For addition and subtraction, it's mainly manually adding and subtracting ordinates (y-values). Once you can work out where the addition/subtraction of two ordinates lie, you should notice a pattern at certain regions of the curve and can join the dots to figure out the curve.
For example, y = x + sin x, you start with a typical y = sin x curve. For small positive x, you shift the y coordinates of sin x up by a small value of x and for large positive x, this shift upwards is greater (resulting in a 'slanted' shift upwards). This is similarly the case for small and large negative x, where the shift is downwards instead. After this, you end up with a "slanted" sine curve.
For multiplication y= f(x)g(x), the most useful technique is to note what the sign of the ordinate is after multiplication. If you have two curves that have the same sign in y-value for a given x-value, then the resultant ordinate is positive. (i.e. f(x) > 0 and g(x) > 0 OR f(x) < 0 and g(x) < 0, then f(x)g(x) > 0)
If you have two curves that have the opposite sign in y-value for a given x-value, then the resultant ordinate is negative. (i.e. f(x) > 0 and g(x) < 0 OR f(x) < 0 and g(x) > 0, then f(x)g(x) < 0)
If one of the curves has an x-intercept (y = 0), then when you mutliply it by any defined ordinate, the result is still y = 0, so the x-intercept is conserved.
The rest is by inspection.