That messes with the simplicity of the system we use here: 1 match, 1 prediction with every match (in a given round) being worth the same amount of points.
If you want to call a winner for a knock-out match you have to make 3 predictions per match: score after 90 minutes, score after 120 minutes, result of a penalty shoot-out. Either you have to roll those into one somehow which really convolutes things and takes draws out of the equation altogether (reducing the outcome to a simple binary decision), or some matches end up being worth more points than others depending on how long they go and you have to deal with questions like if you predict a win for one side after 90 minutes, do you even get to predict the other two? If not, that imbalances the scoring system in favour of draws in the knock-out stage.
And that's certainly a viable way to run a prediction competition, there are plenty of them that do it that way (like Marla's) and there are 1 or 2 relatively elegant ways to handle it (though they make more sense for two-leg knock-out rounds a la the CL), but that's just not how this contest happens to work. Both approaches have their proponents, it's a matter of preference. For possible future editions it could be worth a discussion but unless there is a massive outcry of protest, or an obvious flaw is exposed or unanticipated situation encountered, the rules were set at the start and I don't want to change them mid-way through (if for no other reason than that it would be a nightmare to update the formulas across 40 tabs on a Google Docs spreadsheet).