Chasing Gold III: April FM Election

Alrighty, todays update (4/7) has been added.
As you can see HEM has rebounded from his half a ticket woes to approximately a 20% chance of winning the election. The addition of Peeps, and some significant improvements in terms of the platform were the biggest factors.

Additionally, the First Minister debate was conducted on discord earlier this evening. I think it's fair to say there were no clear standout performances and in a way thats a win for Istillian as the present leader of the race. Additionally though, since it was less a matter of everyone did great and more everyone did mediocre I don't think anyone got put away. It will be very interesting to see the results of the flash poll Calvin is conducting, which you should all definetly take here
 
Update 4/8/20
This update includes Calvin's Post Debate Poll. It shows at best slight changes from yesterdays update, which is really just more good news for Istillian. Unless there is some huge endorsements stacked up on behalf of UV or HEM, I wouldn't expect the model to turn out any different results. Istillian and Hezekon are the clear favorites going into election day. Additionally, given that we live in an era of Europeia that polling captures about half of the electorate rather than a fourth, it would likely definetly take a behind the scenes effort comparable to Rand's or Brun/Rach to outright win the election. I think it is fairly feasible with a regular GOTV that either HEM or UV could drag down Ist's support bellow 50% and force a runnof and then hopefully turn some supporters to eek out a victory then. Regardless both candidates definetly do face an uphill battle to victory and as each day goes by I think it's narrowing as they don’t have any real opportunity to put up gains on ist.
 
So while I am a few days late with this I've been reflecting over the model and what it actually shows. While the percentages are the listed as percent chance to win in reality they resemble more of a projection for vote percentages. Notably they were pretty close to the actual results. The model really is just using qualitative assessments to augment polling results and estimate how the rest of the people who were not polled will vote. That is evidently not really an estimation of who will win the election. However by assuming that the "best" candidate will win each time instead of counting the qualitative assessments as a positive for everyone I reversed it so that if they are not the best candidate in each respective category it counts against them. This implies that if your campaign is not as good as another competitors you are far less likely to pick up the votes of those not already polled or convert them in your favor. This method returns vastly different percentages, with Ist at 76%, UV at 23% and HEM at 0%. Interestingly when applied to the most available past election, Dax versus Calvin, the new model does actually correctly predict a Calvin victory with 57% for him, 38% for DAX and 4% for OD with 0% for cuddles. It would be interesting to see how either model shapes up against past elections and perhaps that'll be my next effort for this outlet!
 
It would be interesting to see a projected winner model and projected vote share (I'm greedy).
 
Agreed that it would be interesting to see both for comparison. It's kinda messing with my brain a bit right now because the two feed into each other a lot; if the vote shares look like they're going to come out very close, minor things unrelated to the candidates' actual quality, such as GOTV, will play a huge role in determining the outcome of the election, bringing the chances of victory even closer together. That's why I think it could be very constructive to be able to see both. Anyway, this is really cool stuff, thanks GraV!
 
Back
Top