The good, the bad, the nitty-gritty, the legislative council and the Dark Knight.

We actually have four elections to recap--one predicted well, one poorly and two I didn't touch. First up:

TASMANIA

Not a huge amount to say here. These were my predictions:

These were the actual results:

The areas I got right really don't leave me much to talk about. They fell the way I said they would, presumably for the reasons I said they would. The only incorrect call is the final seat in Bass. The prediction for this was that the Greens would lead the 4-way race for this last seat. In quote fractions, this was incorrect as follows:

ALP: 0.58
GRN: 0.56

LIB: 0.53
JLN: 0.28

As such, my flow-on predictions did not work out. I did provide a short summary of what might happen if the ALP was ahead at this point, but it was basically a mess depending on who dropped out next and how JLN preferences distributed. At a basic level, though, my prediction failed because I assumed ALP would get 1Q+N votes and the greens M where Q is a full quota, and N < M < Q. in actual fact N > M by a very thin margin--an error of 297 first-preference votes in a seat with almost 65,000 formal ballots. That's pretty darn close to getting 100% on my predictions.

SOUTH AUSTRALIAN LEGISLATIVE ASSEMBLY

A somewhat less pleasing prediction result here, but one that I will break down into specific areas I did well in to soothe my wounded ego. While there may be a little counting to go in some seats, the results look reasonably clear:

That is 36 correct out of 47 or less than 77%. Not great. But there is some silver (or, at least, grey) lining. Normally the independents are hard to predict: they can come and go without warning, with no regular polling available and campaigns limited to individual seats making them very hard to research. Yet of the 15 lower house independent candidates, and with five seats previously held by independents, I managed to predict every single one. That's something of an accomplishment.

Additionally, and now famously, Xenophon's SA-Best party dramatically underperformed compared to expectations. On the numbers, I had 7 seats going to SA-Best but I did also say

My gut says SA-Best might get 3 seats at most, and the others go to the Libs

 I had no statistical basis for this, so didn't bother to specify which of the seven I would give to the Libs; that said, the Liberal party won every one of the SAB-predicted seats so it wouldn't have mattered. Going on my assumption four of those would be won by the Libs raises my hit rate to 40/47 or a respectable 85%. If my prediction that any possible SAB seat lost would be picked up by the Liberals was extended to all seven seats for a standard 2-party dominated election I would have scored 43/47 or 91%.

The other four are Adelaide, Giles, King and Mawson. Two were predicted for Labor and won by the Liberals, and the other two predicted for the Liberals and won by Labor. As a result, my prediction without SA-Best would have been ALP 19: LIB 25: IND 3--the actual result. That's a (rather fudged) 100%. Then again a recount in Adelaide is looking like it might actually pass to the ALP and if so: "Yay! Another seat right!" but also "There goes my post-hoc 100%"

So despite the rather poor result this election, in a race without Xenophon (assuming that did not shift preference flows etc), this would have been a solid result.

Thanks, Obama! Nick!

SOUTH AUSTRALIAN LEGISLATIVE COUNCIL

So I didn't make any predictions for the LegCo due to time constraints. That said, I think it's worth checking in on since I actually prefer voting for upper houses where there is a diversity of great and not-at-all-great candidates to balance.

Newspoll had the vote split as follows 2 days before the election: ALP 31: LIB 34: SAB 17: GRN 8: Other 10 (I am ignoring the ReachTEL poll of the same day since it only considered ALP, LIB and SAB with no GRN or Others). There were 11 seats to fill, and they need to reach the Droop Quota: that's (Total votes/( Seats+1))+1 or just over 1/12th (8.333%) of the vote in this case. So logically you'd expect ALP to win 3, LIB to win 4, SAB to win 2, with the last two up for grabs. GRN are a rounding error away from getting one too. So We could have predicted ALP 3: LIB 4: SAB 2: GRN 1. There is 10 sitting with the Others, but they are too diverse to likely scratch a seat together and it would be impossible to pick which party to give it to (possibly Dignity, based on their past election). Alternatively, The residual votes after the assigned Quotas would look something like ALP 6: LIB 0.333: SAB 0.333. The chance of far left and far right minor parties cooperating enough to outperform the ALP at that stage would have been unthinkable given the absence of preference tickets in modern voting. So the final prediction (postdiction?) would have to be ALP 4: LIB 4: SAB 2: GRN 1.

And that's exactly what the results came out as.

HOLY BY-ELECTION, BATMAN!

And lastly, of course, I did not make a prediction for the much-discussed Batman by-election which gripped most of the nation while SA was voting locally. I didn't even have time to predict the SA LegCo, so what do you expect?

The 2016 election had the ALP hold off the greens with 51% of the vote, 2-party preferred. Polling from February was predicting an ALP win of 53%, the only poll on the by-election's Wikipedia page. Off of that, I should have suspected the ALP would hold the seat, but many commentators were predicting a Greens victory. As a result, Labor's eventual victory is being hailed as a terrible loss for the Greens, but I don't see it that way at all.

There was a recorded swing to the ALP. But then the Liberal party didn't even run, and I don't see those voters (some of whom may have lingered briefly with the conservatives or other minor parties) flowing to the Greens very much. Then again, their eventual landing with the ALP would have happened after the Libs dropped out in 2016. So, yes, there was a swing against the Greens. On the other hand, the Greens still picked up almost 40% of the primary vote. That's not a bad result for a party often left in distant third place.