Most reviews or books will tell you that directly-labeled FISH-probes should cover tens of thousands of bases of target DNA in order to be visible in the microscope. But I haven't found any details for indirectly labeled probes other than that their target DNA can obviously be much shorter. The only commercial gene-specific CISH-probe for which such information was available, was itself as few as about 160 bp long, meaning that its target DNA was just as short and a single 100+ long oligo can work with enough signal amplification