I've gone over my general starting pitcher considerations here.
Selection Walter Johnson
As I've stated, I deduct value from both dead ball era players and pre integration players. Walter Johnson played more than half his career in the deadball era and had nearly all of his brilliant years during that time. In fact there's a distinct drop in his performance (compared to league average) between 1919 and 1920. 1919 was maybe his second best year ever, with a 1.49/2.07/47/66 (ERA/FIP/ERA-/FIP-). 1920, the first year of the live ball era, he managed only a 3.13/2.78/84/78 line. Still good, but the dominance he showed before then never really returned, though his 1924 2.72/3.31/67/84 year is nothing to sneeze at.
Even so, deadball baseball was still modern style baseball. And Johnson may have suffered an injury at the same time the balls became livelier, as he only pitched 143 innings in 1920 after pitching at least 250 and generally far over 300 for the previous 11 years. Or maybe it was just wear and tear, as Johnson clocked a 90+ mph fastball in years before the world's first kinesiology department. One shudders to consider what he may have done after a year of Mike Maddux's or Dave Duncan's attentions.
Either way, Johnson was the dominant pitcher of his time. His career was 21 years long and he still managed a lifetime 2.17/2.36/67/75 line. I'll take it, particularly with a modern defense. The dominant strike out pticher of this day, he had a 12.6% K rate going back to 1916 (before then, batters faced was not recorded) and 6.4% walk rate. Excellent control for any time period. If he's a little too hittable in the starting line up, I'll switch him into the bullpen. With his side arm motion, velocity and strong control, he'd be murder on right handers, for sure.
Look, would you want to be called Grover Cleveland? Him neither. Like Johnson, Cleveland was a dominant deadball pitcher who became merely great after 1919. His career line is a little off Johnson's, at 2.56/2.85/73/81, but he was just as dominant during the deadball era. His best year, in 1916, he managed 1.55/2.12/43/62. Generally, the same caveats apply as Johnson, though the bullpen fallback might not work as well, as I don't think Alexander had such a side arm motion.
One of two pitchers playing today with over 2000 innings pitched with both an ERA- and a FIP- lower than 80* (the other being named Roy as well. If you want your kid to pitch, name him "Roy Johnson"). Oswalt (Weir is he from, anyway?) is a statistical chimera, swinging from flyball to groundball back to fly ball pitcher in his career. And his strikeout rate has fluctuated between over 9 K/9 and just below 6 K/9, his control has remained consistently good at about 2 BB/9. Oswalt has also managed to be hard to square up on, with a lifetime batting average against of .249. It will be interesting to see if he can deal with his back issues enough to make a difference for the Rangers this year. A lifetime 3.21/3.35/76/78 pitcher, Oswalt's best year has probably been 2002: 3.01/2.99/71/68
*Johan Santana will make it in another 14 innings.