It doesn't matter if he is President or not when he says it
Any qualifying remarks must be congruent with the title. Remarks that go further, ie 'We should not help Taiwan' would qualify. Remarks that seem to go less far but imply the title, ie 'We should send military aid' would not qualify.
In the extremely unlikely event that Trump specifies the US should only not intervene if China invades outlying islands this will not resolve YES.
I will not bet and will judge whether the market has been met.
Related questions
Odds seem good that ratuzicat and sanargama, the only people betting "yes" are doing reference class bets https://manifold.markets/chrisjbillington/will-bsp9000-win-the-rootclaim-chal#u4ia0ZpjH6V01UBsSVwg
In reality, foreign policy generally isn't impacted much depending on who is president. People who know foreign policy know this. Trump's "america first" platform will likely not apply here, which is far more militarily serious matter than his statements about not protecting Japan (which nobody wants to invade, at least openly) and reducing US contributions to NATO (which was just a bargaining game and NATO would have been fine either way unlike Taiwan, whose defense revolves around strategic ambiguity).
It doesn't matter whether the military would retaliate against Trump during his campaign, or if he'd be blamed for it if China took the opportunity to invade Taiwan, or even whether the last 3 presidents have been paid actors/spokespeople/figureheads. Trump does not have the incentive to erode strategic ambiguity this far; Biden only eroded it because he had an excuse (slips of the tongue) and even that was in the opposite direction (and in a time of severe escalation between the US and China).
@ooe133 I'm not going to get into the weeds of this and actually argue the point, but I will just note that everyone I know in foreign policy circles in Taiwan views whether or not Trump is elected as extremely impactful (they don't all think it's bad)
@JoshuaWilkes Do you know people in foreign policy circles in the US or China, or Taiwanese foreign policy people who credibly claim to be near TSMC analysts?
@ooe133 much less so for the former two. For TSMC yes in theory, but people I've talked to have given me quite different messages about that relationship so I'm generally not very confident with that side of things.
@JoshuaWilkes hm, that's interesting. I consider that an update, I would have predicted that TSMC-adjacent people think that elections aren't so important. I sold some of my stake and I'm going to do some more research on this area when I have the time.
@Sanargama coined by yudkowsky in the link I gave above, a Reference Class Bet is when someone makes a bayesian calculation based on a reference class, but often just one bayesian update, without going deeper. For example, the stealing brain signals market: https://manifold.markets/IsaacKing/by-the-end-of-2028-will-there-be-a-9a7a1b6d31a7
a reference class bet is when someone decides they only want to spend 20 seconds on that market, and they bet based on the reference class of "things that sound like something a conspiracy theorist would say". This means they aren't likely to outperform others in the market who go deeper, but they save time.
In the brain signals market, Zvi also made a reference class bet, albiet a better one than the example I gave in this comment; some reference class bets are better than others, but specialization is still important.