This is misleading because you're not comparing like with like. You should be comparing chances of at least one win in a "cycle" in both cases. Ignoring the zero, the cycle for a single number is 36 spins, and the cycle for an even chance is 2 spins. But the chance of at least one hit for an EC in 2 spins is 75%, not 50%. So in a cycle, the chance of at least one hit is actually better for the EC's.
Regarding probability your statement is true but you forget something of paramount importance!
The payout for predicting 1 out of 2 EC bet is 0,you get nothing for predicting 1 out of 2.
On the other hand,by predicting 67 times out of 100 a single number with 35 to 1 payout,it would generate these kind of profits:
(1x35 + 1x34 + 2x33 + 2x32 + 2x31 + 2x30 + 2x29 + 2x28 + 2x27 + 2x26 + 2x25 + 2x24 + 2x23 + 2x22 + 2x21 + 2x20 + 2x19 + 2x18 + 2x17 + 2x16 + 2x15 + 2x14 + 2x13 + 2x12 + 2x11 + 2x10 + 2x9 + 2x8 + 2x7 + 2x6 + 2x5 + 2x4 + 2x3 + 1x2 + 1x1 + 1x0) = +1188
While losing 33x36 = -1188
Therefore 0 difference,not worst than your EC selection!
Regarding your system of betting the same as the last 36 numbers one at a time and in the same order, you make a bold claim! how much have you tested it? I can't see any reason why it should give an advantage, except perhaps that you are betting numbers which have recently hit, which is probably better than choosing "colder" numbers.
It's much better to bet this way because you'll never a find a board showing last 37 results like this:
Thus if someone bets the same number all the time,sooner or later he is going to find that this number might be sleeper,by that time it would be late...
The betting selection I'm suggesting is closer to what really happens on each cycle.
There is another more accurate but more complicated way of selection method,but I'm not going to analyze it because I've received only negative comments so far,therefore it doesn't worth to explain you further my strategy.