To answer this question most directly, an argument is fallacious when its premises fail to offer sufficient logical support to its conclusion. In the case of what was actually a bad slippery slope argument which came true in spite of its being a bad argument, we might be surprised, but that only shows that improbable events sometimes do happen. The argument has not become a good argument by virtue of the fact that the improbable events it depended on actually happened. It is not a reason for us to change the standards by which we generally evaluate the goodness of arguments. That probably covers it, but if you want to read more about slippery slope arguments and fallacies, then please do:
The basic underlying feature of almost all fallacious arguments (I say almost, because in some cases the distinction between the logical and material tasks becomes blurred) is that their premises fail to offer sufficient logical support for their conclusions. One point worth noticing is that it is possible to have a deductive argument, an argument in which the truth of the premises guarantees the truth of the conclusion, which follows the form of a slippery slope argument. A deductive argument following this form, if it is valid, cannot be a fallacy.
Fallacious slippery slope arguments are inductive, rather than deductive. As with their deductive counterpart, each premise is logically dependent on the previous one, but the support is only probabilistic. Since each premise is only probabilistically implied by the previous premise, every premise that is added to an inductive slippery slope argument reduces its logical strength. Think of it as multiplying a series of probabilities. If your first premise offers 95% logical support to the second, and the second premise offers 95% support to the third, then the third is only supported 90.25% since .95 x .95 = .9025.
However, this does not show that all inductive slippery slope arguments are automatically fallacious. If there are a small number of premises, and each premise offers strong logical support to the next premise in the series, then we might still think that an argument which has that sort of logical structure can be good.