Thursday, December 13, 2012

Playing with the Projection Machine

So after a couple of projections caused a bit of a stir earlier this week (you can read Royal Review's here and view Replacement Level Yankees' here), I decided to tinker around with the wonderfully nerdy spreadsheet that was provided to Royals nation by Royal Review writer Jeff Zimmerman.

In Zimmerman's projection from earlier this week, he utilized Bill James' individual projections to determine how many wins the Royals should record during the 2013 season. Surprisingly, the projection system spit out a 94-88 record for your Kansas City Royals. Obviously, this caught the attention of the Royals Twitterverse, but many were quick to point out how rosy Bill James' projections are on a yearly basis.

I decided that I wanted to plug a different set of numbers into the projection machine to see how much that win total would swing. Ultimately, I decided to remove subjectivity from the equation as much as possible. Eventually, I decided to simply plug in the rate stats of 2012 while keeping the plate appearance and innings pitched breakdown that had already been provided by Jeff Zimmerman in his Bill James projections. To put it simply, this is projection attempts to quantify what would happen if last year's performances were replicated with a healthy Salvador Perez and Lorenzo Cain and the additions of James Shields, Wade Davis, Ervin Santana and Jeremy Guthrie to the rotation.

For all of the pitchers, I decided to input ERA instead of FIP or xFIP. It could be argued that either FIP or xFIP would be more indicative of true performance, however, I wanted it to be on the conservative side and I found that most of the Royals staff's ERA came in higher than their FIP or xFIPs. This was particularly true in the rotation where Santana, Chen and Hochevar's ERAs were much higher than the Fielding Independent numbers. If anything, this likely made the win forecast more conservative.

I made three exceptions to simply using last season's ERA. First, I switched Wade Davis's ERA to the 2011 4.45 instead of last season's 2.43 ERA. I'm hopeful that Davis's experience in the bullpen will make him better in the rotation, but if nothing else it seems conservative to use his ERA from the last season in which he was a full time starter.

The final two switches I made were to change Felipe Paulino's and Nate Adcock's ERAs to their FIPs from 2012. I did this because both of their ERAs were much better than their FIPs due to small sample sizes and I didn't want this to play a huge role in swaying the projection in a positive favor.

The final note, is that I used Jeremy Guthrie's full season ERA instead of his Royals ERA.

So what was the result?

85-77

Not too bad.

So how much value could certain elements bring to the table? Here's a quick rundown of some things that could happen and how they affect the overall projection:


  • Alcides Escobar's defensive  matches the 2011 version instead of 2012: 87-75
  • Eric Hosmer performs at his 2011 level: 88-74
  • Jeff Francoeur performs at his 2011 level (yeah right): 89-73
  • Luke Hochevar performs at his 2011 level: 86-76
  • Bruce Chen performs at his 2011 level: 87-75
  • Ervin Santana performs at his 2011 level: 87-75
  • Jeremy Guthrie ERA is his career average: 86-76
Not too surprisingly, the biggest change from the initial projection would be if Jeff Francoeur could recreate his 2011 season. Hopefully, Francoeur can channel his inner Alex Rios and achieve just that. Keep in mind once again, the initial projection was based almost entirely on 2012 results. Also, notice that the what ifs that were just provided aren't pulled from a hat. All of them are grounded in past performance and all but Jeremy Guthrie's are based off 2011. 

Basically, what Kansas City needs is health, repeats of 2012, and a breakout or a rebound. If those three things happen, based off of these projections, there's a good chance we will see playoff at the K in 2013.

Again, it is incredibly earlier to be toying around with projection systems and of course they all need to be taken with a grain of salt. But if you are a numbers guy like myself and if you geek out over baseball statistics it is always a bit fun to manipulate the data. Thanks again to Jeff Zimmerman for the awesome spreadsheet. Let's hope it is an accurate one. If it is, 2013 could be hell of fun for Kansas City.

Oh and by the way,  if all of above happens the projection calls for a 99-63 season. 

4 comments:

  1. I wish... 27 years of suffering !

    ReplyDelete
  2. I wish ... 27 years of suffering makes me doubt the Royals, it's like a person with bad breathe.... They stink !

    ReplyDelete
  3. Nice work. What factor in the 2012 season stats for Escobar or was it to 2011 or a blend? I hope it comes to fruition because if it doesn't we have sacrificed 2015 and beyond with the Shields trade. My "eye" test tells me that Hosmer and Moose will not be able to bounce back but what do i know. I do believe that Sietzer's approach did not work for everyone so maybe the new hitting coach can help. We will need it or we won't score enough runs. Another reason we should have given Giovetella more AB's to know if his hitting would be of more value than Getz's "perceived" value the FO.

    ReplyDelete
  4. The original projection had Esocbar's 2012. However, in 2012, Escobar graded out really bad defensively. This is probably more flukey than anything else, so I wanted to take his 2011 defensive metrics and couple it with his 2012 offensive performance.

    ReplyDelete