I wish a very dedicated cosplayer would practice IRL hexes on Rowling.

This is off-topic to the substance of the argument. I was talking to a friend about personality disorders a while ago and realized that there's a common argument/rebuttal pattern applied to any group accused of being a danger: trans people, racial minorities, people with mental health problems, etc. The rebuttal sucks:

A: X group commits lots of [type] crimes
B: Even if that were true, X group is more likely to be a victim than a perpetrator of [type] crime
B (alt): Even if that were true, X group is actually more likely than general population to be a victim of [type] crime

Note that cis men are at once (a) more likely to commit violent crime than general population (b) more likely to be victims of violent crime than general population (b) more likely to be victims of violent crime than they are to be perpetrators of violent crime. That's a farcical MRA talking point. These stats aren't telling us anything meaningful.

Instead of letting the opponent's claim stand like Coleman is doing, I think it is more convincing to either dispute it or offer a good explanation that neutralizes the claim. For instance, assuming there even is some kind of stat that trans women are disproportionate perpetrators:

• Perhaps that stat includes laws that criminalize existing as a trans person, like bathroom bullshit or whatever Rowling is proposing.
• Perhaps there are only a tiny number of trans predators (multiplicative problem: few predators, few trans people) so the sample size is too small to extrapolate reliable rates. Other methodological problems. Sex Crimes Georg got into the dataset and he isn't trans. Etc etc
• Perhaps trans women are poorer than general population (because they're discriminated against, healthcare is expensive, etc). That would mean they're more likely to be involved in any criminal conduct on either side - to put it differently, more highly-policed and likely to be arrested or officially victimized for acts that rich people get a pass for. So stat A is meaningless unless wealth-adjusted.
• Perhaps trans women commit fewer sex crimes than men but genuinely more than cis women, because they were raised in rape culture and some internalized those lessons ("you can ignore consent" learned) independent of gender identity ("because you're a man" discarded). This is not a good argument for trans women, but it does at least disprove the ridiculous bathroom predator arguments, which predict trans women to commit sex crimes vastly more often than cis men even though we just established they don't. So it could be appropriate against that argument, though there are better counterarguments that cede less ground.

There's a lot of stories you could tell. Just saying a different statistic is talking past each other and not convincing to the (entirely theoretical tbh) sharp-eyed reader who is somehow undecided on whether trans people should be allowed to live. If you're gonna ignore such an argument to offer a more convincing argument of your own, I think moving out of the realm of stats altogether is better.