Nate Silver has left 538 but he continues to maintain his Presidential election model which attempts to put a figure on how likely each of the candidates are to win the electoral college and therefore win the election. The model is here (the actual model itself is only visible to subscribers, but the commentary is above the paywall):
https://www.natesilver.net/p/nate-silver-2024-president-election-polls-model
Since the Presidential model came out, the model has shown a tight race which means entertaining swings for those of us anxiously watching the model to see whether our favoured candidate is around 45% or 55%! The main movements have been:
Trump was leading Biden by a wide margin
The first model measuring Trump against Harris gave Harris the lead
Trump overtook Harris at the end of August
Harris regained the momentum after their debate and Trump's "they're eating the dogs" outburst and retook the lead on 20th September
Harris slowly stretched her lead out and the model gave her around a 56% chance of winning throughout late September and early October
The momentum started shifting back towards Trump and he retook the lead today, 17th October
The model currently stands at:
Trump 50.2%
Harris 49.5%
Will Harris regain the lead in Nate Silver's model? If so, in which week will she first regain the lead?
Resolution notes:
The primary source of truth for this market is Nate Silver's blog. Some of his content is subscriber-only, but the results of the model can usually be found in other media outlets including Nate Silver's own Twitter feed.
This market only references Silver's model for winning the electoral collage (ie. winning the election) rather than winning the popular vote or getting ahead in the polling average
If Nate Silver stops publishing a Harris vs Trump model for any reason (including one of the candidates withdrawing) before Harris has overtaken Trump and does not bring it back online by the time of the election, this market will resolve to the "she will not overtake him" option.
Related market - Silver's old home at 538 gives Harris a 52% chance of victory. Will Trump overtake Harris there? /SimonGrayson/when-will-trump-overtake-harris-in-tq3ucuxus2
1,000
3.00There’s a pretty clear consensus here for sticking to the letter of the market and resolving this to the “not before polling day” option on the basis that the updated forecast came out after midnight and therefore ON polling day.
I apologise for messing this market up. I should have considered a pre-election forecast coming out on what was technically election day when I created the market. If you’ve lost Mana on this market as a result of that mistake (ie. Because you had money on that option with an eve-of-election poll in mind, NOT because you lost Mana meta-gaming this market after it became clear it was ambiguous), let me know and I will send you Mana to refund you.
@SimonGrayson fwiw, I think you made the right call. But if you'd gone the other way and resolved according to the spirit rather than the letter, I wouldn't have faulted you for that.
I was one of the meta-gamers, and if my prediction about your resolution-approach was wrong, well so be it, I'd have bet wrong.
Apologies for the delay. I haven’t visited the site for a few days since I was a bit disheartened by being excluded from all of the real money sweepcash action because I’m not in the US!
I have to admit that this market was poorly worded because the wording doesn’t account for a post-midnight update that’s still pre-election.
By the spirit of the market, I feel that this should resolve to 4th November as this was a final-day, pre-election lead for Harris and the “not before polling day” option was meant to mean that Trump would maintain a lead until the end.
As I see it, there are four options:
Resolve to 4th November in the spirit of the market
Resolve to “She will not overtake him before polling day” as 12:30 is not technically before polling day
Resolve 50/50 to those two options as a compromise
Resolve N/A as the market was poorly written and is unenforceable
Does anyone have a view between those options? Particularly people who don’t have a stake in the outcome?
If there’s no clear answer, I may have to ask the mods if they are able to arbitrate!
@SimonGrayson I'm neutral with no $ here and as much as I'd like to see the nod to Harris taking the lead based on today's most recent update, the wording clearly states "before polling day".
@SimonGrayson As someone without a Mana stake, I think YES to Nov 4 best fulfills the spirit of the market. Nate's model did its final run with no polling knowledge, and Harris overtook before polls opened practically anywhere.
@SimonGrayson 4th November seems reasonable to me. And as I mentioned below, there is a textual justification. But I think there's also case for N/A
@SimonGrayson Put me down as a mod in agreement with @Ziddletwix and not @Conflux . This seems pretty clearly “not before polling day” to me. (I don’t think I hold any mana here fwiw).
Polling day = November 5th and the fact that’s an answer and the market is dependent suggests that “week of” in the description referred only to the days of the week of Nov 4th that were not covered by other answers, meaning just Nov 4th. Also the fact that it’s listed as “4th November” and not a range suggests only the 4th.
Maybe one could make a case about time zones, but Nate himself tweeted about running the model at midnight and then posted at exactly midnight ET that the model was running, so I would use the source’s time zone for resolution, or maybe UTC, both which would have been Nov 5th.
@dglid i sold my stake at a loss a while ago because i don't like being involved in contested resolutions but yes IMO this is fairly straightforward—there are two options listed, "November 4th" and "She will not overtake him before polling day". The model updated on November 4th, and she was not in the lead. The model updated November 5th i.e. polling day, and she took the lead.
Extrapolating that the spirit of the market was that "November 4th" was a placeholder for "the week including November 4th" (despite the other options explicitly listing their entire duration?) seems like a huge stretch? if that was the intent, why not just say "She will not overtake him"—there's no need to explicitly include "polling day" there.
Didn't want to have a mana stake here but I don't think this is a particularly ambiguous resolution. I asked about the time zones yesterday because this was a pretty predictable concern (not surprising nate would do a final update on election day?) and I thought "the creator has implicitly chosen another timezone" was the only real ambiguity here (even if I think it should clearly be nate's time zone).
@SimonGrayson I have a little mana on November 4, but I think this can't resolve to that date if the resolution came after midnight. Easiest and best and most predictable to resolve on objective factors when it's possible, as it is here.
@SimonGrayson how are time zones handled? IIUC, nate says there will be one final update at 12:30AM, election day. does that count as "polling day", or could it still resolve November 4th? https://x.com/NateSilver538/status/1853480554517151939
@Ziddletwix People are betting this up, but it didn’t happen on the 4th of November? Nate published an Election Day update. I don’t see why it wouldn’t resolve to “not before Election Day”, unless there’s some reason not to use Nate’s time zone
@Ziddletwix Hmm, this is tricky. At first I thought you were right, but the resolution criteria say "Will Harris regain the lead in Nate Silver's model? If so, in which week will she first regain the lead?" And the answers are "yes" and "the week of November 4th." So it's rather dubious
@Conflux It's not dubious at all. Clearly '4th November' resolves YES according to the resolution criteria.
'She will not overtake him before polling day' probably also resolves YES, but there are some subtle time zone issues to consider.
@biased Only one option can resolve YES.
I hear the argument that “week” implies Nov 4 could resolve YES. But “she will not overtake him before polling day” is the most directly/literally true option. That statement is clearly NO, she overtook him on polling day and was behind him the day before
@Ziddletwix 'She will not overtake him before polling day' resolves YES if she only overtook him on polling day, right?
@Ziddletwix I think not including Nate's final model update -- which took place well before the polls opened on Nov 5th -- is a bit ridiculous. Obviously the market should include his final update.
But that's not how the prop market is worded. So, now that all the info is already known about Nate's model, we instead get to bet on how much of a stickler @SimonGrayson is about wording.
When markets go meta...
@Jubs This is a linked market - the market will resolve to one option once we know the result rather than each option independently resolving to yes or no.
@SimonGrayson
> He will not overtake her before polling day
I could be misunderstanding, but did you mean "She" and "him" here?