Breaking Down David Schoenfield’s Phillies Prediction

Note: Skip down to the bolded text if you’d like to avoid a rant. That’s where the real Phillies-related stuff begins.

ESPN’s David Schoenfield recently wrote an article about the Phillies that evoked some very negative reactions from Phillies fans. It’s titled, “Why the Phillies won’t make the playoffs“. Making that statement about the Phillies, who have reached the post-season in every season since 2007 and have assembled arguably the greatest starting rotation in baseball history, is going to generate some, uh, conversation. Accuscore gives the Phillies a 76 percent chance to make the playoffs, by far the largest in baseball, beating the Boston Red Sox at 67 percent.

Schoenfield definitely went out on a limb, but he did a good job of explaining his thought process that led him to such a conclusion. I haven’t seen his detractors provide much of a counter-argument. The majority of the commenters on his article left me feeling very disappointed, and I’ve seen many others dismiss him outright on Twitter, which is completely undeserved. Unfortunately, this is the downside of giving everybody a platform to make themselves heard. ESPN, of course, appeals to a much more mainstream demographic and Schoenfield’s heavy use of Sabermetrics and rational reasoning seems foreign.

I don’t agree with everything Schoenfield wrote. However, just because I disagree with him and he disparaged my favorite team doesn’t give me the right to question his integrity. A rational discussion between two people who disagree shouldn’t go like this:

  • Person 1: The Phillies will not make the playoffs. Here’s why.
  • Person 2: YOU HATE MY FAVORITE TEAM! YOU ARE BIASED AND COMPLETELY UNDESERVING OF A PAYCHECK!

This post from Tango (as do the rest of the posts on his blog) shows the way a rational discussion should take place.

  • Person 1: The Phillies will not make the playoffs. Here’s why.
  • Person 2: I disagree with your claim because your Reason 1 is based on a faulty premise; you are relying too heavily on one piece of information; your evidence is lacking.
  • Person 1: Here are more reasons that support my claim. (Or: You are right. I will amend my position.)
  • Person 2: Interesting. Those reasons support your claim and make a convincing argument. I will amend my position. (Or: Repeat skeptical argumentation.)

That type of formulaic conversation seems dull to many people, but that is the basic structure of conversations people have every day — it’s just not as obvious as above. I think that, in general, people need to do a better job of being open-minded and willing to have these rational discussions. People are far too quick to bury their heads in the sand or plug their fingers in their ears because they are unwilling or unable to participate in such discourse.

Now that I have that rant out of my system, let’s actually examine Schoenfield’s claims and see how they pan out. I suggest you read his article first. I am not going to do a lot of quoting; instead, I will just highlight his main bullet points and refer to his arguments, assuming you’ve already read the material.

1. Roy Oswalt is around for a full season.

Schoenfield starts out at 95 wins for the Phillies since that was their Pythagorean W-L expectation last year. I don’t agree with that method. Instead, I’d have preferred he base the starting figure on projections. For instance, PECOTA has the Phillies at 89 wins. Given that Schoenfield started at 95 and ended up at 92, that would actually bring the Phillies down to 86 wins given the rest of his logic and that would have enraged tens of thousands more. However, it’s better to base a 2011 win projection on 2011 projections rather than 2010 runs scored and runs allowed totals.

As for Oswalt, Schoenfield correctly notes that a full season — rather than a half-season — of Oswalt is a good thing. He uses pitcher WAR from FanGraphs to derive Oswalt’s value. FanGraphs bases pitcher WAR on Fielding Independent Pitching (FIP). In other words, their pitcher valuation comes from what should have happened rather than what actually happened. This is why I have rarely (never?) used fWAR for pitchers. Value should be based on what happened, not what should have happened. As a projection for 2011, however, using FIP (or xFIP or SIERA) is totally fine.

Baseball Prospectus has Oswalt at 4.6 WARP in 2010 (2.5 with the Phillies), and projects 3.6 WARP this season. Their win metric is based on results, not retrodiction. Schoenfield’s addition of two wins for the Phillies was actually a bit generous, since Prospectus has Oswalt adding only one more win with a full season compared to his half-season.

2. Cliff Lee in, Jayson Werth out.

Schoenfield doesn’t state which projection system he’s using. It’s an important distinction to make because not all projection systems are alike. Bill James’ projections (which you can find at FanGraphs), for example, tend to be overly optimistic about hitters. In fact, analysts say that James’ projections, taken as a whole, are untenable.

In this post, I examined PECOTA’s projections for the Phillies and made my own adjustments as I thought they were a bit too pessimistic with regard to Roy Halladay and Cole Hamels. Last year, Phillies starters combined for 165 VORP, which includes 50 starts from Kyle Kendrick and Jamie Moyer. With my adjustments, PECOTA projects the 2011 starters at 217 VORP, or an increase of about five wins better. Schoenfield again was generous, citing the addition of Lee as a six-win gain.

Werth has averaged about five wins over the last three years. Calling for another five wins in 2011 is not in any way unrealistic. PECOTA has him at 3.7 WARP, which I think is too pessimistic. Schoenfield cites Ben Francisco and Domonic Brown as combined 1-WAR players, which PECOTA agrees with. Generously, that’s a three-win drop. Realistically, it’s four wins.

3. Chase Utley out.

Given the uncertainty around Utley’s condition, I’m not comfortable with any playing time projections at the moment. Schoenfield takes two wins away from Utley, who averaged about 6 WARP from 2005-09. PECOTA has him at 3 WARP in just under 400 PA, which means Utley misses a little over one-third of the season (basically, he’s back in mid-June). If you buy it, that’s a loss of three wins, which is more severe than Schoenfield anticipates.

4. Decline from Carlos Ruiz and Raul Ibanez.

I’ve covered both Ruiz and Ibanez during the off-season. Here, I explained that a lot of Ruiz’s offensive success in 2010 was fluky, based on an unsustainable BABIP. PECOTA sees Ruiz dropping from a .400 OBP and .447 SLG last year to a .347 OBP and .388 SLG in 2011. The OBP drop makes sense because of the BABIP luck, and the SLG drop makes sense for the same reason, but also because his isolaTed Power actually dropped from .171 in ’09 to .146.

PECOTA has Ruiz dropping from 3.5 WARP to 2.3. Schoenfield was in that vicinity.

As for Ibanez, here I discussed why he is likely going to fall around the league average for National League left fielders (~.770 OPS). The arguments against Ibanez usually focus on age and proneness to injury, but no one who uses those arguments ever quantifies exactly how that will affect his production. In the last three years, Ibanez has missed only 28 days due to injury and has had only one stint on the 15-day disabled list.

PECOTA has Ibanez being slightly worse in 2011, going from 1 WARP to 0.7. Schoenfield took a way one whole win. Given the uncertainty with projections, that’s not a big deal.

5. Jimmy Rollins is healthy and will play better.

Schoenfield credits the Phillies with another win with a healthy and better-performing Rollins. PECOTA projects a gain of 0.7 WARP. Personally, I’m more optimistic. FanGraphs valued Rollins at about 5 WAR on average from 2004-08, before injuries and bad luck derailed his past two seasons. He was at 2.7 and 2.3 WAR in ’09 and ’10 respectively. The difference between FanGraphs and Prospectus is defense. Prospectus has him at -18 Fielding Runs Above Average (FRAA). FanGraphs, which uses UZR among other metrics, disagrees. His Aggregate Defensive Rating (ADR), which is a composite of all of the various metrics, had him at +8 last year and +3 the year before.

When the statistics disagree, that’s when it’s a good idea to bring your own eyes and ears into the picture (not before!). I’m sure 99 percent of Phillies fans who have religiously watched most or all of the Phillies games over the past two years will agree that there has been no noticeable decline in Rollins’ defense, injuries aside. When healthy, he is a well above-average defender. So I have Rollins being as many as 2.5 wins better in 2011. Honestly, none of us have a high level of certainty around our expectations of Rollins.

6. Bullpen Issues

Schoenfield says that Ryan Madson and Jose Contreras are unlikely to be as good in 2011 as they were the year before, but he doesn’t provide any reasoning here. SIERA, which does a good job of predicting next-year performance based on current-year information, put Madson at 2.49 (actually only shades lower than his 2.55 ERA) and Contreras at 3.19. Strikeout and walk rates are the best predictors of pitcher performance, and Madson hit a career high 10.9 K/9 and a career low 2.2 BB/9. Contreras posted a 9.1 K/9 and 2.5 BB/9.

PECOTA is particularly harsh to Madson, predicting a 3.64 ERA with an 8.0 K/9 and 2.8 BB/9. I completely disagree with that. Contreras is at a 4.36 ERA with a 6.0 K/9 and 3.1 BB/9. Needless to say, I’m not sold. If there are reasons to expect Madson and Contreras to be considerably worse in 2011 than they were last year, I haven’t seen them.

Additionally, I disagree that the rest of the bullpen will be an issue. The best arms — Madson and Contreras — will be pitching the highest-leverage innings. I’m not comfortable with Danys Baez, David Herndon, and Kyle Kendrick, but given the amount of innings the Phillies’ starters will pitch, they’re just about insignificant. J.C. Romero is very effective when he faces left-handed hitters, which should be a majority of the time.

Schoenfield subtracts two wins because of the supposedly bad bullpen and calls that “conservative”. Actually, Madson-Contreras in the eighth and ninth compared to Madson-Lidge can be considered an upgrade, accounting for leverage. But I’ll call it a push.

That ends the Phillies-specific line of Schoenfield’s reasoning. He winds up at 94 wins, then subtracts two more wins because of the perceived improvement of the rest of the National League East. The decision to take away two wins seems arbitrary, especially since he didn’t do the same line of reasoning for the rest of the teams (which, admittedly, is very time-consuming). Additionally, using a WAR-to-wins conversion isn’t a flawless method. At FanGraphs, Dave Cameron says that the standard deviation between WAR and actual record is about six wins. In other words, if you put the Phillies at 92 wins, they will fall somewhere between 86 and 98 wins about 68 percent of the time. 95 percent of the time, they’ll fall between 80 and 104 wins. In other words, there’s actually a lot of uncertainty, even with WAR.

While I disagree with Schoenfield on some points and with some of his methods, the real takeaway from all of this is that I am not any more sure than he is (and neither are the people who so menacingly disparaged Schoenfield’s writing). People always say that the beauty of baseball lies in its randomness, and nowhere is that more evident than in that standard deviation of six “wins”. Breakout and breakdown seasons, batted ball flukes, and many more factors bring a lot of uncertainty to the table. If you are going to discredit someone else’s projections, especially when they put a lot of time and effort into producing and explaining them, then you need to reciprocate — show your work, then we can move towards a middle ground.

Phillies Stage Comeback, Win on Opening Day

Opening Day appeared to be a cold, dreary, roughly three-hour loss for the Phillies. The Astros chased Roy Halladay after six innings and 101 pitches, then bolstered their lead to 4-0 when they tagged J.C. Romero and David Herndon for three runs. Through six innings, the Phillies could only muster three base runners and mounted no offensive threats against Brett Myers.

Then, in the seventh, they worked counts better, their batted balls found holes, and they made productive outs. A sacrifice fly by Ryan Howard and an RBI ground out from Raul Ibanez brought the Phillies back to 4-2. After an unproductive eighth inning, the Phillies needed to score at least two runs in the ninth against closer Brandon Lyon. After two singles and a stolen base, they had runners on first and third with one out. From there, it was the story of Phillies should-be reserves. Ben Francisco, Wilson Valdez, and John Mayberry each hit RBI singles and the Phillies had completed the highly unlikely: they had come back from a four-run deficit with nine outs to go.

As the win probability graph (via FanGraphs) illustrates, the Phillies had about a 10 percent chance to win at the start of the seventh inning. As the seventh inning came to a close, after the Phillies had halved their deficit, their win probability was still only 13 percent.

Despite the great ending, I was a bit disappointed by the Phillies’ lack of plate discipline. 36 Phillies came to the plate yesterday, 20 of them saw three pitches or fewer. Four swung at the first pitch. Through six innings, the Phillies had seen a grand total of 56 pitches from Myers, an average of less than ten pitches per inning.

With Chase Utley injured and Jayson Werth elsewhere in the NL East, the Phillies do not have many players adept at working counts. The chart below shows their average pitches seen per plate appearance from 2010.

The top-three in the Phillies’ batting order — Shane Victorino, Placido Polanco, and Jimmy Rollins — each saw 3.75 or fewer pitches per plate appearance last year, which is not great. Prorating their averages over 650 plate appearances, Werth would have seen 2,840 pitches while Rollins comes in at 2,420; Victorino 2,387; and Polanco 2,303. The difference of 400 pitches accounts for about one-sixth of their pitches seen.

Seeing pitches has more than the obvious benefit of wearing a pitcher out and getting into the opposing team’s bullpen: it allows hitters to more accurately predict which pitches are coming, which leads to better offensive success. It should come as no surprise that, last year, hitters posted an OPS 270 points higher in hitter-favored counts than in even counts.

Previously, I had remarked that the Phillies don’t have many hitters with good on-base skills. Their relative lack of plate discipline plays a large role in that. If the Phillies intend on remaining near the top of the National League in offense, they will need to have a more disciplined approach in the batter’s box.