You are probably familiar with all of this, but that's a tricky one since dispatchability puts those into different equivalence classes. California ISO has negative LMPs right now, which means building additional solar provides little marginal value to the utility. As a result, even if the "traditional" LCOE for solar (just using capital cost, O&M, and capacity factor) is lower than coal, coal would technically still have higher value to the utility.
You'd have to model real-time demand and fratricide to get a average marginal rate (almost impossible to predict for a 20-30 year horizon right now, given rate of innovation), and then compare that against the amortized capital cost and O&M to get an expected ROI for the plant. An LCOE comparison wouldn't really make sense from a decision making standpoint.
That wouldn't be true if they were both closely equivalent (say NGCC vs NGCT, or coal vs nuclear). Then a strict LCOE comparison would be useful.