Tuesday, January 26, 2010

FBJ's Top 100 Keepers

Recently, FBJ had a post about some of the best keepers based on draft position for the 2010 season. However, our editor over at Fantasy Ball Junkie recently queried the rest of us about putting together a list of the top keepers coming into the 2010 season. I told him I could probably put something together that not only counted the draft slot surplus (or places moved up in Median Draft Position from 2009 to 2010, which is an article already posted at FBJ) but also weighted it by the expected gain needed to move up each of the draft slots. For example, moving from #75 overall to #10 overall is most likely better than moving from #165 overall to #100 overall. The methodology is explained in the article here. And the Top 100 list is here. In addition, I thought it would also be important to ensure we weight by conservative projections for each player in order to not put superb 1-year wonders all the way at the top (for example, J.A. Happ comes in at #53 and Elvis Andrus at #59).

It's important to keep in mind that the Top 100 only includes the Top 200 draft picks for 2010. Any player chosen beyond the 200 mark in the 2010 Mock Draft Central drafts is not included. Assuming a 12-team league, this seemed reasonable to us, since moving from the 300 to the 245 spot isn't really very significant of a talent jump, and there's probably plenty of other guys around there. For this reason, interesting keepers like Randy Wells (315 ADP) are left off in favor of the likes of Kevin Youkilis and Torii Hunter. Any player that moved down in the draft was not included (obviously, since you can get him at a lower price than last year). This fact may make the list somewhat controversial. What's important is to focus on the top 60 or so on the list.

My original inclination was to run a censored/truncated regression on the draft positions. Unfortunately, due to the non-independent nature (and ultimately, zero-sum nature of draft movement) this wasn't really possible. I'm still looking into more non-parametric ways to do something like this, but they're currently beyond my scope.

I began by creating "10-slot weights" by calculating a matrix of the average differences in talent(based on CHONE and FBJ's player rater) from the Top 10 Picks and the 11-20 Picks, the 11-20 and the 21-30 Picks, and so on for any difference (by group of 10) possible. It should be noted that FBJ's player rater values steals a lot higher than most would, so this is one reason we see Michael Bourn so high.

Interestingly, there were some NEGATIVE values when moving up in the draft in terms of expected performance for 2010. Because of this, I just roughly smoothed out the differences for those areas with a local linear regression of those jumps around that point (or, really, a simple average). The higher the difference, the more difficult the move up that far in the draft. Using this, I multiplied the number of draft positions moved up by the given weight. This is a rough score of weighted draft position.

From there, I felt it necessary to also weight this rough score by the player's projected performance for 2010. It's important to also somewhat take into account what you're getting for the price, rather than just what a bunch of other people think. Because the player rater is built to cap around 10 (though, Pujols put up a whopping 13+ last season), I didn't simply multiply by this number, but used "RaterScore^1.5". This way, we can exaggerate the expected performance for better players in order to push small increments up a little bit.

From there, I was torn. Using this simple calculation left me with Joe Mauer down at the 60th-ish spot for keeper value and Justin Upton at only #16, while Michael Cuddyer came in at #12 and Kurt Suzuki came in at #19. I'm not convinced that's accurate. In order to further correct this, I got a little subjective. Since each player in the order 1-300 should be harder and harder to jump, given a certain performance, I used the following to divide the resulting score above:

(CurrentMDP/300)*(MDPJump/PreviousMDP)

In other words, my further scaling weight is the product of 2 ratios. First is the 2010 Median Draft Position divided by 300 (the censored cap for draft position). The other is the increase in MDP from 2009 to 2010, divided by the 2009 MDP. At this point, I wish I had more justification for this (and there may be a little better way, I'm still tweaking things). This gave me the scores shown over at FBJ, and they seem reasonable. However, I think they allowed the likes of Chase Utley (#29 going from MDP of 12 to MDP of 4) and Mark Texeira (#58, going from MDP of 11 to MDP of 8) jump up too high. However, I think it properly weights Mauer and A-Rod (who came at a HUGE discount for a lot of people last year) as some of the better keepers around. Anyway, I'd love to hear any thoughts anyone has on this, or a more methodological way of doing things. This was a lot of just sticking things together, and I think I got lucky that it turned out so well, with the following Top 20:

1. Mark Reynolds
2. Aaron Hill
3. Michael Bourn (remember, FBJ rater overvalues steals)
4. Matt Kemp
5. Justin Upton
6. Adam Lind
7. Joe Mauer
8. Alex Rodriguez
9. Kendry Morales
10. Zack Greinke
11. Troy Tulowitzki
12. Ben Zobrist
13. Chris Carpenter
14. Tommy Hanson
15. Prince Fielder
16. Felix Hernandez
17. Nyjer Morgan
18. Gordon Beckham
19. Andrew McCutchen
20. Jacoby Ellsbury

One last thing to keep in mind: this does not take into account likely breakouts or significant steps forward. Honestly, I think a guy like Yovanni Gallardo is in for a large jump in ability. He didn't crack the Top 100 on this list because he only jumped from an MDP of 115 to about 96. Given the talent from 96 to 115, Gallardo's conservative projection, and the fact that 15 spots just isn't that much that late in the draft, he doesn't get a huge score based on these calculations. The calculations also ONLY take into account the price paid in 2009, and DO NOT take into account the future possibility of keeping a player (in which case, I would say younger players are much better). This is ONLY a 2009 to 2010 keeper ranking.

No comments:

Post a Comment