CON 126
LIB 92
BQ 52
NDP 36
OTH 2
GRN 0

See Commentary About Our Final Projections

My wife sees a silver lining in all this, the cons will accelerate civilization’s decent to rock bottom, were people finally figure out that there is something wrong with this.

Ridings they may all have wrong:

West Nova
Papineau
Ahuntsic
Edmonton Strathcona
Oakville
Vancouver Quadra

(I am sure there might be others)

Election prediction project:

125 CON
94 LIB
36 NDP
51 BLOC
2 IND

Compared to what I have said above, they seem to have gone largely with incumbents in the ridings that were still too close on Friday.

Democratic Space, as of last night, had

128 CONS
92 LIB
34 NDP
52 BLOC
2 IND

niXtuff

124 CONS
88 LIBS
42 NDP
1 Green
2 IND
51 BLOC

- a bit more radical. It might be fun to go through the three and look where they have made different calls, and at cases where all three might yet prove to be wrong (Edmonton Strathcona ?)

If any of the above scenarios is true, it is hard to see how Harper can convince the GG that parliament, which “wasn’t working” enough to justify an election call in violation of his own fixed election date law, is now going to work any better. No significant increase in seats, and possibly a drop in popular vote. No seats in PEI, NFLD, or (at least according to niXtuff) Nova Scotia. Perhaps fewer than 10 seats east of Ontario, and losses in Quebec.
Not much of a mandate, though still better than the other parties.

I am stilling waiting for 1-2 % of the green vote to break for whichever party is most likely to beat the tories in several key ridings. My guess is

CONS 123
LIBS 93
NDP 40
BLOC 50
IND 2

He has 54 ridings still in too close territory.
20 in Ontario, 6 in Quebec, 10 in BC, 4 each in NB and NS, 3 in Sask,1 in AB, and 4 in MB, and 1 in PEI and 1 in NL.

His current call is for
128 CONS
91 LIBS
34 NDP
53 BLOC
2 IND

Of the 54 that are too close, he has called

CONS 18
LIBS 25
NDP 6
BLOC 4
IND 1

In addition to those he has predicted for each party, the realistic chances of winning “too close” seats currently given to another party are:

Green 1
CONS 24 for a maximum of 152
LIBS 16 for a max of 107 (same as my previous post !)
ND 18 for a maximum of 52
BLOC 2 for a Max of 55
IND 0 for a max of 2.

Worst case scenario (lose all close seats) for each party:

CONS 110
LIBS 66
NDP 28
BLOC 49

MEDIAN PROJECTION (AVERAGE OF MAX AND MIN)

CONS 131
LIBS 92
NDP 40
BLOC 52
IND 2

(which doesn’t add up to 308, but 317)

They are now down to 35 seats still labelled as too close to call (so about the number they usually get wrong; it was 30 last time and I imagine the majority were in that TCTC group). Some they have already decided may still be in play though, like Edmonton Strathcona and Palliser.

Their current predictions

Cons 118
Libs 77
NDP 29
BLoc 47
Independent 2
TCTC 35

Of the 35 that are still too close, they are currently held by

LIBS 20
CONS 7
NDP 5
BLOC 3

So if all parties hold their close seats, we would have:

CONS 125
LIBS 97
NDP 34
BLOC 50
IND 2

Who stands to gain those seats still called too close if they are not holds ?
Some are 3 way races, so these numbers won’t add up.

CONS 19
LIBS 10
NDP 9
Bloc 2
Green 1

This would put an upper limit on each part of

CONS 144 (hold all too close plus win 19)
LIBS 107
NDP 43
BLOC 53
IND 2
Green 1

I might try this again with Democratic space’s predictions, using all of his
ridings still labelled too close.

By definition, in any meta-analysis, there MUST be one measure that will have a higher projected seat total for each party than all other measures; and there MUST also be another measure that will have a lower projected seat total for each party than all other measures.

Seat predictions are toward the bottom of the page. Because of their methods, their prediction changes daily, occasionally drastically. The CPC, for instance, have lost 20+ seats in the prediction this morning compared to a prediction made just Monday.

I know that the UBC election forecaster is not included because they use the exact same methodology as the Paulitics geometric analysis.

As for your question about the Geometric analysis, you’re absolutely right to point to potentially bizarre results of a Geeometric projection in ridings where any one given party is abnormally strong.

Because of this, it is sometimes possible for a Gemoetric riding-level projection to add up to more or less than 100%. In these instances, however, determining which party is projected to win is, by definition, never a problem since one is always way above the others. Thus I still go by whichever party has the highest vote projection.