pulsetrain = repmat( [1 * ones(1,125), -1 * ones(1,125)], 1, 4);
(125 + 125) * 4 = 1000 samples.
1000 cycles / 500 cycles/s = 2 seconds.
2 s * 2 cycles/s = 4 cycles. Each cycle has 2 phases, so 2*4 = 8 half-cycles. 1000 / 8 = 125 samples per half-cycle.
If you want to add high frequency components, you can make the same sort of calculations to find what to add. Do remember, though, that the addition of 1's and -1's can add up to greater than 1 or less than -1; you will need to decide what you want to do about that.
