2007 Brewers: Mediocre Posers or Crew of Champs?

About a week ago, I linked to the projected 2007 standings provided by two of the most respected sites for baseball analysis, The Hardball Times, and Baseball Prospectus. The former has the Brew Crew finishing 3rd, at 78-84 and behind the Cardinals and the Cubs. The latter has them tied with the Cubs for the NL Central title with a record of 85-77. (Note: At the time of my aforementioned blog entry, the Brewers were actually one game better than the Cubs and projected to win the division, but since PECOTA is a fluid thing, it’s changed since then. The Cubs became one game better while the Brewers projected record stayed the same.) Regardless of where they end up in the standings, a that’s a 7-game difference in the records. No matter how weak the division is, 78 wins probably isn’t going to cut it for the title. 85 wins, on the other hand, could very well win it for the Crew. Let’s look at the difference between how The Hardball Times and Baseball Prospectus arrived at their respective projections in order to get a better idea of what might need to go right for the Crew to win the division, what might be waiting to go wrong, what we really have to look forward to as Brewer fans in 2007.
First, let’s look at how each site arrived at their projections. The Hardball Times starts with the projected WAR (Wins Above Replacement) for a team’s players, which “were the result of several months’ work by David Gassko and Chris Constancio” and rely heavily on Baseball Info Solutions’ batted balls stats. In order to projects a team’s record, John Beamer then adjusted the WAR projection for each player to better fit expected playing time, so that each team’s pitchers only totaled 1440 IP, and the position players totaled 6165 PAs. He then established a “reasonable” replacement level and viola: a W-L record. Baseball Prospectus keeps more of their magic to themselves, but I’m sure they do something fairly similar to that, using player projections to create a projected RunsScores/RunsAgainst for each team, and then making a Pythagorean record based on that. The real difference is in the player projection method. PECOTA, Nate Silver’s baby, relies more on comparable players, player types, age, handedness, etc. than anything. So far, nothing has approached its accuracy.
Now that we see how(sort of), let’s look at what(hopefully). The first thing I though of when I saw this difference in standings was Corey Hart’s generous PECOTA. Obviously that’s not going to account for 7 wins, but it’s a place to start, maybe accounting for one win. His weighted mean PECOTA calls for a .870 OPS and 3.6 WARP in 553 PAs. The Marcels, Tango Tiger‘s invention featured at The Hardball Times, project Hart at a .798 OPS in 334 PAs. While projections that Gassko & Constancio came up with are not as rudimentary as the Marcels, which rely heavily on Regression to the Mean, it’s probably safe to say that they’re not as generous with Hart as PECOTA is. Even Nate Silver himself said, “I see plenty of scenarios where he matches PECOTA’s .288/.353/.517 projection, but that looks closer to a 75th percentile outcome than a weighted mean.” Anyone who’s seen Corey’s sweet stroke knows that 50+ XBH aren’t out of the question at all, but of course it needs to actually happen first.
Another point of note might be the Ben Sheets projection. PECOTA shows a weighted mean 3.29 ERA and 1.06 WHIP in 180 IP, while the Marcels show a 3.63 ERA in 129 IP. Nate Silver says: “Ben Sheets’ 3.29 ERA projection is higher than where most other forecasting systems have him, but lower than where most laymen have him. Sounds just about right to me.” I think he meant it the other way around, but either way, it bodes well for the Crew. The main issue here is the playing time, which looks good at this point: Benny has made all of his spring training starts without complication, and is set to pitch at Miller Park on in just a few days (I will eat nachos in honor of his voracious appetite while watching him K 20 Dodgers). The difference in these two projections amounts to about 8 starts, which means that the Brewers will have a better chance to win 8 more games under BP’s system.
It’s a pretty similar situation for Rickie “Skittles” Weeks. BP anticipates 610 PAs of .815 OPS while THT expects only 448 PAs with a .784 OPS. The difference of 160 PAs could make a huge difference, especially if Skittles is anywhere near .800 in OPS, which he should have no problem doing. We need a OBP>.360 at the lead-off spot, and Rickie can provide.
While those are the three biggest differences between PECOTA and the Marcels, PECOTA also gives higher ratings to Dave Bush (4.06 ERA in 185 IP vs. 4.22 in 177), Prince Fielder (.904 OPS in 638 PAs vs. .850 in 530), and Bill Hall (.868/633 vs. .836/559). The trend here seems to be that while the Marcels knock everyone back a bit through regression, PECOTA anticipates growth from a young team with a year of experience under its belt and its injuries behind them. Just as I’m more likely to take Howie Kendrick than Jeff Kent in my fantasy leagues, I’m also on PECOTA’s side with young players, and (albeit as a huge fan), think that the Crew’s record will be a lot more like what BP predicts than what THT does. If we look at the predictions for the other teams in the division, this ideas seems to hold up. THT and BP agree on the Cubs record, but they’re way off on the Cardinals: the Prospectus says 81-81, while the Times says 85-77. Again, I think THT might regress players like Edmonds to the mean, which means he’d actually improve from last year, and quite a bit at that. Plus, while Edmonds was plagued by injuries last year, unlike the Brewers with injury questions, Edmonds hasn’t played this spring and is set to start the regular season on the DL. THT also overvalues the veteran scrapitude of David Eckstein.
So what have we learned from this? Really, what we already know: that the Brewers need healthy, full seasons from Ben Sheets and Rickie Weeks, and they need to see a very reasonable amount of growth from their young players. Geez, sorry for wasting your time.

Quantcast