Clicking on this match and then on the advanced analysis link, shows that their match forecast is:
41.6% Korea
29.0% Draw
29.4% Algeria
So you would think that they would assume a win for Korea. But what seems to be happening is that instead, they are using their most likely correct score to determine the match winner. For this match, their most likely score is 1-1 (13.0% chance)
That seems a pretty flawed way to pick the most likely winner...
I think you might be misinterpreting the graphic. What they're doing makes sense to me. Could you explain why you're saying "they are using their most likely correct score to determine the match winner"?
For example, take a look at group C. Japan has three most likely scores of 1-1, whereas Greece has two 1-1s and a 0-1, yet Greece is expected to advance. It seems clear they're using the win probabilities, not the most likely scores, to determine who advances.
I may be getting confused here, but the visualisation doesn't help things. (Plus the original message I'm replying to seems to have been deleted)
The top-level 'front page' of the table seems to have a colour code of blue for match winners. But in the group H KOR:ALG game, both teams are coloured white. Yet, their percentage chances indicate that Korea is the most likely winner of this match.
Though the 'who advances' bit does seem to be properly calculated from their win chances, as you say.
Clicking on this match and then on the advanced analysis link, shows that their match forecast is:
41.6% Korea 29.0% Draw 29.4% Algeria
So you would think that they would assume a win for Korea. But what seems to be happening is that instead, they are using their most likely correct score to determine the match winner. For this match, their most likely score is 1-1 (13.0% chance)
That seems a pretty flawed way to pick the most likely winner...