The problem with this is that its a rolling average, it may look similar on a poll of polls like this but not if you isolate the last few days. We're 9 days before the election and the last 10 polls have a con lead of: 12 (newest), 7, 9, 13, 9, 15, 10, 8, 6, 7 - an average of 9.6 and a low of 6. Compare that to the same point in 2017: 6, 3, 6, 10, 12, 4, 6, 7, 14, 12 - an average of 8.0 with a low of 3. In the coming days in 2017 we saw a lot more single figure polls including one showing a 1pt CON lead and another in fact showing a 2pt LAB lead. The 2017 gap continued to close whereas right now it seems to be stalling slightly
The poll of polls really doesn't tell the whole story, last time there was a really large range in polls, with the ones showing a large CON lead bringing up the average. However it turned out the close ones were correct last time, and quite a few of them accurately predicted the 2-3pt lead. This time the close polls are a bit further away and there are fewer of them.
True but that doesn't change the fact that in the last few days they have 51% of young people telling them they won't vote, and that should account for all the people who registered up to a week ago. On that basis their methodology is accurate, they would have to predict a 49% turnout of 18-24 year olds. Of course it is a very small sample in just one poll, but it seems like the big problem is that young people are actually not wanting to turn out on the level of 2017, and these pollsters may be right.