Rating system (glicko, elo, etc): is 'predict the outcome of games' the same thing as 'reduce the number of upsets'?

Question

From Lichess page 'Chess rating systems':

Question:

Which rating system is best?

Answer:

The purpose of rating systems is to predict the outcome of games. Therefore, they can be objectively better or worse, according to their ability to make such predictions. Glicko 1 makes better predictions than Elo, and Glicko 2 makes better predictions than Glicko 1

My question: Is 'predict the outcome of games' the same thing as 'reduce the number of upsets' ?

Upset here is simply lower rating defeating (or drawing against?) higher rating. Thus, assuming all ratings are integer valued and no 2 players of equal rating play against each other, what I understand is that

'1800 is more likely (or at least as likely) to beat a 1700 if the ratings are glicko 2 as compared to glicko 1'

if and only if

'Glicko 2 makes better predictions than Glicko 1'

Is there anything I misunderstood?

Answer 1

Basically correct. (Note that these rating systems don't natively deal with draws, only with binary wins/losses. Draws are incorporated by summarizing results as point-score = wins + 0.5 * draws.)