RN
Three different things are being stacked on top of each other in that question, and they come apart cleanly once you name them.
There is what your body does in response to an image. There is who you are drawn to over time, in and out of a browser. And there is the word you use for yourself in front of other people. Researchers keep these apart because they behave differently and they do not reliably predict each other. The question in the title assumes all three are one measurement, and the reason it feels unanswerable is that it is three questions wearing one coat.
First, the word in the title
The genre tag is an industry label, not a description of anyone. GLAAD's media reference guide lists it among defamatory terms and describes it as language used to denigrate and hypersexualise trans women.
That matters here for a practical reason rather than a scolding one. A category name that was written to sell a fetish is doing a lot of quiet work on how the material is framed, how the performers are presented, and how a viewer interprets their own reaction to it. If your read on your own response is partly a read on a marketing taxonomy, that is worth knowing before you draw conclusions from it. The performers are trans women; the tag is a shelf label a distributor chose.
Arousal is measured, and it is not the same instrument as orientation
The laboratory version of this question has been run. In a 2004 study in Psychological Science, Chivers, Rieger, Latty and Bailey compared genital and self-reported arousal across groups and found a sex difference in how specific the response is: men's genital arousal tracked the category of stimulus fairly closely, while women's did not, with women showing substantial genital response to material that did not match their stated preference.
Two things follow, and the second is the one usually left out.
Genital response can occur without the corresponding wanting — later reviews in this literature describe response to material that participants found unappealing or aversive. So a physical reaction is not a verdict. But the same literature says men's arousal is more category-specific than women's, which means for male readers a persistent pattern of response is not pure noise either. Anyone telling you arousal means everything is overselling it. Anyone telling you it means nothing is skipping half the finding.
The honest position is narrower than either: a response tells you something about what you respond to, and nothing at all about what you should call yourself.
What the study on this specific question actually found
There is research aimed directly at men with a sexual interest in trans women. Rosenthal, Hsu and Bailey published an internet survey in Archives of Sexual Behavior in 2017 — 314 men reporting the interest, 211 controls — and their headline conclusion is the one that gets quoted: the pattern is best understood as an unusual form of heterosexuality rather than a separate sexual orientation. The men reported far higher attraction to women than to men, and on average were equally attracted to both groups of women.
The paper reports two further results that rarely travel with that quote. The same men reported slightly higher bisexual feelings than the control group. And they scored markedly higher on a measure of autogynephilia — arousal at the idea of oneself as a woman — which the authors treat as a further variant of heterosexual attraction.
Whether that last construct is valid is genuinely contested; it has been the subject of years of sustained criticism, and this article is not going to settle it. Note also what kind of evidence this is. A self-selected online survey of a few hundred people recruited by interest is a description of who answered, not a population measurement, and the authors' interpretive frame is doing real work in the conclusion.
So: the most-cited study on the exact question does not return "no." It returns "mostly heterosexual, with some complications, according to one contested framework, in one self-selected sample." Which is a real answer, just not a tidy one.
Why the question feels urgent at all
Gayle Rubin's 1984 essay "Thinking Sex" describes a charmed circle: the sexual practices a society treats as normal sit inside it — heterosexual, monogamous, marital, non-commercial, reproductive — and everything else sits outside, where it attracts suspicion in proportion to its distance from the centre.
That framework explains the anxiety better than any finding about arousal does. The distress in this question is almost never about the content. It is about the suspicion that a preference has moved you across a line that carries a social cost. The stakes are in the hierarchy, not in the video.
Which is also why no research result will fully answer it. A study can tell you what a sample of men reported. It cannot tell you whether you will still be welcome at a particular dinner table.
The part you actually have to do yourself
Orientation, as researchers use the term, is a durable pattern of attraction — who you want, over time, with the browser closed. It is not established by a viewing history, and it is not revoked by one either. Categories are shelf labels; a shelf label is a claim about where a video is filed, not about who you are.
If you want a word for yourself, the word is yours to choose, and it is allowed to be provisional. If you would rather not have one, that is also available. Neither decision is made by a tag.
And if the question is not really curiosity but persistent distress — the kind that stays after the tab is closed, and starts affecting sleep, mood or how you treat people — that is a different problem with a different address, and a therapist is a better one than an article. There is no evidence base behind anyone who claims a genre preference can be corrected, and being told what you are by a stranger is not a substitute for working out what you want.
The material itself is catalogued at trans porn sites and its premium listings, and the separate gay category is separate because it is a different thing — which is, in the end, most of the answer.