(Warning: this is long-winded. It's a slow work day.)
Using ERA is also flawed. It may capture some of the ability to limit HRs that is not taken fully into account by things like xFIP and SIERA, but at the cost of muddling it up with luck and variance that are mitigated with advanced metrics (BABIP, team defense, order-of-events things like LOB%, etc). Some of that order-of-events is deserved though, so guys like Greinke and Scherzer who are prone to having meltdowns might have that removed as variance. It's the nature of the beast... nothing is perfect. SIERA is also not park-dependant. Attemps to adjust for park factors resulted in effectively no changes versus using plain old SIERA, indicating that it's inherently adjusted for pitching environment. And unlike FIP/xFIP it doesn't simply use K/BB/HR, but rather K/BB and batted ball data. Which explains why it is a better ERA estimator than xFIP.
Cueto is definitely an interesting case though. Over the past 2 years he's significantly outperformed his SIERA. He, Lohse, and Vogelsong are in a class of their own in that respect (well over a run difference). Maybe they're the outer edges of normal distribution due to variance, or maybe there's a serious skill there that's not being measured. Almost certainly it's a mix of the two, but we don't know how much of each. If we go back to the previous 2 year span, Jair Jurrjens leads the league in SIERA-FIP with a mark that's pretty close to what Cueto has done in 2011-2012. Jurrjens has also had 2 recent seasons with a worse ERA than his SIERA, so how much of it was actually skill, and how much was luck/variance? I guess next year we'll see if Cueto is more like Matt Cain, or Jair Jurrjens.
Cueto's ability to hold runners is also a significant factor. He's only allowed one stolen base all season (and only one last season) and has an excellent pickoff move. This article
is pretty interesting.
Overall I'd say it's still between Kershaw/Dickey/Cueto. I'd lean towards giving it to Dickey, but it'll come down to what happens in the last month.