I'm not going to make the argument that 7 AL teams should have higher median outcomes than the White Sox, but I don't think it's an implausible argument to make, and I certainly don't think it's so "off" that it's proof that a system is garbage like some are arguing.
Hendriks is absolutely better than Colome, and fWAR supports this. However, fWAR is a DIPS-based statistic, and what that is measuring is how much of the "work" of the outs that are occurring can be directly attributed to the pitcher versus the defense/circumstance. Hendriks is better, essentially, because he allows much less contact and thus eliminates the factor of defense, resulting in fewer baserunners. However, that does not mean that his performance is guaranteed to produce better results, simply that it is more likely to produce better results. And what we saw in 2020 with Colome was nearly perfect results, even if he didn't actually pitch as well as Hendriks is likely to, and even if he relied on good defense and batted-ball luck to do so. So, if the question is "Is the 2021 White Sox closer role likely to allow fewer runs than it did in 2020?", the answer is almost certainly no. I agree that Hendriks could be deployed in a multi-inning role with some regularity, which would improve his gross production, but I'm very skeptical that the White Sox will do that. And even if they did, it would have to be a LOT more in order to make up ground toward a season of 0.81 ERA.
In this instance, we are comparing a projection to what actually happened last year. Even if what actually happened last year was unlikely, it's still our baseline if we're trying to find out if the team got better.
Yes, Lynn is exactly what they needed.
But they GAVE UP something they also needed in order to get it. In the process of filling that glaring hole, they reopened the SAME hole. The White Sox problem is that they did not have ENOUGH pitching, so they needed to acquire pitching by using a resource OTHER than pitching. In fact, it's even worse than that, because since Lynn makes so much more money than Dunning they actually paid BOTH in current pitching talent and in money to make the upgrade. If you have some oranges but you need many more oranges, you should buy oranges with money or trade for oranges with something else. You should not trade oranges you already have, and you CERTAINLY should not trade both oranges and money. You may still end up with more oranges than you started with, but you have drastically reduced your total yield in oranges.
This behavior isn't something we normally have to unpack as fans because it makes so little sense that it hardly ever happens in real life.
"But Lance Lynn is better than Dane Dunning."
Yes, but now you only improved your rotation by the DIFFERENCE between them. FanGraphs seems to be down right now but I think Lynn projects as about a 3.5 fWAR pitcher and Dunning projects at about a 2 - 2.5 fWAR pitcher (though to arrive there with ZiPS you have to adjust his IP, which is projected as like 70 innings or something). Most Sox fans see adding Lynn as adding 3.5 wins, but really it's adding 1 - 1.5 because the WHite Sox decided to SUBTRACT Dunning's 2 - 2.5 directly in the process. They spent $8m to add a win, and they still need just as many pitchers as they did before the move.
"But maybe they could ONLY have gotten Lynn by including Dunning."
Then get a different pitcher. There will SO MANY other options. How many 2 win pitchers signed for $8-12m this offseason? How many other, better pitchers were traded for prospects instead of MLB talent? If the guy at the fruit stand will only accept oranges in the deal, he's an idiot and you should go to a different store.
I would amend this to say some areas of LIKELY regression. And it is much easier for a player who just played over his head to return to his norms than for a young player to learn new things to get closer to his promise. In truth, each instance of potential regression and upside is a unique event with its own probability. We, as humans, are particularly ill-suited to parse and sum them objectively to arrive at a net result. Which is why we use mathematical models, like... you know, PECOTA.
That's certainly a plausible claim, but I don't know how you could possibly have evidence for it at this point, so I don't think it qualifies as a very good conclusion. Certainly not one that could survive scrutiny based on the information we have today.
Yeah I mean, that's a fine opinion to hold, but we all need to understand that's as far as it goes. It's a hope we share. It's completely possible, but here's the kicker -- even if the every player on the team has a career year and the Sox win it all, it does NOT say anything about PECOTA. All that would mean is that a whole bunch of unlikely things occurred at once. PECOTA is not a genie claiming the Sox will end up the 7th best team in the AL. It's giving you its MEAN outcome out of tens of thousands of trials. Real life is ONE trial, and all kinds of things can happen.
What it illustrates for us is that for the Sox to win, they need to achieve higher percentile outcomes. In other words, they nearly all of that upside to manifest. And this is, honestly, exactly where they were last year -- and it's pretty much status quo for this front office.