Yes, you got lucky.
Basically, what you're saying is this:
Flip a coin. Observe what it comes up... say, for the sake of argument, it's heads.
By your method, you would bet that a tails would come up next. That is,
P(heads on second) = 1/4 and P(tails on second) = 3/4.
The problem with that is this: imagine somebody else shows up after you flip the coin the first time, and has no idea what you're doing. If you ask him what the probabilities of heads vs tails is, he would likely respond:
P(heads) = 1/2 and P(tails) = 1/2.
Of course, he's right, because it doesn't matter what happened in the past... only what is still going to happen.
Does that clarify it any?