The actual main finding of that study seems to be that we lack any kind of good data source especially given that trans questions were removed from the census:

Dinno suggests that the most surprising finding, that transgender people overall do not face greater homicide risk than do cisgender people, is because of transgender people becoming adept at strategies to reduce potential harms, such as avoiding certain neighborhoods or being cautious around new people. Although research has found that many transgender people do employ these strategies that limit their freedoms in favor of safety, a claim that transgender people are successfully avoiding their own murders remains dubious. Because of the uncertain nature of both the numerator and denominator in these relative risk calculations, a far more parsimonious explanation would be that these estimates have substantial limitations and that better data and further research are needed to understand the relative risk of homicide for transgender and cisgender people in the United States. In that regard, Dinno’s work cannot provide a definitive answer to the “transgender murder rate” but instead provides a compelling argument for the advancement of more comprehensive data collection strategies to improve the safety of transgender people.