The Fans and Marcel in 2011: Pitchers

Monday, I looked at how the 2011 Fans and Marcel projections compared for position players, and today I will do the same for pitchers. For position players the fans project more playing time than Marcel. The fans also project higher rate stats (wOBA) for most players , but for below-average players actually project a lower wOBA than Marcel. Most of this can be explained by a the fans not regressing towards the mean as much as Marcel does. Let’s see whether the same holds for pitchers.

Again, I am looking just at players with more than 15 fan ballots, and grouping all fan ballots regardless of the fans’ favorite team. Again, Marcel is on the x-axis and the fans on the y-axis. The red line is when the two systems project the same, so the closer to the line the more similar the projections. I separate out relievers from starters by considering a pitcher a starter if the fans project more than half of his games as games started. Starters are solid points while relievers are empty circles.

Here are the innings pitched projections:

Almost all pitchers have more projected IPs by fans than by Marcel, just like position players with PAs. The difference is most noticeable for pitchers who are expected to spend their first full year in the rotation (Jeremy Hellickson, James McDonald, Daniel Hudson), so again the fans bring more depth-chart knowledge than Marcel. But the difference is seen throughout — again, I assume because the fan’s don’t regress as much as Marcel.

Marcel also doesn’t know about Johan Santana’s September 14th surgery, or the fact that he hasn’t started to throwing yet. But the fans do and it looks like the Mets fans know it even better than everyone else. They see Santana throwing just 106 innings versus 161 by other fans (Marcel thinks 176). Here, unfortunately, it looks like the Mets fans have a better handle on Santana’s time table.

Turing to ERA:

With just one exception (Joel Peralta) the fans project lower ERAs for relievers than Marcel does, and by 0.4 runs on average. A big part of this is that the fans are much more likely to project good relievers: projecting closers and setup men, not middle relievers. If they did project some middle relievers maybe some would have higher fan-projected ERA than Marcel-projected ERA. The amount Marcel regresses is inversely proportional to the number of innings pitched, so all relievers are going to get lots of regression. Since the fans don’t seem to regress towards the mean as much they will give closers and setup men (who had low ERAs in the past) very low ERAs compared to Marcel.

Looking just at starters:

Here the fans are actually in broad agreement with Marcel. There are nearly as many pitchers with higher fan-projected ERA, 56, as with higher Marcel-projected ERA, 60. And on average Marcel ERA is just 0.0007 runs higher. So this is the case were the fans have the smallest overprojection compared Marcel, effectively none. Though there is still a trend for the fans to rate above-average pitchers better than Marcel does.

So, overall, we see that the fans are much less likely to regress than Marcel, which is a natural human tendency — i.e. to take what we observe as true talent. But I do think that Fans bring an important piece of information with their knowledge of depth charts and injuries. They are probably better trusted for playing time, once you discount their projections somewhat for regression. In terms of rate stats, fans project much better performance from position players and relievers, but, strangely, not starters. Across all groups of players the fans project above-average players to do much better than Marcel projects them, and below-average players worse than Marcel projectes.