started Smith most of the season and benched him for Pettite while Pettite was healthy, and then plugged Smith back in after Pettite got hurt. You'd get all 75 of Pettite's great innings, plus 30-40 from Smith. The credit for you being able to do that goes to Pettite for condensing his value in a short time frame.
I guess I'm comparing apples and oranges, since player rater scores known performance, and you're talking about projection systems.
I'm not sure whether or not player rater overrates middle relievers, I just know that it gives the successful ones a much higher score than I would expect. There are a ton of middle relievers in the top 192 last year. I don't know exactly what the algorithm is for Player Rater, but I think it is based purely on the final stat line, with no correction for how many weeks it took to accumulate that line. McGee's 55 innings with 73 Ks and 5 Ws would be an amazing stat line to see from one of your starting pitchers halfway through May, but when you spread that same stat line over a full season it's less valuable.
Ah, I found an example to make a direct comparison. Joe Smith (reliever) and Andy Pettite (starter) finished with very similar player rater scores (1.60 and 1.78), and very similar final stat lines. However, I think Pettite was much more valuable, since he put up his numbers over 2 months, instead of over the whole season. You could've started Smith most of the season and benched him for Pettite while Pettite was healthy, and then plugged Smith back in after Pettite got hurt. You'd get all 75 of Pettite's great innings, plus 30-40 from Smith. The credit for you being able to do that goes to Pettite for condensing his value in a short time frame.
I think player rater must implicitly assume that each player gets put in your starting lineup the whole season. That's why it gives Sergio Santos a -2.17 score last year. Most of that negative score doesn't come from ERA and WHIP, but actually from Ks (-0.89) and Ws (-0.78). It doesn't make any sense to give large negative scores for those categories for a guy who in reality only spent a couple weeks in your lineup.
Luke is right. The player rater doesn't really deal with partial seasons very well. Ideally, there would be some sort of fantasy replacement level, and for missed time, it could add it in. I mean, no one is going to start Sergio Santos all season. It also doesn't really address roster spot scarcity, but it's a fairly good way to look at two guys with similar playing time.
Absolutely. In a similar vein, Nolan Reimold, for example, was a contributor for Spencer last year despite his poor showing on the player rater. He hit a bunch for the two weeks he was playing, then didn't do anything the rest of the year.
BUT, I think we tend to underestimate the impact of excellent middle relievers on pitching stats, in part because they never fare well in projections. Thanks to his sub 2.00 ERA and sub 1.00 WHIP, I do think that last year McGee was more valuable than Matt Moore, as the player rater indicates.
As best I can tell, the player rater is based exclusively off of Z scores, plus some kind of weighting for innings pitched. The extremely low ERAs that middle relievers can have produce very, very high Z scores, leading to their high overall value. That, coupled with the fact that replacement level strikeouts is actually pretty low and a few lucky wins, and you've got a top 20 arm, at least looking at the past.
Post a Comment