The evolution of sex chromosomes and their differentiation from autosomes is a major event during genome evolution that happened many times in several lineages. The repeated evolution and lability of sex-determination mechanisms in fishes makes this a well-suited system to test for general and predictable patterns in evolution. According to current theory, differentiation is triggered by the suppression of recombination following the evolution of a new master-sex determining gene. However, the molecular mechanisms that establish recombination suppression are known from few examples, owing to the intrinsic difficulties of assembling sex determining regions (SDRs). Forward-genetics data and the development of long-read sequencing have generated a wealth of data questioning central aspects of the current theory. Here, we demonstrate that sex in Midas cichlids is determined by an XY system, identify and assemble the SDR by combining forward-genetics, long-read sequencing and optical mapping. We show how long-reads aid in the detection of artifacts in genotype-phenotype mapping that arise from incomplete genome assemblies. The male-specific region is restricted to a 100 kb segment on chromosome 4 that harbors transposable elements and a Y-specific duplicate of the anti-Mullerian receptor 2 locus, a known sex-determining gene. Our data suggests that amhr2Y originated by an interchromosomal translocation from chromosome 20 to 4 predating the split of Midas and Flier cichlids. In the later, it is pseudogenized and translocated to another chromosome. Duplication of anti-Mullerian genes is a common route to establishing new sex determiners, highlighting the role of molecular parallelism in the evolution of sex determination.