You can't really do it w/o the help of a computer. If you look at a ton of skew-Ts, you can probably begin to eyeball/estimate CAPE pretty well. I suppose you could print out a skew-t and use tiny pieces of paper to "estimate CAPE" by doing some "calibration" beforehand. For example, all the paper shreds equal 5000 CAPE, so if you only use 1/2 of the scraps to fill in the area between the parcel trace and the environment, then you could estimate the sounding (for the parcel you choose) to have about 2500 j/kg CAPE. I suppose you could use your calculator to numerically solve for CAPE by going through LFC to EQ at, say 250m increments), but that's not really practical.