I was without internet for a week, but I am back now with an updated prediction map.
As you can see I still suspect inbuilt errors with polling. In particular, that group of voters that seems to have not gotten correctly polled last time is still not being correctly polled, or so it seems. The NDP vote, however, does appear to be a real thing, but due to this "missing" group, that pushes down the NDP.
I suspect the real numbers to be closer to a tie for the Tories and NDP with Wildrose close behind.
The Liberals are really suffering. Their leader recently had to fend off accusations he was killing the party, something there is no good way to spin. With the NDP trending very heavily in Edmonton, Swann may be the only Liberal to come out of this with a seat, and with the NDP growing in Calgary, even that is becoming less and less likely.
The reality is that even if the polls are wrong, they are right. The Liberals have managed to miss a large number of ridings in which to find candidates; this indicates serious trouble in the party machine; a lack of volunteers, a lack of enthusiasm by loyal backers, a lack of hope for the brass, or some terrible combination of all three.
With all the polls showing the NDP as the bastion of progressive support this election, I fully expect many Liberals will vote NDP on election day. Some may vote PC to stop Wildrose, and others may vote Wildrose just to get rid of the PC Government.
In the end we run into a problem; math. The NDP vote is inefficient to the extreme. As noted on the map, even with just a lack of 1% in popular vote, the NDP ends up a full 20 seats behind the PC Party.
The problem for the Tories is that this is the limit where any more Wildrose voters will result in dozens of ridings flipping. I've included the ratio grid on the map; you can go to the website yourself and punch in .23 as opposed to .22 for wildrose and see how many ridings actually change due to such a small adjustment.
In terms of polling, I will say this. In short; ignore 1AB polls. They are of use, but not of use by their raw numbers. It's simpler to say ignore them than to double the remainder of this post trying to make specific exemptions for them.
As for the next 2 weeks, I see a few possible paths.
1 - Rebound
It is quite possible that voters will be turned off by the idea of both a Wildrose government and an NDP government. In such a case, the PC Party could rebound. Even just a few extra points would be enough to gain them a dozen or so extra seats above and beyond this prediction. To know if this is happening, watch for the PC Party to pull closer to even in the polls with Wildrose and the NDP. Remembering that I suspect the PC Party is being underpolled; an even split in the polls means a PC Lead.
2 - Wildrose Breakout
A bit unlikely given the problems the party had last time, but it is quite possible that given the threat of an NDP win, right-wing voters will back Wildrose. The idea of an NDP government could be just the spoiler that Wildrose needs to convince PC voters to stay with the party on election day. To see this happening, watch the party clear 35% in the polls regularly.
3 - NDP boost
It is quite possible for the NDP to win the most seats, and most votes, in this election. Less likely is a majority. The NDP would need to regularly poll in first place for this to become a reality, but it is something that can happen, especially with a collapse of the Liberal vote.
4 - Stability
This would indicate that voters want a minority government. As noted, very slight number changes can result in many seats flipping and with so many inaccurate polls, and, the history of voters not usually being so great at forcing a minority, this could result in a few forms of chaos, including an accidental majority for somebody. Polls that come out before the 28th will determine if this is the course we are headed on or not.