I don't think there's a "be-all, end-all" stat for anything, but OPS is a darn good one to use. OBP accounts for both hitting ability and patience; slugging accounts for ability to hit for power. Put them together and it gives you a decent sense of who the most dangerous or effective hitters are.
Adjusted OPS and RC are both excellent to use, although RC needs to be adjusted for park and league context. To do this, you need to multiply by the park factor, and then divide that number by the amount of average runs created per game in that league.
Example: In 2000, Mike Piazza created 117 runs, in a park that lowers run production by 7%. Multiplying 117 by 1.07 gives us ~125 RC. In the NL in 2000, the league average of runs scored per game was 5, so Piazza created as many runs that a typical team would score in 25 games.
TPR, Linear Weights, and VORP are the metrics I have the most faith in. Win Shares gets lost in its own formulas and doesn't account for Loss Shares. RC is not too bad if put into the RC/27 context.
OPS is a good napkin stat, pretty easy to figure and much more accurate than the grandfather stats. OBP is definitely understated in basic OPS but not as much as many will put forward something close to (1.5 * OBP) + SLG. It should really be (OBP * teamSLG) + (SLG * team OBP) for a more accurate measure but you really start to lose the ease of calculation with that.
As a numbers guy, my main beef with OPS from a mathematical perspective is that it benefits those with higher AVGs moreso than those without, and really doesn't compensate for it elsewhere in the equation. It takes batting average twice into the equation, with the extra from SLG and OBP once apiece. I think if we added (OBP -AVG) and (SLG - AVG) to OPS, we'd have a more accurate figure of value. But maybe that's just me.