Something that I haven't brought up until this point, but which I think is worth considering, is the matter of comparing commanders by CEV.
This is not the only way of comparing commanders, but it's a good one IMO because what it does is that it specifically factors out the issue of numbers. Everything else is fair game, and a commander on the defensive will naturally have a higher CEV on average (for example); in another war it would be useful to try and work out if one side had better troops, but in this case since most large ACW armies were broadly similar it's worth looking at.
The way that the CEV works is that you take the force ratio of the two armies, and square it; this is the expected casualty ratio going the other way. So if army A is twice the size of army B, you square the ratio A:B (2:1) and it becomes 4:1.
This is the expected casualty ratio in favour of the larger army.
Then you look at the actual casualty ratio (let's say that A inflicts twice as many casualties on B as B inflicts on A) and compare them. The amount that an army does better or worse than the expected value is the CEV of that army.
So a 2:1 advantage in favour of A looks good, but it's actually indicative of a CEV of 0.5; B has done well to hold itself to that few casualties.
Achieving a high CEV is not by itself a way to win a war. If Army A and Army B in the above example kept clashing then Army B would be worn down to a stub quite quickly. But it is a useful way of comparing commanders by looking at how well they did relative to the forces they had available.
First, as a quick benchmark, the Union CEVs in most of the big battles in the East are listed below.
1st Bull Run 0.74
2nd Bull Run and Chantilly 0.24
Fredericksburg 0.2
Chancellorsville 0.26
Gettysburg 0.88
Mine Run 0.22
Wilderness 0.22
10th May at Spotsylvania 0.39
12th May ditto 0.35
Cold Harbor 0.09
Petersburg 0.15
Crater 0.12
In some cases not a lot of fighting took place but that which did can still be analyzed.
Note that these don't actually consider whether or not the battle was a victory, though most of these are actually (at least tactical) defeats - this is just about the fighting itself. The manoeuvering is separate, though good manoeuvres can lead to a good CEV or can win a battle in spite of a bad CEV.
Now, here's how George Thomas did as an independent commander.
1) Mill Springs.
CEV here is 4.06, which is really good. It's on the defensive and a small unit action, which can be quite "swingy", but it's definitely a good sign.
2) Peachtree Creek.
CEV here is 1.2. This is actually better than Gettysburg, and it's a much larger battle than Mill Springs - Thomas was defending against a slightly smaller army here, but he acquitted himself well.
3) Nashville.
Sadly this isn't so good. Thomas had slightly more than double Hood's force size, which means that in spite of doing a lot of damage he comes out at about 0.5 for this one. It's a pretty good offensive battle, though, and it's actually better than any offensive battle in the above sample.
Here, however, are McClellan's battles.
Seven Pines 1.6
Mechanicsville 4.5
Gaines Mill 3.55
Peach Orchard, Glendale and Malvern Hill 1.28
South Mountain 0.64
Antietam 0.7
The interesting thing about these is that we have five McClellan-Lee engagements. Lee's CEV for the latter five is just the inverse of McClellan (which is how the CEV works) and we can thus determine the characteristic CEVs for these commanders attacking and defending against one another, and the way it comes out is that - purely in terms of the ability to use the "whole force" available - McClellan slightly outfights Lee.
This is the case for no other Union commander, indeed nobody comes close to McClellan's offensive CEV against Lee and whenever someone (mostly Meade) got a chance to try it on for a defensive CEV against Lee they came out with a CEV of less than one.